Home

Instructor`s Manual Communication Systems Modelling

1. 48 FM deviation 50 FM and Bessel Zeros sees reeset toners oie 51 FM demodulation with the PLL ee einne en i RE A R 51 The Costas lOOp 52 FURTHER AND ADVANCED DIGITAL EXPERIMENTS 55 BER measurements in the noisy channel c cescessceesecseeeeeeeeeeseeeeeeeeeeeeeeeenseees 57 BER instrumentation macro Model ccceecceesceeeceeeceecesecesecseecaeecseeeaeeneeeeeeeerenrens 58 Bit clock regeneration ashe HR hae Rd NER 59 Carrier acquisition Meche a as eg WR Rk 59 DPSK carrier acquisition and 60 PCEM MOM gatas esl isthe Soest ate oa est RCL 61 Block coding and decoding cceccesccesecseeesceesceeeceseceseceaecaecseecaeecseseaeeneeeeeeeereneees 61 Block coding and Coding 62 Instructor s Manual Convolutional Coding E i tendons ii 62 63 PPM and PWM 63 QAM and 4 PSK 63 Multi level QAM and PSK EOR 64 Spread spectrum DSSS amp 66 Digital utility 5
2. Q3 not the amplitude spectrum which is what is usually thought of and usually seen on a spectrum analyser This is a linearly modulated signal The bandwidth is twice that of the message The phase cannot alter the message bandwidth Q4 no the baseband filter will impart a certain shape to the message sequence spectrum This shape will be reflected in mirror image form in either side of the BPSK spectrum no matter what the carrier frequency or phase This is a linearly modulated signal Q5 it removes the sum components from the multiplier output around twice the carrier frequency and should be wide enough to pass sufficient of the signal to allow the detector to regenerate accurately The narrower the passband the less the noise but the more the degradation of the signal A compromise Q6 delay through the filter changes with bandwidth area under the delay frequency curve remains constant Depending upon the order of the filter it can change multiples of p So the phase between the received signal and local carrier could rotate a significant amount and so the amplitude could go up if not already at a maximum Q7 phase between the local carrier and the resultant of each pair of sidebands This is easy to show using phasors It is not easy to measure directly impossible but very easy to optimise Instructor s Manual Signal constellations Read the Advanced Modules User Guide To save time set the following on boar
3. b say Because the sampling rate is effectively halved as compared with the normal usage the allowable message bandwidth is also halved in this mode Q7 advantageous where the message is speech PCM decoding TRUNKS None It would be advantageous to demonstrate the effects of companding with speech as the message Unfortunately the processors used in the PCM modules restrict the system bandwidth to well below that of speech Oscilloscope Triggering Using the frame synch signal FS for oscilloscope triggering is useful for showing frames but if the sweep circuits miss a beat then adjacent frames will swap positions on the screen Sometimes this is a nuisance If the FS signal is divided by 2 this cannot happen Use the divide by two sub system in the BIT CLOCK REGEN module or divide by 4 in the LINE CODE ENCODER WARNING Please note that the maximum clock rate for this module is about 10 kHz Thus it is safe to operate at the clock rate of 8 333 kHz PCM DECODER For message reconstruction an LPF operating down to about 200 Hz is required The TUNEABLE LPF will not tune as low as this Version V2 of the PCM DECODER module has a built in LPF for this purpose so it is advantageous to supply this up dated version if possible Instructor s Manual 25 answers to tutorial questions Q1 there will be a delay though a bandlimited channel The phase adjustment acts as a delay adjustment to the stolen carri
4. sess se ose Co on O Thus for 95 of the total power only the first pair of sidefrequencies is required Instructor s Manual 43 For 5 the entries side relative relative cumulative of frequency amplitude power sum of reference powers power a as mos 9934 Thus for 95 of the total power a bandwidth to include the first 5 pairs of sidefrequencies is required Q4 from the sidefrequency locations the message frequency is 2 kHz To find the power it is necessary to start from the carrier and obtain a cumulative sum of the power in each sidefrequency until all significant components have been included This can be decided by waiting until the cumulative power approaches a limit freq volts relative cumulative denormalized a ae a 9 gt 10 11 12 13 14 15 16 17 18 The relative power is entered as the square of individual components This should be corrected by a factor which incorporates the load resistance and takes into account the rms value of the amplitude But for the present purposes this is unnecessary The amplitude E is equal to the square root of the cumulative power so E 5 98 volts 44 Instructor s Manual Using this to denormalize we get the final column of entries which are the denormalized amplitudes equivalent to the Bessel coefficients associated with each spectral comp
5. MODEL DECISION UTILITIES outa je REF DIODESLPF RE LPF el 5 ERROR COUNTING UTILITIES SEQUENCE _polarity inversion FREQUENCY COUNTER MASTER SIGHALS 5 9 g z sin TRANSMITTER NOISY CHANNEL RECEIVER INSTRUMENTATION modified system without encoder amp decoder ERROR COUNTING UTILITIES module This module contains two sub systems an X OR and a pulse counting facility The former has been met in the experiment entitled PRBS generation These are both described in detail in the Advanced Modules User Manual Condensed descriptions suitable for this experiment are to be found in the experiment entitled Digital utility subsystems under the headings Exclusive OR and Timed pulse warning make sure the default settings have been selected with the on board switches SW1 and SW2 TRUNKS No signals required answers to tutorial questions Q1 all SEQUENCE GENERATOR modules are identical They are driven by the same clock They are not influenced by the noise or received sequence in any way There is no reason why they would step out of alignment Q2 the long sequence introduces more patterns and exercises the system more thoroughly thus giving a more realistic error measurement Q3 matching the signal to the detector threshold which is offset from zero volts by approximately 25 mV BER instrum
6. Q5 Adaptive delta modulation TRUNKS Speech is optional either via TRUNKS or using student s own SPEECH modules Delta sigma modulation The operation of this modulator demodulator system can be described in simple terms But this will probably fail to reveal the reasons for its application in the CD player where both cost and performance are paramount The experiment works but it has been left to you the instructor to flesh out the experiment details to complement the level of your course work Instructor s Manual 27 28 Instructor s Manual FURTHER AND ADVANCED ANALOG EXPERIMENTS 29 30 Instructor s Manual Amplitude modulation method 2 TRUNKS Speech is useful perhaps students have their own SPEECH modules answers to tutorial questions The phasor diagram represents the AM signal y t A 1 m cosut cosm t The amplitude of the AM signal 15 the resultant of the three phasors The resultant of the DSBSC is co linear with the carrier phasor since the two angles are equal The trough of the AM signal occurs when this sum is at a minimum You can see the sum will be zero when 180 deg and m 1 If the resultant of the DSBSC is not co linear with the carrier but offset by an angle not zero then the resultant phasor of the DSBSC can never combine with the carrier phasor to produce zero amplitude Q2 you cannot use a normal commercial phase meter It would not know the
7. answers to tutorial questions Q1 for synchronous demodulation once the carrier has been acquired then the bit clock is available by division Q2 Q3 the bandwidth of the ASK is twice that of the message the data Refer to the bandwidth of a DSB signal It has nothing to do with the phasing of the message So the bit clock carrier phasing has no influence upon the amplitude spectrum of the ASK FSK frequency shift keying The BIT CLOCK REGEN module is essential In addition e an extra AUDIO OSCILLATOR or VCO as a sinewave source in the range to 5 kHz is convenient This is the SPACE signal for transmission But alternatively you can supply a suitable sine wave at TRUNKS e extra UTILITIES and TUNEABLE LPF modules are optional Required if both channels of the demodulator are to exist simultaneously This experiment has no step by step Tasks Only a collection of models to evaluate 20 Instructor s Manual Preparation by students is important especially with regard to the asynchronous demodulator This uses the two BPF in the BIT CLOCK REGEN module Because the filters are required to be on different frequencies one must be tuned by an external oscillator the other to 2 083kHz internal clock The BPF bandwidths are approx 2 of centre Having one tone arbitrarily called the MARK at 2 083 kHz sets the bit rate many times below this The divide by 8 in the BIT CLOCK REGEN module is used to lower an AUDIO OSCILLATOR to
8. 1 Then 1 The term in square brackets can be expanded by the binomial expansion to give the approximate result k cos ut 1 k sin cos ut k sin ut Further approximation small leads to kcos ut 1 ka cos ut ka cos ut The first term represents the wanted term the remainder unwanted Notice that for a 0 odd and even harmonic distortion will be introduced which is otherwise zero Q6 from the in phase and quadrature form defined in the answer to the previous question the Armstrong envelope is Instructor s Manual 49 a t Jia k sin cos k cos cos ut k sin cos which when 0 is very small simplifies to a t Ju 2k sin cos ut cos ut This can be expanded using the binomial expansion but perhaps the exercise is getting a little tedious One can at least see that there will be a fundamental while is not zero FM deviation multiplication This experiment requires a Version V 2 of the 100 kHz CHANNEL FILTERS module You may consider this experiment too short If so then it could be extended by continuing on to the next one entitled FM and Bessel zeros That experiment might also be considered too short so combining them has its merits TRUNKS 130 kHz sinusoid from your VCO See the answer to Q4 below answers to tutorial questions Ql Q2 Q3 in practice harmonic multipliers requiring
9. It is easy to measure the depth of a null for a tone but less so for speech If the null of the tone is better than say 40 dB as it should be then it could end up below the system noise and one can then only give an estimate of the null eg better than 40 dB Take care not to overload the MULTIPLIERS else non linearity will spoil performance by introducing distortion components which cannot all be nulled at once at the receiver Do not allow the TRUNKS signal level to be such that the received signals at each TIMS 301 are above TIMS ANALOG REFERENCE LEVEL otherwise over load at the receiver will degrade the achievable depth of null A BUFFER AMPLIFIER at the receiver could of course be used to adjust the amplitude from TRUNKS Instructor s Manual 41 There is no need to trim the two carriers to exact phase quadrature The sin and cos outputs from the MASTER SIGNALS module are quite close enough to quadrature You can note the frequency of the AUDIO OSCILLATOR and make its measurement part of the experiment requirements For more mature students this could be a two tone audio signal the measurement of the frequency of which requires a little more ingenuity patching diagram MULTIPLIER Pac fe MASTER SIGHALS PDM output 100 7 sin es 100kHz cos oy PDM Generator answers to tutorial questions Q1 assume all noise output is independent of the local carrier phase Thus we are concerned with the wante
10. 8 67 Instructor s Manual 69
11. Note that not much attention was devoted to a change of clock speed You may prefer to insert a separate non inverting amplifier between the INTEGRATOR and the ADDER rather than use the ADDER gain g to change the loop gain A change of ADDER gain g alters the step size but one cannot see this as the change occurs inside the ADDER TRUNKS Speech is optional either via TRUNKS or using student s own SPEECH modules answers to tutorial questions Q1 stable picture message is a sub multiple of the sample rate Q2 Q3 highest message frequency determines the maximum message slope Q4 V3 Q5 on e bit per word Instructor s Manual Delta demodulation TRUNKS Speech is useful either via TRUNKS or using students own SPEECH modules answers to tutorial questions Q1 for a given message slope and sample rate there is a step size below which slope overload will occur To avoid this the step size must be increased When the message slope is low a small step size will result in smaller errors Thus avoidance of slope overload requires large steps which will increase the error in the regions of low slope giving rise to what is referred to as granular noise Q2 this scheme allows the measurement of signal noise and then noise alone The method is described with reference to Figure 2 in the Experiment Q3 Q4 cancellation of more than one frequency component simultaneously would require more than a simple phase shifter
12. a multiplication of 9 are seldom used since the percentage separation of the 7th and 11th nearest unwanted odd harmonic terms makes the filtering too difficult For a tripler it is the fundamental and fifth which are unwanted thus making less demands upon the filter Working at 100 kHz and below and with speech messages the desired bandwidths are large fractions of the carrier frequencies which makes the filtering quite difficult These relationships are not common in commercial situations where carrier frequencies are considerably higher The assumption is made that the bandwidth of the wanted output signal 33 kHz for the tripler is approximately three times that of the input bandwidth at 11 kHz 0 dB 50 dB frequency gt diagram is not to scale harmonic multiplier output The three rectangles represent the bands occupied by the significant components at the fundamental third harmonic and fifth harmonic of the 11 kHz carrier The heavy rectangle 50 Instructor s Manual represents the minimum requirements of the bandpass filter to select the third harmonic components from the nearest unwanted components It is shown having an attenuation of 50 dB at the frequency where it overlaps the unwanted components This is in excess of the 40 dB specified This is based on the fact that at the output of the harmonic multiplier the amplitude of the harmonics fall in proportion to the harmonic number Thus the c
13. answered experimentally during the experiment Thus purely analytical questions were to be avoided As experiments were added this philosophy was sometimes forgotten and analytical questions crept in These questions have not been edited out but answers are not always supplied note taking No advice has been given about note taking This does not mean that note taking is unimportant You should have a definite policy as to how you want the results of each experiment recorded Students should be made aware of this laboratory programs TIMS is capable of supporting courses in communications at all levels These courses are typically given in two often independent strands namely analog and digital However it is perfectly feasible to give a general course on communications without introducing this split What ever the course structure TIMS can support the theoretical side of any such course with related experiments Many instructors will prefer to write their own laboratory material extracting those parts of this Text which suit their purposes Others might prefer to build the theoretical side of their course around a selection of the experiments contained herein For contents please see end of file Page numbers refer to those in the footers and NOT those displayed by the PDF reader INTRODUCTORY ANALOG EXPERIMENTS Introduction to modelling with TIMS It is recommended that this Chapter be read by all students prior to
14. attempting their first experiment be it analog or digital Modelling an equation TRUNKS Depending on the quality of your students and the time available you may feel the experiment is long enough already Otherwise it is instructive to supply a bandlimited speech signal at TRUNKS Preparation of such a signal is discussed in the Introduction to this Instructor s Manual Alternatively you can use a SPEECH 1 module There are then two options suggested you can think of others 1 to the speech add a single tone from an AUDIO OSCILLATOR The tone frequency should lie within the speech band say 1 kHz You can set the relative level of the speech and tone to any value which you find convenient try the experiment and see which you prefer When this corrupted speech signal is used as the input to the nulling model it should be possible to null the tone from the output This adjustment will put a hole in the speech near the tone frequency but the speech will still be intelligible 2 add a second tone to the speech or replace the in band tone this one being above the filter slotband This would be removed when the LPF of the HEADPHONE AMPLIFIER is switched in giving a demonstration of noise removal when the wanted and unwanted signals are non overlapping in frequency oscilloscope triggering As stressed both in the introductory chapter of the student text and in this experiment it is important for students to develop
15. can be simulated with a 101 kHz output from a VCO A trick message 15 a two tone audio which makes the SSB look like DSBSC yet behaves like SSB eg recovered message amplitude is insensitive to local carrier phase Do this only if the students are sufficiently experienced to cope with it It can be simulated with two VCOs say on 101 and 102 kHz ISB you can make this with TWO generators each of the phasing type and using different speech messages Alternatively ONE genuine SSB generator for the upper sideband say derived from speech and for the other sideband ADD a sinewave say 99 kHz from a VCO This simulates a lower SSB derived from a 1 kHz tone The ISB signal will show up the shortcomings of the synchronous demodulator under study since it cannot differentiate between upper and lower sidebands The true SSB demodulator will be examined later in the experiment involved with SSB demodulation and CSSB DSBSC SSB message VCO 102kHz three possible TRUNK signals You can think up other signals if you want to add your own personal touches For example a good trick signal is 8 Instructor s Manual CSSB compatible SSB This has not been defined in any experiment to date but has probably been met in formal lectures This is made by adding a large carrier component to an SSB If the SSB was derived from a single tone then CSSB looks like AM with a small depth of modulation
16. depth of null is close to 40 dB Q2 the PHASE SHIFTER introduces different phase shifts at different frequencies To make an analog wideband constant phase shift network is hard enough but to make it adjustable as well is very difficult impossible Periodic signals in the noise other than the one for which the cancellation has been achieved are unlikely to arrive at the ADDER exactly out of phase so would not cancel although may be reduced in amplitude Circuit noise being wideband will likewise not cancel over a wide bandwidth Q3 the method without refinements would be approximate only Measure the total power out when there is no cancellation one ADDER input removed then cancel the fundamental and measure remaining power Assumes random noise power is even lower than the harmonics otherwise is measuring signal to signal plus noise and distortion ratio Q4 since the phase shift introduced by the PHASE SHIFTER varies with frequency the null will not be maintained Refer to an earlier question Q5 in principle yes But the PHASE SHIFTER must compensate for small phase shifts within each module and so itself may need to introduce slightly more or less than exactly 180 This the INVERTING AMPLIFIER would not do DSBSC generation TRUNKS The experiment does not require any TRUNKS signals However if you like you could send speech so that students can see what a DSBSC signal looks like when derived from such a message P
17. filter characteristics ISI can be minimized by certain critical spacings Outside the scope of TIMS cost complexity required SNR a TTL LOW is the green socket on the VARIABLE DC module e post 1998 TIMS have a TTL high available from the VARIABLE DC module e the DIGITAL DIVIDER module provides a TTL HI Some digital modules when unclocked have their outputs at a TTL HIGH e the COMPARATOR in the UTILITIES MODULE with REF grounded e DELTA MODULATION UTILITIES output no input or clock connected e 02 of the TWIN PULSE GENERATOR Q1 is a TTL LOW Instructor s Manual 21 Alternatively use a ve output from the VARIABLE DC module amplified and reversed in polarity through a BUFFER AMPLIFIER Q7 an SNR advantage BPSK binary phase shift keying 22 TRUNKS An unknown BPSK near the carrier frequency used in the experiment could be transmitted If you derive a stable clock from the system 100 kHz master then part of the requirement would be to report its frequency Identification of the message sequence might be a problem there would be no start of sequence information although you might be able to generate a very short sequence answers to tutorial questions Q1 agreed It is transmitted by an analog transmitter and it has a narrow sub octave bandwidth Q2 some line codes may have DC components AC coupling of the MULTIPLIER will inevitably introduce base line wander following repeated series of 1 or 0
18. first experiment with a complete transmission system does not use any line coding but uses the LINE CODE ENCODER and LINE CODE DECODER in the straight through condition ready for line coding in a following experiment If you do not have LINE CODE ENCODER and LINE CODE DECODER modules then some modifications to the set up for this experiment are necessary You will need two UTILITIES modules The SEQUENCE GENERATOR could be clocked by the 2 084 kHz sinusoidal message from MASTER SIGNALS But a TTL version is required by the DECISION MAKER so it is converted first to TTL by the COMPARATOR in the UTILITIES module The sequence input to the channel is then taken from the yellow analog output of the transmitter SEQUENCE GENERATOR Note that this is an inverted version of the TTL output This simplifies the SEQUENCE GENERATOR circuitry but can be an occasional inconvenience especially if not observed The sequence inversion must be accounted for at the X OR gate of the ERROR COUNTING UTILITIES module You will observe that the instrumentation SEQUENCE GENERATOR TTL stream and that output from the DECISION MAKER are inverted one with respect to the other To correct for this a BUFFER is inserted at the channel output Its gain should be set to unity The modified system is illustrated below 57 data UTILITIES rT COMPARATOR RECTIFIER DIODESLPF RE LPF CHANNEL
19. from an appropriate LPF and dropping the scaling factor are Instructor s Manual 5 LPF output berg ASIN Ignoring the DC term then the amplitude ratio of the wanted fundamental to the only unwanted 2nd harmonic term is 4 ratio m This for m 1 is not very good but for smaller values of m it is acceptable apart from a scaling factor the unwanted components lie near 2 for the full wave rectifier but near for the half wave Thus the filtering requirements are different This is perhaps of little significance when 0 gt gt u but otherwise may be important Q4 radio signals can arrive at an antenna via one or more paths If more than one this is called multi path reception If the paths are of different lengths the resultant will be the phasor sum Particularly in the case of ionospheric paths these all change with time measured in seconds So the received signal strength varies with time It can be shown that components even very close in frequency as are sidebands in a narrow band signal will suffer different fading effects hence the term frequency selective fading It is possible for the carrier of an AM signal to fall even if only momentarily to zero amplitude while sideband amplitudes remain finite Thus the envelope suffers severe distortion Thus the output from an envelope recovery circuit delivers a non linearly distorted message intelligibility is s
20. heading multipliers and modulators page 13 SSB demodulation the phasing method TRUNKS SSB ISB and DSBSC signals are required 1 the SSB should be derived from speech 2 the ISB should have at least one sideband derived from speech 3 the DSBSC should be derived from speech with a sinewave from VCO added to one sideband to simulate an interfering carrier Its amplitude should be of similar magnitude to the DSBSC itself or otherwise as you decide It does not matter if the speech is the same for all although for proof of identification purposes separate messages would be preferred But this will require more modules at your end An economical method of producing these signals is to use an arrangement based on that used in the previous Product Demodulation experiment It needs three ADDERS Thus 1 USSB from the above generator 2 one side of the ISB is the above SSB into an ADDER together with the other supplied by a VCO in the range say 97 99 kHz 3 DSBSC can come from within the phasing generator into an ADDER into which is connected the interference from a VCO the same as for ISB on top of either sideband If you have more modules available then it would be preferred that the three signals looked a little less alike after demodulation being derived from different speech and having tones of different frequencies answers to tutorial questions Q1 suppose the lowpass filter bandwidth 15 B Hz Let the
21. its suppressed carrier Q3 the FREQUENCY COUNTER counts level crossings in one direction only in a given time interval set by front panel switch This level is above zero but well below the TIMS ANALOG REFERENCE LEVEL The zero crossings of a DSBSC signal are not uniform You should consider how you might attempt to prove this or otherwise define the zero crossing locations The DSBSC signal is not periodic at least not in a time interval comparable with the message period let alone the carrier period A quick answer is that you don t know what the counter will display but it surely won t be the carrier frequency Secondly it will probably vary from count to count Q4 see previous answer Q5 three pairs of DSBSC components of relative amplitude and Absolute amplitudes unspecified Q6 analytically it is not necessary But if oscilloscope displays are to be compared with normal expectations then it is necessary See the experiment entitled Envelopes in Volume A1 Amplitude modulation If you have them WIDEBAND TRUE RMS METER modules could be useful eg see Tutorial Question TRUNKS Send speech but if students have SPEECH modules this is unnecessary Instructor s Manual 3 answers to tutorial questions Q1 while watching the envelope and the trough to trough distance Q increase m from below unity to above unity If you imagine Q changing sign as m passes through unity then it b
22. of the receiver is infinite The superiority of the synchronous demodulator under these conditions is demonstrated But this can be down graded if there is any non linearity in its circuitry typically in the multiplier since this will cause intermodulation products to be generated some of which could pass through the LPF There is no inherent selectivity in the lower receiver The audio filter is of no help in separating signals at RF This would have to be done with a pre detector filter which would of necessity be an RF bandpass type Q5 the receiver should ideally be linear a if operating at the TIMS ANALOG REFERENCE LEVEL the addition of a small amplitude signal at 90 kHz should not be noticed b amp c eventually as its amplitude is increased overload of some circuit will occur probably the multiplier new components will be generated some of which will be observed Q6 see answer to similar question in the experiment entitled Envelope recovery The synchronous demodulator would suffer from linear distortion individual frequency components would be of the wrong amplitude audible and wrong phases not detectable Instructor s Manual by ear although the waveform shape would be incorrect Importantly there are no new frequency components which would occur with non linear distortion With speech the distortion could be of minor significance Q7 see the chapter entitled Introduction to modelling with TIMS under the
23. start of synch signal from the SEQUENCE GENERATOR and the oscilloscope ext trig input The following methods are worth thinking about although each proves inadequate at clock speed of 2 kHz as used in this and in many later experiments e use the TWIN PULSE GENERATOR using the delayed pulse to trigger the oscilloscope But the delay available at typical 2 kHz clock speeds is insufficient e use the DIGITAL DELAY sub system within an INTEGRATE amp DUMP module This provides an adjustable delay up to 1 5 ms but this is inadequate e TUNEABLE LPF followed by the COMPARATOR in the UTILITIES module will introduce a delay variable with bandwidth Once again only a few clock pulses at most However for demonstration purposes the clock speed can be increased even to 100 kHz if you really must see the next n bits where n is not too large Then all methods are useful the rectangular pulse retains its approximate shape if the filter bandwidth is at least 1 27 Hz where T is the width of the pulse in seconds synchronized if they use the same clock aligned if they are the same sequence pattern and both start at the same time where the actual start occurs is a matter of definition the filter will introduce a time delay This depends upon its order size and bandwidth The order is fixed but the bandwidth and so transmission delay can be changed in general for a longer sequence
24. the message Measure their resultant This gives a point on the envelope amplitude versus time curve If the carrier is drawn vertically then for the special case of carrier and resultant being co linear amplitude modulation the graphical construction of the envelope is simplified by projecting a parallel line from the resultant peak across to the envelope curve But if the DSBSC resultant is not co linear with the vertical carrier phasor this simple construction technique projecting across is invalid The resultant amplitude must be 4 Instructor s Manual measured and used for the envelope amplitude at that time but the point on the envelope curve cannot be located by simply projecting across horizontally Q3 radio signals can arrive at an antenna via one or more paths If more than one this is called multi path reception If the paths are of different lengths the resultant will be the phasor sum Particularly in the case of ionospheric paths these all change with time measured in seconds So the received signal strength varies with time It can be shown that components even very close in frequency as are sidebands in a narrow band signal will suffer different fading effects hence the term frequency selective fading It is possible for the carrier of an AM signal to fall even if only momentarily to zero amplitude while sideband amplitudes remain finite Reference to the relevant phasor diagrams will show that the
25. until eventually there would be no effective pulse at all At the receiver as the sample widths fell so would the signal to noise ratio of the recovered message until eventually it would reach an unacceptable level Q3 see the answer to the previous question The guard band allows for errors in the location of the samples and of the switching pulse in the receiver Practical pulses may have skirts which will overlap so the guard band can help to reduce this source of crosstalk Q4 the linear sum of the messages from all channels Q5 see above patching diagram Power measurement 14 TRUNKS Students should make their own test signals so that they have control over all parameters However an extra audio sinewave would be useful as would some speech answers to tutorial questions Q1 1 single tone 2 two tones 3 DSBSC 4 AM 5 Armstrong s signal 6 angle modulation 7 bandlimited speech is often quoted as having a peak to average power ratio of 14 dB The entry in the rms column is based on this the meter is based on the AD637 which is described in data sheets available from Analog Devices Instructor s Manual INTRODUCTORY DIGITAL EXPERIMENTS 16 Instructor s Manual PRBS generation TRUNKS No signals required answers to tutorial questions Ql Q2 Q3 Q4 Q5 Q6 Q7 Q8 use your oscilloscope delayed time base facility if available Otherwise endeavour to insert a delay between the
26. what has been decided to be the optimum decision instant An instantaneous comparator may detect a multiplicity of transitions within a symbol interval if there is a high level of ringing eg due to a sharp cutoff filter with large phase distortion or noise Q2 timing jitter the decision instant has a random variation relative to the ideal The result is that a decision may occur at a value that has a reduced margin relative to the noise baseline wander see answer in previous experiment Q3 as per text book Line coding TRUNKS No signals required duobinary encoding There is no procedure given for investigating duobinary encoding although this is incorporated in the line coding modules You may wish to add something to the experiment to cover this topic Instructor s Manual 19 answers to tutorial questions 1 the answer to this question could fill a book Topics would cover e spectrum shaping can minimize energy near DC especially for telephone line applications e enhancement of timing information for timing recovery at the receiver e elimination of the DC component to control baseline wander eliminate waste of power and so on Q2 processing delay by both the LINE CODE ENCODER and LINE CODE DECODER modules Q3 reduction of baseline wander qv Transmission systems are often AC coupled and so a DC component would be undesirable ASK amplitude shift keying TRUNKS No signals required
27. 0 BPSK binary phase shift 22 Signal Constellations E E E ereaaveneed ct 23 Sampling with 1 1 23 PEM encoding E ENE E E AEE 24 PEM dec dihg ea 25 Delta mod latioi ne E 26 Delta demodulation 27 Adaptive delta modulation n i eeen E A E NE 27 Delt sigma mod lati onasini E EE O N R E E 27 FURTHER AND ADVANCED ANALOG EXPERIMENTG 29 Amplitude modulation method 2 31 Weavers SSB Generator ET E 31 Weaver s SSB 32 Carrier acquisition and the 33 Spectrum analysis the WAVE 34 Amiplifier overloads 36 Frequency division 37 Phase division 1 41 Analysis of the R E 43 Introduction to FM using VCO 45 and the synchronous 47 Armstrong s phase
28. 0 kHz To obtain 50 kHz you can divide the MASTER CLOCK by 2 and use the 60 kHz LPF Otherwise use a VCO carrier A copy of the 50 kHz carrier will be useful but certainly not necessary for comparison purposes Beware though if the phase between recovered and stolen carriers is measured either the 50 kHz or the 100 kHz example that phase shifts via TRUNKS will make the answer less meaningful 59 answers to tutorial questions Q1 the SQUARER would not be required nor the divider at the output Q2 there is a large DC component in the output of any squarer This is not required If present it could overload any following stages here a MULTIPLIER Q3 not essential But if as might be the case in practice the signal is accompanied by other signals and noise these could degrade the performance of the SQUARER in its main task DPSK carrier acquisition and BER TRUNKS Two signals a DPSK and a sinusoidal carrier DPSK A DPSK signal based on a carrier of f 50 kHz Obtain the carrier by divide by 2 of the 100 kHz TTL MASTER then filtering by a 60 kHz LPF module A TTL will not overload this module if the gain is set for an output within the TIMS ANALOG REFERENCE LEVEL limits There will be a large DC component This can be removed by passing the filter output via a PHASE SHIFTER module Alternatively the following MULTIPLIER may be set to pass AC only the NRZ M signal should have no DC component The message w
29. 34 ANALYSER This experiment introduces students to a new module SPECTRUM UTILITIES module This module is not in the TIMS BASIC MODULE SET It uses a centre reading moving coil meter movement preceded by a buffer amplifier filter The moving coil meter is indeed a mechanical filter and will not respond to an AC signal above say 10 Hz certainly not to 100 Hz It is this that gives the analyser its selectivity Instructor s Manual The measurement of the amplitude of spectral component takes a little skill in tuning the VCO to within 1 Hz or so of the unknown component and then estimating the peak reading of the meter as it oscillates slowly about the centre zero You might agree that to expect more than a 10 reading accuracy would be unreasonable It would seem natural then to include some electronics to capture and hold the peak meter reading This facility has been included in the module s circuitry But it must be used with caution and with the full understanding of the likely mis interpretations The sample and hold circuit connected to the signal across the meter has no frequency selective properties it will hold what ever voltage is present If the peak meter reading without the sample and hold and the sample and hold reading are to agree there must be only one component of significance across the meter This condition can be approximated provided 1 spectral components are separated by at least say 500 Hz 2 the a
30. 90 at the transmitter The demodulator recovers both channels simultaneously so that their messages appear added at the demodulator output Their relative amplitudes are determined by the phase of the demodulating carrier When the amplitude of one message is reduced to zero nulled that of the other is maximized In this way one or the other channel can be recovered independent of the other Thus it is not the wanted channel which is maximized but the unwanted channel which is minimized There cannot be more than three or more channels in the PDM system since it would be necessary then to minimize two or more simultaneously which would be impossible In the spread spectrum signal there are literally thousands of DSBSC signals The correct demodulating sequence contains thousands of individual carriers all at the correct frequency The identical messages from each of these DSBSC appear added at the demodulator output If the relative phasing of each of the components in the demodulating sequence is correct these contributions are combined so as to optimize the resultant amplitude Otherwise they will combine to produce a much smaller resultant The selection process is an enhancement of the wanted message unlike the PDM case where it is a minimization of the unwanted message Q4 Q5 Q6 more confusing to the interceptor See answer to Q2 Q7 Q8 one less clock to recover influences the spectrum Digital utility sub systems Thi
31. CODER module This has an in built message reconstruction filter System requires 12 slots if the optional WIDEBAND TRUE RMS METER is used TRUNKS none Convolutional coding 62 The experiment is divided into two parts A and B Depending on the time available each part could be considered as a separate experiment although this is not essential However it is essential that the experiment entitled BER and the noisy channel should have been completed at another time and during a full laboratory session before Part B be attempted That experiment itself depends upon the successful completion of other experiments Part B uses more modules than can be accommodated in a single TIMS 301 system Either a second TIMS 301 or a TIMS Junior will be required An alternative is to place the transmitter at TRUNKS with stolen clocks provided but this option has not been considered Without a doubt it is preferable for each student to have free and easy access to both ends of the system Each TIMS320 DSP DB module should already have installed in it the two EPROMs containing the convolutional decoding algorithms The four MEMORY SELECT jumpers should be in the upper A locations Jumper J1 near EPROM US selects the decoding algorithm for CODE 1 or CODE 2 please start in the L position this is CODE 1 The alternative option of access to a PC via the front panel SERIAL LINK is not considered in this experiment However i
32. Instructor s Manual to accompany Communication Systems Modelling with Volumes A1 A2 D1 amp D2 Instructor s Manual to accompany Communication Systems Modelling with Volumes A1 A2 D1 amp D2 Emona Instruments Pty Ltd HUM is a registered trademark of ABN 79 069 417 563 Amberley Holdings Pty Ltd 86 Parramatta Road ABN 61 001 080 093 Camperdown NSW 2050 a company incorporated in the State Sydney AUSTRALIA of NSW AUSTRALIA INTRODUCTION this manual This TIMS Instructor s Manual is intended to help you the Instructor use and supervise each experiment It contains notes on each experiment as well as answers to most Tutorial Questions Each of the experiments described in the four volumes of Communications System Modelling with TIMS Sufficient details are supplied including background information for the block diagram to be modelled and adjusted to perform as required Some experiments are written with the expectation that earlier experiments will have already been completed This is stated under the prerequisites Most experiments include some suggested measurements Depending upon the level of your course work you may find it necessary to add your own list of supplementary measurements the TIMS laboratory A well equipped laboratory will have all TIMS SYSTEM UNITS interconnected via the TIMS TRUNKING SYSTEM with one TIMS SYSTEM UNIT serving as the MASTER the remainder being
33. Its message can be recovered with an envelope detector for small SSB carrier amplitude ratio the distortion won t be noticed and a synchronous demodulator but the output from the latter is not dependent upon the local carrier phase This is a trick signal and should test the student s understanding note that the CSSB signal described above does not need an SSB generator for its production just ADD a small sinusoidal component from a VCO on say 102 kHz to a large 100 kHz carrier Adjust relative amplitudes to make a reasonably undistorted sinusoidal envelope say in the amplitude ratio 1 3 or less The student should be able to identify each of the signals identify the messages each carries For the case of the two tone message it is possible to identify the frequency of each tone for example filter off the lower of the two with the TUNEABLE FILTER then recover the envelope of the two tone message to obtain the difference frequency write an expression for each of the received signals no TRUNKS It is important to have a TRUNKS system for this experiment To expect the student to generate the unknowns as well as identify them is not satisfactory The unknown element is lost and setting up would take an unnecessarily long time and if badly done could lead to undesirable results Measuring of Easy in principle multiply the 100 kHz reference with the VCO output Filter with the RC filter in the UTILITIES module Me
34. ODER See the note re this module in the PCM DEMODULATION experiment answers to tutorial questions Q1 using the frame synch signal FS for oscilloscope triggering is useful for showing frames but if the sweep circuits miss a beat then adjacent frames will swap positions on the screen If the FS signal is divided by 2 this cannot happen Use the divide by two sub system in the BIT CLOCK REGEN module Block coding and decoding TRUNKS None WARNING Please note that the maximum clock rate for the block code modules is about 2 kHz for which frequency the modules have been optimized It is not safe to operate at a clock rate of 8 333 kHz Instructor s Manual 61 answers to tutorial questions Ql assuming errors are reasonable rather than catastrophic then the frame must have already been identified Thus the LSB is already without error The error detector is looking for errors in the bits representing the message Q2 because of the alternating 0 and 1 pattern in the LSB position adjacent frames will never be identical With a DC message every other frame is identical So a synch signal is needed which will look for every other frame Dividing FS by two achieves this If a divide by 2 sub system is not available you can use a divide by four see the experiment entitled Digital utility subsystems for other sources of digital dividers Block coding and coding gain comments Need a version 2 or above PCM DE
35. SLAVES Some experiments require that the MASTER generates signals which are sent to the SLAVES via TRUNKS For building very large systems a TIMS Junior can be added This will expand the capacity from 12 modules of the standard TIMS 301 to a total of 20 modules 100 kHz MASTER clock Unless there is a special reason for not doing so all TIMS SYSTEM UNITS in the laboratory should be synchronized to the 100 kHz clock in the MASTER TIMS SYSTEM UNIT 1 sometimes referred to as the Student Text messages Many experiments require signals to be derived from messages There are many sources of messages including analog messages a SPEECH MODULE will be introduced during 1999 AUDIO OSCILLATOR analog output from the VCO the 2 083 kHz message from the MASTER SIGNALS module two tone signals add any two of the above band limited recorded speech see below the lowpass filtered analog output from a SEQUENCE GENERATOR DC to suggest a telemetry signal in TDM say a ee ae digital messages Almost exclusively digital messages will come from a SEQUENCE GENERATOR module However there are exceptions bandlimited speech We recommend that you provide yourself with a source of taped speech as this is always useful for qualitative observations and is essential for some experiments two or more tracks are required for TDM and FDM Speech can be obtained by recording radio programs althoug
36. TIPLIER modules per student group then the solution is for you to provide a QPSK with which they can test their demodulator answers to tutorial questions Ql Q2 Q starting with a DSB signal whose bandwidth is twice that of the message from which it was derived the QAM signal adds a second channel occupying the same bandwidth and spectrum location Twice as many message channels in the same transmission channel the required phase is that between the resultant phasors of the two DSBSC There are typically no components at these frequencies So a direct measurement is not possible We make indirect measurements by noting the phase difference between the two carriers which will null each DSBSC at the output of a product demodulator channel discrimination is determined by the phase difference between the two suppressed carriers The unwanted channel is nulled The other channel will be of maximum amplitude and SNR the noise is phase insensitive if the two DSB were at 900 If they were at 809 say nulling can still be achieved but the wanted amplitude would be slightly reduced by a factor of sin809 sin90 this is quite small Smaller deviations from 90 are relatively unimportant although only one channel demodulator was modelled had there been two they would have acted independently The carrier phasing adjustment of one is independent of the phasing adjustment of the other there being a 90 phase difference th
37. UNKS Two or more DSBSC at least one of which is based on a 100 kHz carrier and one which is not Costas loop filter it would be preferable if the RC FILTER in the UTILITIES module had a longer time constant Being near 3 Hz it is not really suitable as a loop filter in this application A later version of this module may have a jumper on board to parallel in a larger capacitor In the meantime make the message frequency as high as possible There is a preferable loop filter in the BIT CLOCK REGEN module The SPECTRUM UTILITIES module is very useful for measuring a DC voltage in the presence of AC components Supply one if available answers to tutorial questions Ql Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 the squaring loop a PLL preceded by a squarer passes DC but if too narrow acquisition takes longer not only acquires a carrier but adjusts it in the I arm to maximize the output amplitude this is in fact the message so Costas acts as a demodulator in addition to acquiring the carrier yes the third multiplier the roles of the I and Q arms will be reversed the roles of the I and Q arms will be reversed during pauses in speech there is no signal It would depend upon the dynamic properties of the loop as to how long a pause could be tolerated before performance was compromised if the passband is not flat then sideband amplitudes will not be equal and complete cancellation is not possible differential en
38. aE cos ut cos at Instructor s Manual 35 Now we are only interested in components near so it is not necessary to carry out the complete trigonometrical expansion Observe that the original signal is present in the output scaled by aj The distortion components represented by cos ut form the message to a DSBSC on both and 3 which we know will come from the cos mt term We only need consider the first of these DSBSC Thus components near are 3 ae cos u t cos u t cos Lt cos at Expanding this further the spectrum near 0 is frequency amplitude See the experiment entitled Amplifier overload Q4 only turn on HOLD when within a few Hz of the term being sought Keep turning the HOLD facility off to ensure the correct signal is being held Amplifier overload TRUNKS No signals specified answers to tutorial questions Q1 the tones are often kept close to ensure that the intermodulation products won t be spread apart too far It also makes the appearance in the time domain more like a DSBSC However there are no hard and fast rules Other applications make the two tones far apart and not necessarily of equal amplitude Q2 the DSBSC method of making a two tone signal is very convenient It enables them to be moved across the spectrum by varying the frequency of the higher frequency term and their separation by varying the low frequency term At the same time it provides a signal
39. age from being aliased into the passband Q2 samples would have been taken at a different time This will introduce a phase angle into each of the AC components of s t but not the DC term Hence each of the components of the sampled signal spectrum will be changed in phase except the wanted message term which arose form the produce with the DC term of s t So there will be no change to the recovered message Instructor s Manual Q3 the switching function will have a faster pulse rate so the unwanted spectral components will move further away from the edge of the reconstruction filter Since the pulse width remains fixed the ratio dt T is increasing Figure 1 A of experiment the amplitude of the recovered component will increase PAM amp time division multiplex TRUNKS You must provide a four channel TDM signal Two channels of a TDM transmitter may be generated by the arrangement shown below The sampling rate will be 8 333 kHz so a frame occupies 120 us If you make the guard time between samples equal to the sample width then the sample width will be 15 us But it could be made smaller say 8 us and the active channels bunched to one end of the frame to show more channels could be added say 8 in total if required PULSE MASTER SIGHALS GEHERATOR 100kHz sin 100kHz 100kHz TTL sample 95 message sin two channel TDM two channel TDM This two channel TDM can be combined in a
40. al so its bandwidth needs to accommodate transmitter frequency drifts It is harder to design more complex to build difficult to make tuneable Instructor s Manual 33 The PLL has a fixed frequency LPF easy to design the PLL is tuneable and will track The PLL output is of constant amplitude remains there if signal fades momentarily Q2 the signal has major and minor envelope peaks Let the major peak to peak amplitude be P and the minor Q Knowing the spectrum is as illustrated below and given the carrier is in phase with the sidebands then it follows that 2A B P and 2A From Figure 1 P Q 2 2 2 0 Thus A 1 05 B 0 1 Relate the carrier peak amplitude to the signal peak amplitude carrier level 2010910 4 2 This evaluates to about 26 dB Q3 the tuning of the PLL is by an oscillator simple To tune a BPF is far more difficult The design of and realization of a BPF is more complex than that of a LPF Q4 Q5 the two signals are a carrier on 100 kHz and in effect a DSBSC The DSBSC from TRUNKS can have undergone delays phase shifts different to those experienced by the system 100 kHz clock So these two may no longer be in phase at a remote point Q6 Q7 the PLL will have constant amplitude will be present during fades although may drift off frequency if the fade is too deep will track the signal if it drifts in frequency Spectrum analysis the WAVE
41. an early appreciation of the need to choose a suitable signal for oscilloscope triggering A suitable signal must at least be of constant amplitude That is it should not change as the various adjustments are made to signal levels throughout the model This obviates the need for constant re adjustment of the triggering circuitry l available 1999 Instructor s Manual rev1 1 1 Typically this signal will be periodic The oscilloscope trigger controls must be set to accept this external trigger signal and preferably switched to auto mode if available answers to tutorial questions 1 the depth of null needs to be described with respect to a reference This is generally the amplitude of the wanted term Here there is no wanted term So why not use the sum of the two components at the ADDER output as the reference This makes the calculation easy since after measuring the null the difference just flip the 180 degree switch on the PHASE SHIFTER front panel and turn the cancellation into an enhancement which displays the sum as required It is usual to express this in decibels thus depth of null 20 log 9 sum difference If the two signals out of the ADDER are Vj cos t and V gt cos t 0 then when the signals are subtracted the difference amplitude is difference IM cos a V sin a sum cos sin a Setting V7 1 01 Vz and a 1 degree then
42. ance of the receiver from that achievable with an input image rejection filter But in the measurement situation we have control over the noise we merely have to add about half as much at the input to the receiver as would otherwise be required for a given amount at the detector input where SNR is to be measured This is a problem but can be accounted for Note that a 2 dB change at the NOISE GENERATOR attenuator would make more than a 2 dB change at the measurement point How much more In an attempt to obtain enough noise at the detector input it is tempting to amplify it at source But in the present example this would be before bandlimiting Beware of overloading the multipliers Spread spectrum DSSS amp CDMA 66 TRUNKS See experiment A DSSS signal where the PN sequence is derived from a VCO near 100 kHz This will be a signal sharing the channel but not part of a CDMA system Instructor s Manual answers to tutorial questions Ql Q2 each of the spectral lines of the PN sequence locates the carrier of a DSBSC signal Each of the sidebands of this DSBSC is a frequency translated version of the baseband message one erect the other inverted Depending upon the spacing of the lines in the PN sequence these DSBSC may or may not overlap If they do not overlap the job of an interceptor eavesdropper is simplified Q3 there are two DSBSC in a phase division multiplexed signal The two DSBSC are phased nominally at
43. asure the frequency of the filter output But problem 1 the analog input of the COUNTER is AC coupled So put the filter output through the COMPARATOR and connect to the counter TTL input This works fine problem 2 the counter needs a full second to respond for frequencies 1 Hz and above even longer for smaller frequencies One soon finds the method unsatisfactory as df becomes small solution the two sliding traces on the oscilloscope is by far the preferred method of observing and estimating the frequency difference answers to tutorial questions Q1 try it Different observers tend to report different opinions and text books have quite differing views Instructor s Manual not observable no observable distortion but amplitude will momentarily fall to zero every 5 seconds 10Hz probably not observable not probably not observable horrible 100 Hz still intelligible especially if twee _ speech is translated high Q2 the synchronous demodulator as examined is not a true SSB demodulator although it can demodulate SSB if an SSB is present alone The synchronous demodulator examined in the experiment has a window either side of the local carrier frequency of width equal to twice the bandwidth B of the associated LPF that is from B B A true SSB demodulator must be able to look at one side of the carrier whilst ignoring any signal including noise o
44. below 50 Hz Such low speeds make standard oscilloscope viewing difficult but not impossible choice of two frequencies If spectral analysis facilities are available much work can be done regarding spectrum control by suitable choice of the two signalling frequencies as a function of bit rate This would require the ability to generate such frequencies a facility which TIMS currently cannot conveniently provide In addition some more appropriate filters might be required This is perhaps getting too far from the intent of TIMS However if users desire such a capability a suitable module or modules could be developed for the purpose Suggestions are welcome TRUNKS See above 1 asine wave in the range 1 to 5 kHz say 2 5 kHz 2 an FSK signal f on 2 083 kHz and f nearby say 2 5 kHz 3 perhaps a low frequency bit clock would be a help answers to tutorial questions Ql Q2 Q3 Q4 Q4 Q5 Q6 it is the sum of the two spectra carrier frequency not stable enough for communications applications otherwise OK the filter centre frequencies must be far enough apart for the filters to be able to separate the two tones There are optimum spacings for minimum ISI Signalling speed is limited by the inequality that requires 1 T lt B where B is the filter bandwidth and T the period of the clock bandwidth would set a maximum bit rate Minimum separation would be determined by the response shape But depending on
45. bove you should be aware of it if problems arise recording tips take care with setting up for the recording Avoid earth loops and other sources of mains hum and noise Check the final result by doing the experiment yourself Unnecessary noise and extraneous signals must be minimized so the student will not be distracted at all times during the preparation of the FDM signal and later when trunking it avoid any form of overload and thus crosstalk make sure all messages are different and bandlimited 300 to 3000 Hz Whilst music makes a change from speech it tends to have occasional high level passages which can overload any part of the system and thus cause crosstalk if the tape recorder has any noise minimizing schemes disable them These often assume certain properties of the signal to be recorded based on the assumption it will be music which the FDM signal does not have determine the amplitude V k of a sine wave into the recorder which just overloads the tape recorder Do not rely on any built in record level metering but play back the tape and observe the waveform of the output When making recordings ensure that the peak signal level into the recorder is at least half Vok to avoid non linearities and thus crosstalk between channels ensure all SSB generators are well balanced especially for carrier leaks ensure the peak amplitudes of each channel into the ADDER of Figure 1 2 or 3 are equal when combining the s
46. chemes of Figure 2 and 3 make sure peak amplitudes into the combining ADDER are equal make sure the peak amplitude of the FDM into the TRUNKS system does not exceed the TIMS ANALOG REFERENCE LEVEL variation of the experiment If the amplitude of the FDM from TRUNKS is too large it will overload the demodulator This is investigated in the experiment circa Task T7 It will be the MULTIPLIER which will overload 40 Instructor s Manual You may prefer to simulate a channel overloading by interposing the COMPARATOR from the UTILITIES module as was used in the Amplifier Overload experiment This has a more controllable non linear characteristic answers to tutorial questions Ql Q2 Yes The DSBSC demodulator will examine a window 3 Hz either side of its local oscillator frequency In each case there is a wanted channel on one side and nothing on the other The carrier frequencies used would be 8 and 16 kHz Q3 in principle the true SSB demodulator will have a signal to noise advantage since it will not add the noise from the unused sideband Q4 Yes But a DSBSC demodulator scheme needs a synchronous carrier Carrier acquisition would be a problem but pilot carriers could be sent Alternatively a master carrier could be recorded on the tape say 16 kHz from which the others could be derived by division Carrier acquisition circuitry would need to track the carriers they would move due to the non constant speed of the tape rec
47. coding Instructor s Manual Q10 would be hard to notice the difference The loop would still endeavour to produce no output from the third multiplier When this is achieved the wanted output would be reduced by a factor sin 80 sin 90 Instructor s Manual 53 54 Instructor s Manual FURTHER AND ADVANCED DIGITAL EXPERIMENTS 55 56 Instructor s Manual BER measurements in the noisy channel Instructor s Manual The LINE CODE ENCODER and LINE CODE DECODER and the DECISION MAKER have been designed to work together as compatible interfaces between the digital input sequence from a SEQUENCE GENERATOR and output sequence from the DECISION MAKER of the system with an analog channel in between The DECISION MAKER has been optimized to operate at or around 2 kHz and so the 2 083 kHz message from the MASTER SIGNALS module is an ideal clock source A preferred method of deriving the bit clock is to use the LINE CODE ENCODER and its compatible partner the LINE CODE DECODER when working with a complete transmission system Part of their coding job is to convert TTL level signals to TIMS compatible analog levels The LINE CODE ENCODER is driven by the 8 333 kHz TTL MASTER SIGNALS clock The LINE CODE ENCODER provides an 8 333 kHz divided by four TTL output which of course is at 2 083 kHz This serves as the bit clock for the system and is a rate ideal for the DECISION MAKER no LINE CODE ENCODER This
48. d jumpers e M LEVEL ENCODER J3 to NORMAL position e M LEVEL DECODER the range to HI for an 8 333 kHz clock TRUNKS No signals required answers to tutorial questions Ql Q2 Q3 Sampling with sample and hold TRUNKS None But you may like to supply speech if SPEECH modules not available answers to tutorial questions Q1 an engineering estimate is that one can see the presence of the smaller amplitude component when it is about 30 dB below the larger component by CRO observation Certainly a 40 dB ratio is not visible Q2 put the output of the reconstruction filter into an ADDER Combine with the original message via a PHASE SHIFTER Null the message remainder will be at least 40 dB below all that TIMS claims probably better than 50 dB below Can t see any distortion due to the S amp H process Q3 the slot band is the sum of the passband width and the transition band That is from DC to the start of the stopband Nyquist assumes a brick wall characteristic zero transition band So the re written Nyquist criterion assuming a bandlimited message would say that the sampling rate should be twice the message bandwidth plus the transition bandwidth Including the effects of practical message bandwidth limiting the sampling rate would be twice the message bandwidth plus twice the transition bandwidth assuming similar filters for bandlimiting and reconstruction The first of these engineering definitions is applicabl
49. d signal amplitude only The system is adjusted to null the output from the unwanted channel The amplitude of the signal from the other channel will be proportional to the cosine of the phase angle between its resultant and the carrier At best the angle will be zero and the multiplying factor will be cos 0 1 When off by 45 degrees the factor will be cos 45 0 707 Thus the amplitude ratio will be 0 707 and the power ratio will be 20 log 9 0 707 which represents a degradation of 3 dB Q2 the system depends on nulling the unwanted channel what ever remains is declared as wanted Thus a three channel system would have to null two channels at once This is not possible Q3 consider each channel to carry a single tone of the same amplitude Assume the two channels are at 90 degrees at the transmitter From the answer to Q2 above show that the crosstalk ratio would be ratio unwanted to wanted or approximately Thus a degree error would make the crosstalk power ratio approximately 35 dB Q4 the alternative arrangement as suggested in this question enables each channel to be adjusted independently whereas in the arrangement of Figure 2 the phase shifters will be interactive So which arrangement is preferable What if the phasing was altered at the transmitter How would the two compare 42 Instructor s Manual Analysis of the FM spectrum answers to tutorial questions Q1 perhaps but the ampl
50. e peak amplitude would be V2 volt one less clock to recover the modulator is based on a multiplier Typically multipliers are susceptible to overload which gives rise to the generation of unwanted spectral components non linear distortion Multi level QAM and PSK 64 Read the Advanced Modules User Guide which includes details of the following adjustments e set the on board jumper J3 of the M LEVEL ENCODER to the NORMAL position e set the on board range jumper of the M LEVEL DECODER to HI for an 8 333 kHz clock e if BER measurements are to be made check the adjustments of the on board trimmers RV1 and RV2 according to the procedure described in the Advanced Module User Manual e make sure the Z modulation facility has been set up to suit the oscilloscopes in use rack space If each student position has a TIMS Junior as well as a TIMS 301 then there is sufficient room to model both a transmitter and a receiver However the experiment has been written as though such extra space is not available Even so the complete receiving system including noise and error rate measurement requires 13 slots Instructor s Manual It is possible to manage with 12 slots however by noting that once the signal to noise ratio has beeen set at the M LEVEL DECODER the WIDEBAND TRUE RMS METER can be removed and replaced by the ERROR COUNTING UTILITIES module since these two are not required simultaneously DC offsets You may co
51. e when testing with a single sinewave f since the message appears to be brickwall filtered to fo Instructor s Manual 23 Q4 cancellation of the aperture effect with an equalizer possible because the reconstruction filter introduces inear frequency distortion See sinx x correction PCM encoding TRUNKS No signals required oscilloscope triggering Using the frame synch signal FS for oscilloscope triggering is useful for showing frames but if the sweep circuits miss a beat then adjacent frames will swap positions on the screen Sometimes this is a nuisance If the FS signal is divided by 2 this cannot happen Use the divide by two sub system in the BIT CLOCK REGEN module WARNING Please note that the maximum clock rate for this module is about 10 kHz It is safe to operate at the clock rate of 8 333 kHz for which the module was optimized but no faster answers to tutorial questions 1 the clock rate to the module is 8 333 kHz Frames are 8 clock bits wide The input signal is sampled once per frame So the sample rate is 8 333 8 or about 1 kHz To satisfy the sampling theorem the message must be limited to frequencies below half the sampling rate or less than 500 Hz Q2 the module is driven by an external clock Samples of the input analog waveform are taken every eight clock bits and coded A to D into either 4 bit or 7 bit words These words are fitted into a time frame together with a frame synch pulse FS w
52. ecomes negative and the formula can still be used Q2 this is not easy to explain in a few words Set it up with TIMS and work out an explanation yourself the phasor diagram represents AM signal y t A 1 m cosut cosot This can be represented in phasor form as in the figure When 180 degrees the resultant gives the trough amplitude which equals Q 2 Q 2 A Am 2 Am 2 When 0 degrees the resultant gives the peak amplitude which equals P 2 P 2 Am 2 Am 2 Substitute for P and Q from these two equations into the right hand side of eqn 9 and show it is equal to m Q4 the AC DC switch is a dual switch It controls signals into the MULTIPLIER It passes AC only in the AC position but AC and DC in the DC position In the present vase removal of the DC term in the message eliminates the carrier term see eqn 1 so the output signal changes to DSBSC Q5 in the phasor diagram the three amplitudes will be 4 1 and volt peak The total power in 1 ohm will be 0 75 watt Thus the rms meter will read V 0 75 volt Q6 a reversal of the phase of the carrier term Envelopes TRUNKS No signals specified answers to tutorial questions Using phasors to predict envelope shapes is a useful skill to acquire The technique is often not fully appreciated until a non standard case is examined such as that of Q2 Q1 2 draw the phasors at various positions throughout a period of
53. entation macro model 58 This Chapter is not an experiment Instructor s Manual Bit clock regeneration TRUNKS One or more line coded signals from a SEQUENCE GENERATOR and LINE CODE ENCODER combination should be sent via TRUNKS A 208 kHz sinusoid is an option see the Experiment for details answers to tutorial questions Q1 if the amplitude is varying then there must be two or more components in the signal Assuming the largest component is at carrier frequency smaller components can vary the amplitude and the zero crossings Call these smaller components sidebands These can be resolved into in phase and quadrature pairs or symmetrical and anti symmetrical components In phase components introduce linear modulation and do not influence the zero crossings Quadrature pairs introduce angle modulation non uniform zero crossings and timing jitter Q2 non uniform zero crossings Q3 amplitude jitter will not influence the zero crossings Q4 Q5 Carrier acquisition TRUNKS Instructor s Manual modulated A modulated signal on a carrier of 100 kHz and another at 50 kHz are required from which carriers will be recovered A carrier of 50 kHz is unusual but convenient for the acquisition VCO which with the squarer will be tuning to 100 kHz You should nominate which signal is which The signal on 100 kHz should have a carrier component present but this is not necessary to be avoided in fact for the one on 5
54. envelope suffers severe distortion See multi path propagation Envelope recovery TRUNKS No TRUNKS signals are specified If no SPEECH modules then supply it if you consider it useful You could also send signals with envelopes to be identified These could be derived from speech or tone or tones whose frequencies must be determined answers to tutorial questions Q1 amp 2 fora fullwave rectifier in the ideal envelope recovery circuit the requirement is that lt 20 uy For a half wave rectifier it is that u lt y This is derived by reference to Figure 2A of the experiment In discussing envelopes and distortion in the output of envelope recovery circuits one should be clear as to what is required By definition there is no distortion of the envelope recovered with an ideal rectifier filter combination assuming an appropriate filter If one is actually looking for the message then that is dependent on the nature of the signal Thus for an AM signal it is necessary that m lt 1 If m gt 1 then the envelope will be a distorted version of the message Thus in this case the output from an envelope detector will be a distorted version of the message but not a distorted version of the envelope The square law detector can be analysed by considering its transfer function to be 2 Cout ein For the case of the input being the AM signal y t 1 m cos ut cos at then the audio output terms
55. er to compensate Q2 the two waveforms as seen on an oscilloscope look at a first glance to be the same The Q3 Q4 Q5 a number of steps in each waveform is exactly the same since sampling rate is the same The difference is in the accuracy with which the amplitude of the steps represents the amplitude of the sampled message The more accurate 7 bit the less will be the distortion of the recovered message nothing b increase the number of quantizing levels This requires a faster bit rate and so more bandwidth if the word rate is to remain the same But see next Question Q6 more quantizing levels price to pay To keep the same word rate the bit rate would need to increase This means more bandwidth BUT in the present system the bit and word rate remains the same because of the way the frames are formatted For both 4 bit and 7 bit coding schemes the words are all 8 bits wide So the data rate is the same the channel is not used efficiently Q7 the advantages show up when speech is the message Intelligibility measurements are non trivial so you may or may not be convinced by a quick listening test Delta modulation 26 There are many parameters which can be varied in the delta modulator and investigating the effects of all of them is time consuming and perhaps unnecessary If you have certain preferences you could instruct the students to concentrate on these rather than following the experiment as written
56. everely impaired whereas the output from a synchronous demodulator exhibits only linear distortion effect upon intelligibility may be mild SSB generation the phasing method TRUNKS Speech is interesting Send it if students do not have SPEECH modules answers to tutorial questions Q unnumbered There is no reliable method of positively identifying in the time domain a signal as either SSB or DSBSC when derived from speech Q1 put the PHASE SHIFTER in the path to the other MULTIPLIER There are others Try not to implement a change which makes two reversals thus cancelling each other Q2 Q3 frequency division multiplex FDM systems were used in vast quantities for analog telephone systems before being superseded by time division multiplex TDM One such system used channels spaced by 4 kHz in the range 60 to 108 Hz Channel separation was accomplished with bandpass filters the cost of which was low due to the quantities involved There were thus filters located either side of the TIMS 100 kHz carrier and these were used by TIMS as LSB and USB filters These filters are no longer easily obtainable Q4 by eye one would have to detect when a circle had degenerated into an ellipse Say a difference of 5 in the major and minor axes Not good enough to detect 1 degree errors but OK for a quick check for major problems 6 Instructor s Manual Q5 no The Hilbert transformer is a mathematical transform which conceptually is a si
57. for oscilloscope synchronization the low frequency term But the weak link is the MULTIPLIER which is likely to be the cause of intermodulation distortion The test signal will thus be impure before it even reaches the circuit under test Combining two tones with an ADDER is a much more linear process 36 Instructor s Manual Q3 adjustment of the relative amplitudes of the two tones to equality is a simple matter of obtaining a symmetrical DSBSC like display on the oscilloscope Only when they are of equal amplitude can they add to zero and produce a null in the envelope Q4 unstated in the question and possibly in the advertisement is the accompanying amount of distortion at the quoted power output Presumably the 60 watt amplifier will generate more distortion when delivering an output of 60 watt than will the 50 watt amplifier when delivering 50 watt One business may sell for the HI FI market and the other PUBLIC ADDRESS applications The requirements of these applications are quite different Frequency division multiplex TRUNKS An FDM signal is required You can model the generator with TIMS and connect the output to TRUNKS The experiment says the signal can be recorded on tape So it will be more convincing if you do this and then play the tape back during the experiment It will certainly save time for future presentations of the experiment Reference to the experiment will remind you that two schemes are sugge
58. h such signals tend to have wide variations in level A better solution is to make a voice recording under studio conditions from suitable non copyright text A segment of only a few minutes is adequate This should then be copied onto a second cassette and repeated for as many times as required so that the second cassette will last for the duration of the experiment By repeating the text the student soon becomes bored with the contents and can then concentrate on the experiment Two or more speech channels are useful for multiplexing experiments The topics chosen for each track should differ as widely as possible This enables the student to identify from the context which channel is being recovered If you are preparing your own tapes then readings from technical fictional sporting texts and so on can satisfy this requirement The speech must be bandlimited to the range 300 to 3000 Hz Attenuation of frequencies above 3 kHz is essential for many experiments eg TDM and FDM A TIMS TUNEABLE LPF can be used for this purpose whilst making the recording To cut the low frequencies a 3 dB corner at around 300 Hz is suggested This can be achieved with an external RC network inserted in the signal path during the recording Unless the low frequencies are attenuated as well as the high frequencies above 3 kHz the intelligibility will suffer tutorial questions The initial aim of the Tutorial Questions was to ask questions which could be
59. hich is 8 clock bits wide The frame slots are numbered 0 to 7 as shown below For the case of a 4 bit word its four bits fit into the frame as shown Do being its least significant bit in slot 1 Slots 5 6 and 7 are empty Slots 5 6 and 7 are occupied when 7 bit code words are used a frame gt _ D D D D Fs bit 7 bit 0 MSB LSB Q3 the sequence of frames could be stored and re transmitted at a slower rate This would be advantageous under poor SNR conditions for example 24 Instructor s Manual Q4 all samples of a DC message are identical If the oscilloscope is synchronized to show one frame per sweep to overlap consecutive sweeps then the display will be stable Consecutive samples of an AC message will differ and so when overlaid the display will not be stable Q5 Q6 if alternate frames of a PCM signal were omitted then it would appear that the input signal A say 15 being sampled at half the original rate A second analog message B say sampled in the same manner but with its frames timed to interlace with those of the first could be added to it The message A could be recovered by a decoder if the decoder examined only the A frames likewise for the B message The TIMS PCM ENCODER has an operational mode TDM were alternate frames can be omitted two such encoder modules can be set up to operate as master A say and slave
60. icant sideband criterion 6 pairs of sidebands would be included so B 12 kHz Q4 the counter counts level crossings in a given time slot Since the zeros are not evenly spaced in time neither will be the count made by the counter However the count will depend on ratio of carrier to message frequency message frequency magnitude of the frequency deviation as well as the period of the count Try it Q5 Q6 a forn 1 the first zero for B between 3 8 and 3 9 the second between 7 and 8 b for n 3 there is a zero for between 6 3 and 6 4 FM and the synchronous demodulator TRUNKS An FM signal derived from a single tone based on a 100 kHz carrier Armstrong See the experiment entitled FM Deviation Multiplication for a suggested method of generation Note that you will need an FM UTILITIES and a 100 kHz CHANNEL FILTERS module The latter should be version 2 or later V2 marked on circuit board Filters shipped pre June 1993 were V1 but this was not marked on the board The message tone should lie somewhere in the range say 500 to 1000 Hz This is high enough to simplify resolution with the WAVE ANALYSER not so high as to exceed the bandwidth of the filters you will need to use with the multiplier stages following the Armstrong modulator Use a frequency deviation to produce at least a few sidebands of significance Ensure there is a component at carrier frequency so that the demodulator will have a DC output The aim of the ex
61. ignals The object is to demonstrate that when the demodulator is set up to receive a LSB then a 99kHz signal will produce an audio output 1kHz while there will be no audio output with the 102kHz input signal answers to tutorial questions Q2 in principle complete cancellation of a single unwanted sidefrequency can be achieved if we have control of both the amplitude and phase of the signals into the summer For this to be achieved over the whole sideband then the two filters would need to be identical else degree of cancellation of the unwanted sideband will vary with frequency For the single filter case the possibility of overload of the multiplier is increased since both the difference and the sum frequencies will be present at its inputs Instructor s Manual Carrier acquisition and the PLL TRUNKS 1 a nominal 50 kHz sinusoid This avoids the need for students to have two VCO modules 2 a DSBSC based on a 50 kHz carrier with a single tone message 3 A DSB based on a 100 kHz carrier derived from a single tone with some pilot carrier added This is conveniently made by using as the message to a DSBSC generator a tone plus a DC voltage which have been combined in an ADDER The DC voltage will provide the carrier term Keep the ratio of the DC to AC amplitude small so that the pilot carrier will be relatively small See Figure 1 of the experiment 4 a DSBSC based on a 50 kHz carrier with a speech message Th
62. ill be from a SEQUENCE GENERATOR set on a long sequence and clocked at 2 083 kHz fy 24 kHz A suggested model is illustrated below SEQUENCE LINE CODE TUHEABLE MULTIPLIER LPF ENCODER 50 kHz sinusoidal carrier to TRUNKS TTL 8 333 kHz DPSK generator carrier and bit clock Students will not normally have a 50 kHz sinusoidal carrier signal so this should also be sent via TRUNKS answers to tutorial questions Q1 the SNR 15 adjusted at the DECISION MAKER input If it is of comparable power to the signal at this point then it will be much larger prior to bandlimiting which occurs 60 Instructor s Manual immediately before the DECISION MAKER Hence there is a danger it may overload some parts of the system between the noise source and the filter The noise into the MULTIPLIER is wideband with components near 50 kHz desired but also around 100 kHz These latter components will also be translated into the passband of the bandlimiting filter this is the image response Being white noise it will be combined with the desired noise and appear as noise Q2 impure local oscillator means more noise or other unwanted components out of the MULTIPLIER PCM TDM TRUNKS None Speech perhaps Or students may have SPEECH modules WARNING Please note that the maximum clock rate for the PCM modules is about 10 kHz Thus it is safe to operate at the clock rate of 8 333 kHz PCM DEC
63. input signal be and the local oscillator be cos Wot For simplicity we assume these are very much greater than 2 7 After taking the product of the high frequency signals the difference frequency component which is the only one likely to pass through the filter is cos 00 Provided 00 lt 2 7 B then the difference signal will pass through the lowpass filter Thus can be either above or below Wg for this condition to be met and so it can lie in a window of width 2 B Hz Q2 let the oscillator be cosmt and the input signal cos t this is above rad s Instructor s Manual 11 Although a single filter is shown at the output of the summer this could be replaced by two similar lowpass filters one each in the P and Q arms Then the lowpass terms into the summer will be from the P arm gt cosut from the arm gt sinut which is changed to cosut by the 900 network There is a finite output if these are subtracted Now suppose a signal appears at the input lower in frequency that w rad s For the same circuit conditions the lowpass terms into the summer will be from the arm gt cost since cos ut cos ut from the Q arm gt sinut which is changed to cosut by the 909 network There is no output since the summer is acting as a subtractor Q3 you can show the maximum is broad ill defined but the minimum is narrow well defined Q4 if there is interference on one sideband o
64. is is required for the final part of the experiment which you may elect to demonstrate instead see below Since only three TRUNKS channels are available the first signal will need to be removed towards the end of the experiment and replaced by the fourth The VCO is described in the User Manual Pertinent information for this experiment is as follows To obtain a 50 kHz output before plugging the VCO in locate the on board control RV8 toggle the on board switch SW2 to FSK mode plug in the VCO leaving two free slots on the left for hand access to the board connect a TTL HI to the DATA input select HI on the front panel toggle switch Dy en PS OD SO rt tune the on board control RV8 the front panel f control is inoperative in FSK mode for 50 kHz using the FREQUENCY COUNTER the experiment You may elect to instruct students to miss the final part of the experiment which requires them to have three MULTIPLIER modules Instead you could set up a demonstration for them to examine The loop filter for the PLL is in the UTILITIES module This has a 3 dB point at around 3 kHz It would be preferable if this were lower in frequency There is such a loop filter in the BIT CLOCK REGEN module If you have any such modules please supply them to students Version 2 of the UTILITIES module offers a choice of corner points via an on board jumper answers to tutorial questions 1 the BPF will not track the incoming sign
65. itude spectrum is missing the sign phase of the Bessel coefficients as they are often called So it is not a straightforward exercise Some trial and error might help Q2 the amplitude E is a peak value so Q3 in En 71 ower Watt 2 50 This is true for any value of B but is obvious for 0 Should you be tempted to obtain the power by summing the contributions from individual components don t forget there is a single term at carrier frequency but the others come in pairs considering the power in an angle modulated spectrum one uses the fact that it is independent of the degree of modulation The power in the unmodulated carrier B 0 is taken as a reference When power ratios are involved there is no need to convert amplitudes to rms values Neither is there a need to include the amplitude as in eqn 7 To answer the question one refers to Bessel tables and takes the reference as 1 Then from the column for the appropriate value of B one adds the squares of the entries until the sum reaches the desired amount ie 0 9 in the present case The precaution necessary is that the carrier contribution is included once whereas subsequent entries are included twice to account for upper and lower sidefrequency pairs So for 1 the entries are side relative relative cumulative of frequency amplitude power sum of reference powers power 7652 5855 5855 58 o
66. lack of linearity between input and output In some cases two measurements would not be sufficient in the case for example of fortuitous cancellation this was not discussed in the aforementioned experiment Q3 assuming uniformly distributed noise power then the noise power would halve thus the meter reading will be reduced by 1 V2 Q4 5 2 1 2 volt Q5 25 2 volt Instructor s Manual b V 25 5 volt Q6 the square of the meter reading will give the power in one ohm This does not lead to a linear conversion factor Q7 on noise alone the meter would read 0 458 volt rms Thus SNR 6 8 dB Changes of less than 10 would be more difficult to read Thus unless the SNR is say 10 dB or less the measurement of small changes to SNR can become difficult Q8 3dB b reduced to 1 V2 of initial value Q9 using a sine wave as a test signal will show a delay unless it is an exact multiple of 360 but cannot distinguish between delays of and n 360 Use the output from a SEQUENCE GENERATOR clocked at well below B Hz where B is the filter bandwidth Then individual bits can be identified and their transmission delay noted There are other methods Signal detection with the DECISION MAKER TRUNKS No signals required answers to tutorial questions 1 the clocked gated strobed comparator makes only a single decision during a symbol period be it correct or otherwise This decision is made at
67. module If the input to the QPS is the 2 083 kHz sinewave from the MASTER SIGNALS module then the P and Q outputs are in phase quadrature There is no phase trimming possible but this can be an advantage since it removes one degree of freedom in the setting up procedure The disadvantage is that the presence of the QPS module might make some students think they are modelling some sort of hybrid phase shift method of SSB generation One of the points often omitted from text books is the fact that Weaver s method if out of alignment places the unwanted products in the same band as those wanted so they do not interfere with adjacent channels The unwanted components being in your channel will seldom go unnoticed by you as they might if they fell outside it answers to tutorial questions An important feature to notice when analyzing Weaver s generator is the fact that the unwanted components fall on top of occupy the same spectral space as the wanted components Thus an imperfect generator degrades its own channel and not an adjacent channel Weavers SSB demodulator 32 TRUNKS A sinewave just above and a sinewave just below 100kHz would be useful as test signals These could come from two VCO modules say 99kHz and 102kHz These would simulate a 1kHz message on a LSB and 2kHz message on an USB of a 100kHz carrier If TRUNKS is not available then supply each student with a second VCO to generate either of the above s
68. more trials are necessary until alignment is achieved the counter counts transitions in one direction only Each bit of the sequence starts at a clock HI and lasts for a clock period If the sequence was a series of Os and 1s there would be half as many sequence transitions as clock transitions Thus the counter would display half the clock frequency If in the sequence some of the zeroes became ones or vice versa the sequence would contain less transitions it cannot contain more Thus the counter would indicate less than half the clock frequency The pattern is such that it reads half of half the clock frequency if the TTL sequence is either 0 or 5 volt it would generate half the power of a continuous 5 volt signal Thus power in one ohm would be 12 5 watt Thus rms value V 12 5 volt the clock period is 1 2083 s there are 2048 bits per sequence one bit per clock period So the delay is almost 1 second Instructor s Manual 17 Eye patterns TRUNKS No signals required answers to tutorial questions Q1 baseline wander is usually the result of an AC coupling It is accentuated when the data has long sequences of identical values which with AC coupling introduce a DC shift wander This upsets the decision device threshold Appropriate line coding can transform the data so that such long sequences are avoided Q2 they are meant to be filters satisfying the same channel allocation requirements This is more likely to be b
69. mplitude ratio of adjacent components does not exceed say 100 1 The second requirement is included to cater for the case where a small component is being measured in the presence of a larger adjacent one To aid the mechanical filter the buffer amplifier has a lowpass characteristic with a corner at about 10 Hz and beyond this an attenuation rising at 24 dB octave TRUNKS Unknown signals are required for spectral identification You can choose those signals which you think would be appropriate for your students Put them in both the audio and 100 kHz region Use an ADDER to combine signals from the VCO AUDIO OSCILLATOR and so on Suggested signals could be e two tone audio e alow audio frequency square wave e 100 kHz DSBSC plus an output from a VCO e etcetc answers to tutorial questions Q1 the analyser has a window 2 Hz wide This is tuned and down the input spectrum The point of the question is to show that the bandwidth of the instrument is the same as that of the filter Q2 the instrument is of the same form as a product demodulator for DSB signals It looks out over a width B either side of the centre frequency So its bandwidth is twice that of the associated LPF or 2 Hz Q3 before the signal reaches the analyser it is distorted new components are generated Where are they Let DSBSC ut cos at With this as the input the non linear amplifier output will be Cout QE cos ut cos at
70. n ADDER with another two channel TDM to make a four channel system Some phase shifting of the 8 333 kHz signal to one of the TWIN PULSE GENERATOR modules will be required More actual channels are unnecessary but more virtual channels can be implied by grouping these four down one end of the frame implying more inactive channels sampling pulses to analog switches this could simulate an eight channel system Instructor s Manual 13 Sources of suitable messages for each channel are discussed in the introduction to this Instructor s Manual answers to tutorial questions Q1 if the switching pulse is initially narrower than the sample then more of the sample will be passed to the reconstruction filter as it is widened and the amplitude of the reconstructed message will at first increase linearly until the switching pulse becomes wider than the sample From now on the recovered message amplitude will remain constant but the signal to noise ratio will decrease There will be an abrupt increase of crosstalk when the receiver starts accepting parts of the samples from the adjacent channel The above remarks assume the switching pulse is located in time so as to straddle as much of the wanted sample as possible Q2 there is a minimum pulse width that is obtainable from the module so this sets one limit If this width could be narrowed further then eventually the rise and fall times would become a significant part of the pulse width
71. n the other It must be able to select sidebands Q3 a small DC would appear at the demodulator output This would not be detected by ear Q4 refer to the diagram The top receiver uses a synchronous demodulator whereas the lower one uses an envelope detector Assume the input signal is free of any significant noise Provided there is only one input signal present then there would be no significant difference between the two audio outputs as observed by a listener Instrumentation might detect a difference in signal to noise ratio There is a significant difference between the two receivers however when it comes to selectivity Suppose the bandwidth of the LPF was 0 3 kHz and that the original AM signal had 3 kHz wide sidebands Consider the presence of a second signal of similar bandwidth but 10 kHz higher in frequency e upper receiver would frequency translate it both up and down The down products would not pass through the LPF since they would be in the range 7 to 13 kHz The audio output from the wanted channel would remain unchanged The bandwidth of the receiver assuming linearity is B the audio filter bandwidth e the lower receiver would be operating on the linear sum of the two signals The envelope of this combination would not be the sum of the individual envelopes It would bear little no resemblance to either envelope as an analysis will show Thus the audio output would be unintelligible The bandwidth
72. nd 3 This builds up the recording in two stages on two different tracks The two tracks can then be combined onto a single track of a separate high quality recorder or the two tracks can be combined in an external ADDER at the time of transmission via TRUNKS message two channel message Instructor s Manual message two channel message Figure 3 method 1 scheme 2 It is possible to record all FDM channels simultaneously on a single track according to the scheme illustrated in Figure 4 below Notice that one generator produces a lower sideband and the other an upper sideband message LSSB message GENERATOR message method 2 scheme 2 A less elaborate scheme for preparing the recording is suggested in Figures 5 and 6 This builds up the recording in two stages on two different tracks The two tracks can then be combined onto a single track of a separate high quality recorder or the two tracks can be combined in an external ADDER at the time of transmission via TRUNKS message two channel message Figure 5 Instructor s Manual 39 two channel USSB Message GENERATOR 12kHz A Figure 6 other recording schemes You may think up other schemes for preparing the FDM signal Be aware that the isolation between tape recorder tracks is often not great with stereo type cassette recorders Although this should not be a problem with the two track scheme suggested a
73. ngle input single output device It has an infinite bandwidth extending down to DC Practical realizations have a finite bandwidth which does not extend down to DC The QPS is composed of two complementary devices each with a single input and single output The phase shift through each device varies differently with frequency but at any one frequency these phase shifts sum to 90 a small typically equiripple error Q6 the arrangement of Figure 4 would be a quick check for serious errors The most realistic check is to use it in an SSB generator or receiver and to measure sideband suppression or rejection Q7 the requirement here is to find the ratio of the sum and difference of the two phasors V and where they are degrees off cancellation sum magnitude V1 V2 2 V1 V2 cos a difference V1 V2 2 V1 V2 cos a suppression ratio 10 2 difference For the case of a phase error and this simplifies to 2 2cos a SSR 1010 810 5 2 cos 1 cos sin SSR 10 logio dB 1 cos sin which further simplifies to SSR 20 7 For a 1 degree this gives SSR 41 2 dB Q8 let the signal be y t where y t 20 0 cos wt where the amplitude ratios are 20 1 This can be expanded into y t 1 a cosut cosat a sinut sinot a t cos t B where a t V 1 2 c cos
74. nly the true SSB receiver can demodulate the other sideband The improvement in output signal to noise ratio can be large If the interference is removed there is a 3 dB advantage to be had if both sidebands are recovered in a true synchronous demodulator double the amplitude four times the signal power but only twice the noise power Q5 Q6 the carriers need to be at 90 degrees if and only if the QPS is perfect and there are no other although small and fixed phase shifts elsewhere in the circuitry To account for these small imperfections then the phase of the two carriers into the two MULTIPLIER modules needs to be adjustable Q7 no a 10 Hz error goes undetected for speech With a 100 Hz error speech is still intelligible to the experienced operator For larger errors an experienced operator can still communicate but it will now depend on other factors including the sign of the error It is easier to listen to an upward shift of say 200 Hz than downwards as the percentage change is significantly different Try it Note that the tolerable SSB carrier offset is an absolute measure Thus the higher the carrier frequency the tighter the tolerance on the receiver local carrier The sampling theorem 12 TRUNKS Supply speech unless students have SPEECH modules A two tone audio test signal else from the student s SPEECH modules answers to tutorial questions Q1 it prevents any out of band noise accompanying the mess
75. nsider it important to fine trim the DC offsets into each of the M LEVEL DECODER modules used by the students For details see the Advanced Modules User Manual at your discretion The experiment brings the student to the point where the system can be set up and demonstrated It does not describe but leaves plenty of room for many more measurements and observations For example for the various multi level signals available e estimation of bandwidth e estimation of maximum transmission speed for a fixed bandwidth noise free system as judged by eye pattern quality e determination of maximum speed for a BER equal to or better than a specified figure for a noisy fixed bandwidth system Some of these require access to both ends of the transmission system and so more than 12 slots as provided by a single TIMS 301 rack Reference should be made to the use of these modulation techniques in modems especially those used over fixed bandwidth low noise telephone lines TRUNKS At least two signals should be sent to TRUNKS Each will require an M LEVEL ENCODER and two MULTIPLIER modules Data clock of 8 333 Hz and carriers of 100 kHz For the first a 4 QAM signal is required as expected by students This gives the minimum of levels in the two paths of the demodulator output and so makes for an easier setting of the phasing Send two others of your choice Each signal will use the TTL output X from a SEQUENCE GENERATOR Use a
76. omponents at the fundamental will be about 9 dB above those at the third harmonic thus suggesting more than 40 dB attenuation is needed although those components at the upper edge of the fundamental band will be smaller than those in the centre suggesting some relaxing of the requirements On the other hand a typical bandpass filter will probably find the specification on the high side the most difficult to meet meaning that these low side considerations may not be of importance What all this means is that each case must be studied on its merits For the case B 0 33 then on the 1 significant sideband criterion from the tables 3 pairs of sideband are required to be kept At 100 kHz 3 6 sideband pairs are significant Notice this bandwidth increase is not in proportion to the increase of B For a 3 kHz top message frequency the filter bandwidths would be 2x3x3 18 kHz and 2x6x3 36 kHz This is far wider than that available with the TIMS filters typically 6 kHz and 20 kHz bandwidths so the top message frequency would have to be held below about 1 kHz Q4 the principle which comes to mind 15 to use the BFO principle multiply the VCO with a fixed frequency and take the difference frequency via a 60 kHz LPF The VCO tunes 70 to 130 kHz Subtracting from 100 kHz gives an output range of 0 to 30 kHz Unsuitable So you the Laboratory Manager should supply 130 kHz at TRUNKS from your VCO Students can then tune thei
77. onent The formula quoted Q1 can be used to determine beta with the proviso that we don t know the signs although it is probable that they are positive for the first three terms If you try a few successive groups of three adjacent components you will find that most cases yield 7 25 If the signal is then this is the peak phase deviation For an FM signal this is Af u Since u 2 1 f and f 2 kHz then Af 91 1 kHz By examining the final column from the bottom up until an amplitude exceeding 0 05 is reached 5 of 1 the amplitude of the unmodulated carrier the number of sidefrequency terms satisfying the 5 significant criterion is seen to be 9 Thus on that criterion bandwidth 2x9xmessage freq 36 kHz Q5 this represents a 180 deg phase change not detected by the usual spectrum analyser Such a phase shift on one component of a DSBSC pair of components has the effect of moving the resultant of the DSBSC by ninety degrees quadrature But when effective on both the net result is to reverse the sign phase of their resultant Remember J nB CD Jn Q6 the spectral lines will be spaced 1 kHz apart with relative amplitudes obtained from the Bessel tables for 1 5 and 10 respectively Q7 the number of sideband pairs required are respectively 3 7 and 13 Thus the bandwidths are 15 35 and 65 kHz respectively based on a 2 5 kHz message The bandwidth of a PM signal does increase linearly wi
78. onents was measured by summing the contributions of two sidefrequency components around 100 kHz The DC component which you were responsible for by ensuring a significant component at carrier frequency is derived directly from the single carrier term and so does not need to be halved in amplitude The amplitude spectrum of course does not show absolute phase nor recognise the relative phases between components the power in the FM signal is proportional to its amplitude squared It is also proportional to the sum of the squares of all significant components Knowing the latter from the amplitude spectrum the former E can be found The amplitude of the nth sideband from the carrier n 0 is known and is E J B Thus can be found J B Not quite since many values of will satisfy the relationship But by trying several values of n and using a little ingenuity the value of B can be isolated the DC term arose from translation of the single component at carrier frequency All other terms arose from the translation of two components which after translation fall on top of each other and reinforce to give twice the amplitude which would have been due to one of them alone Armstrong s phase modulator 48 answers to tutorial questions Ql expand y t a t cos a t 0 a t cosd t cosm t a t sind t sinw t P cosw t Q sino t a t V P Q o t tan Q P thus foreqn 3 P E and E m
79. order and or phase jitter In principle this sounds easy in pratice it will present problems in implementation An asynchronous SSB demodulator could recover the individual channels this would be adequate for speech This is an acceptable solution both in principle and in practice Q5 the guard band takes into account the finite width of the transition band of the channel filters and to a lesser extent the frequency stability of the demodulating carrier oscillators Q6 how does one measure the peak amplitude of a speech channel How does one measure the peak amplitude of an FDM signal The oscilloscope as a readily available and affordable instrument is very useful Specialised instruments also exist but are expensive If the peaks occur only occasionally could not one operate at a higher level than the peak level Phase division multiplex Whilst PDM does not find much application as a method of multiplexing speech in the 1990s the principle involved is very important This experiment is meant to illustrate the principle in a strictly analog environment A later experiment eg QASK shows a common application in digital communications TRUNKS For this experiment you must send a PDM signal via the TRUNKS Each TIMS 301 has a copy of the TIMS 100 kHz clock for use as the stolen carrier Make one message a single tone and the other speech Students can then assess the relative merits of nulling on a single tone and on speech
80. periment is principally to introduce the indirect baseband method of spectrum measurement answers to tutorial questions Q1 since the message will be in the 1 kHz region a sweep speed of about 1 ms cm would be suitable for showing up a kHz envelope Instructor s Manual 47 Q2 Q Q4 5 Q nN Q8 the effect upon the amplitude of the wanted components would be slight since it would be the difference between being scaled by sin 85 say rather than by sin 90 However the components which should have been reduced to zero will now have small but finite amplitudes being scaled by cos 85 rather than cos 90 although the filter output would be at message frequency there would not be a linear relationship between its amplitude and that of the message back at the transmitter Worse since the single tone message could lie anywhere in the audio range say 300 to 3000 Hz a filter suitable for 300 Hz would be useless for 3000 Hz and vice versa Also this is a special case of a single tone message for a more complex message the question is almost meaningless this would have been apparent when first odd then even harmonics of the message were identified with the SPECTRUM ANALYSER first draw the amplitude spectrum of the baseband signal These amplitudes will be the maximum attainable in each case The corresponding 100 kHz relative amplitude spectrum is obtained by remembering that each of these AC comp
81. phase of the resultant of the DSBSC signal unless it had special features for accepting such signals Phase meters typically compare zero crossings of the two input signals which need to be periodic A DSBSC is seldom ever periodic Convince yourself that this is so Q3 this is a philosophical question which you should consider at your leisure Q4 the TIMS FREQUENCY COUNTER counts the number of times the input signal passes through a pre determined amplitude level in the same direction This level is not zero a finite amplitude level is chosen to avoid corruption by noise and false readings with no input It can be shown that provided the depth of modulation of an AM signal is less than 100 its zero crossings are uniformly spaced at the carrier rate Thus provided the depth of modulation is not so high that the envelope troughs fall below the level at which the counter counts it will read the carrier frequency Weaver s SSB generator TRUNKS Speech is interesting but perhaps students have their own SPEECH modules the experiment This experiment as noted requires an extra pair of MULTIPLIER modules an extra TUNEABLE LPF and an extra PHASE SHIFTER module Instructor s Manual 31 The phase shifter at message frequencies is required to produce the inphase and quadrature carrier pair at 2 083 kHz If you do not have a spare PHASE SHIFTER module these two can be produced with a QPS module rather than a PHASE SHIFTER
82. pt Using the binomial theorem or otherwise this for small approximates to a t 1 02 4 07 4 cos2ut Amplitude ratio of unwanted to wanted AC components is 0 4 What size would a need to be for the presence of the second harmonic to be just noticed Say 0 1 Set it up on TIMS and find out Instructor s Manual 7 Q9 the spectrum is two lines same amplitude similar frequencies In the time domain this looks exactly like a DSBSC signal The envelope peaks will be separated by a time tg where to 2 7 Of and of is the frequency difference between the two audio tones Q10 it is useful for generating circles on an oscilloscope screen perhaps for a circular time base Product demodulation sync amp async TRUNKS You should generate the signals to be demodulated and send them via TRUNKS Remember the students have been told each signal is based on a 100 kHz carrier of which each TIMS 301 has a copy This will become the stolen carrier the students require for demodulation Required signals are DSBSC choose any suitable message typically speech or a single tone A two tone audio is legitimate if students are sufficiently experienced to cope with it It will look unusual but its behaviour with respect to phase adjustment must reveal it as true DSBSC SSB use a phasing type generator with speech as the message If you prefer an USSB with a 1 kHz tone message then the signal
83. r VCO through 70 to 130 kHz and obtain 0 to 60 kHz from their 60 kHz LPF modules Their FREQUENCY COUNTER will give them their exact output frequency FM and Bessel zeros Refer to the previous experiment entitled FM deviation multiplication It was suggested there that this experiment might be added on to that experiment TRUNKS No signals required FM demodulation with the PLL TRUNKS Set up an FM signal of your choice For qualitative measurements use a single tone as message or perhaps a two tone Otherwise speech may be preferable stable carrier If you wish your generator to be based on a stable 100 kHz carrier then Armstrong s method is indicated Use the generator modelled in the experiment entitled FM Deviation Instructor s Manual 51 Multiplication Note you will need an FM UTILITIES and a 100 kHz CHANNEL FILTERS type 2 module modulated VCO If you want really wideband FM you could use a VCO as the source Although its carrier will not be stable this may not concern you After all that is what the PLL is good at ie tracking a drifting carrier depth of course work The analysis of the PLL is not a trivial matter although the general principle of operation is simple to explain You can add tasks to those given their degree of difficulty depending upon the depth of understanding you require of your students Likewise you can add some more demanding Tutorial Questions The Costas loop 52 TR
84. referably they could have SPEECH modules Task T5 requires the creation of a text book like display of a DSBSC This means that both the envelope and the DSBSC itself are stationary on the screen Use a message which 2 Instructor s Manual is a sub multiple of the 100 kHz carrier This can be done with the 2 083 kHz sinusoidal message from the MASTER SIGNALS module this is a 48 1 ratio There is also an 8 333 kHz ratio 12 1 TTL signal at the MASTER SIGNALS module The TTL can be used instead but its fundamental must first be extracted with the TUNEABLE LPF Overload of the analog filter with a TTL input is not a problem as long as the output is near sinusoidal it is acceptable Switch the MULTIPLIER to AC coupling else the DC from the TTL will result in a carrier term at the MULTIPLIER output and the output signal will no longer be a DSBSC answers to tutorial questions Q1 The product V1 cos t and V gt cos 7t k V1 V2 2 cos 1 02 cos 02 where k is a property of the MULTIPLIER In the multiplier module has been set so that when Vj and V gt are at the TIMS ANALOG REFERENCE LEVEL so also will be the peak amplitude of this product Thus working in peak values we want 1 Vy which gives k Q2 since a DSBSC contains more than one component the question is meaningless unless perhaps if the DSBSC was periodic See next two questions Of interest is often the frequency of
85. s Chapter is not a conventional experiment Instructor s Manual 67 68 INTRODUCTORY ANALOG EXPERIMENTS 1 Introduction to modelling with 1 1 DSBSC BemeratrOm ET 2 Amplitude modulation a e a A KEE ETETE 3 EMV ClOPOS A EE 4 E EETA T SAEN 5 SSB generation the phasing 6 Product demodulation sync amp eceeccesceseceseceseceecseeeseeeeceeeeeeeeeeeneeneenaeenaes 8 SSB demodulation the phasing 11 The sampling EEA 12 PAM amp time division multiplex cccecccecccesceseceseceseceeeceeeseeeeeeseeeeeeeaeeneeeeeeeereeeens 13 Power measurements oes 14 INTRODUCTORY DIGITAL 15 PRBS generati om s 2 enina e R ERE 17 Ey patterns wei EER concave Wee ite 18 The noisy channel model E 18 Signal detection with the DECISION MAKER ceeecseseccseeeeceseeeceseeeeeeceaeeeeeneens 19 Lime CODING 3 55 evasion R ea R E 19 ASK wamplitude shift Keyin droen rein a E R TEE E 20 FSK frequenicy shift keying 5 O E E EA 2
86. short sequence for the 4 QAM longer for the others Use the patching shown to ensure matching relative phases at the students demodulators as instructed in the experiment Instructor s Manual 65 SEQUENCE M LEVEL ENCODER TRUNKS 8 333kHz cosat a 4 QAM transmitter answers to tutorial questions Q1 the decision point is set according to the waveform at the output of one of the filters The same point in time is used for the decision point of the other waveform Unless the filters are reasonably closely matched especially with respect to delay the decision point of the other waveform will not be optimum Q2 reasonable or not Quite OK under high signal to noise ratio conditions as is demonstrated by zero errors For marginal SNR it would best be decided by measurement if only because not enough information is known about the response of the PHASE CHANGER The approximate phase adjustment method could ensure that the q signal could be nulled from the output of the i channel but not that the i signal could be simultaneously nulled from the output of the q channel for example If this is significant it would show up in the BER under poor SNR conditions Q3 if wideband noise enters the input to the receiver not only will components at around 100 kHz be frequency translated to baseband along with the incoming signal but also noise components around 200 kHz This would degrade the perform
87. sinut so envelope 1 m sin2ut and phase tan m sinut Instructor s Manual Each of the last two expression can be expanded as harmonic series in terms ofu See appendix B for some ideas Q2 no It can typically only measure the phase between two sinusoidal components For the DSBSC it is the phase of the resultant which is required peak to peak V 42 52 volt trough to trough 5 volt Q4 the phase adjustment is made when the amplitude ratio of DSBSC and carrier is unity Call the carrier amplitude and the DSBSC maximum resultant amplitude E P Esin Ecos As the DSBSC resultant goes through its two maximum values twice per message cycle the resultant amplitudes are P and Q where E Esin 0 Ecos Q JE Esin cos 2 When 0 these two are equal but otherwise not The amplitude ratio is 1 5 Q 1 sina On an oscilloscope one could surely see a 5 amplitude difference In this case solving for P Q 1 05 gives a 1 4 degrees Q5 let the Armstrong signal be defined as y t sin at k cos ar cos ut where is the error from quadrature and the DSBSC to carrier amplitude ratio k Expand this into the in phase and quadrature form and get y t 1 sin ut sin at k cos cos ut k sin cos ut cos at which leads to k cos 0 sin 0 1 sin acos ut As first approximation let cosa
88. sted There is little difference in the difficulty of making either recording your responsibility but the method of demodulation student responsibility is significantly different The experiment has been written on the assumption the simpler arrangement of Scheme 2 will be used Although it is not in principle necessary to build two demodulators it may be more convincing to do so and so the notes recommend this FDM recording For the recording you will need a good quality audio tape recorder Bandwidth per channel should be at least 16 kHz and signal to noise ratio at least 60 dB In addition at least two general purpose cassette recorders are required or four if the more ambitious scheme is attempted together with cassettes of different passages of band limited 300 3000 Hz recorded speech The preparation of these has been described earlier method 1 scheme 1 It is possible to record all FDM channels simultaneously on a single track according to the scheme illustrated in Figure 1 below Notice that an upper sideband is required from each SSB generator Instructor s Manual 37 38 message USSB meee USSB USSB Figure 1 Whilst the above scheme looks elaborate it is well worth the trouble to get the complete FDM signal onto tape From then on setting up of the experiment is greatly simplified method 2 scheme 1 A less elaborate scheme for preparing the recording is suggested in Figures 2 a
89. t is an acceptable alternative Instructor s Manual Although some of the information contained in the Advanced Modules User Manual is reproduced within the experiment text it would be helpful if this manual was made available to students TRUNKS None TCM trellis coding PPM and PWM Because almost all of the required functions are in the single INTEGRATE amp DUMP module there is very little patching to be done There are no detailed step by step instructions given Students have to plan their own procedures Their aim apart from getting a feel for the signals involved is to make sufficient measurements to enable them to answer the Tutorial Questions TRUNKS None Speech perhaps Or students may have SPEECH modules answers to tutorial questions Q1 why not Q2 each pulse of a monostable pulse generator can be started by a clock pulse fixed position and stopped by the PPM signal making a variable width pulse Q3 more sensitive to revealing distortion Q4 Q5 the loop filter is too narrow for message frequencies in the range 300 3000 Hz QAM and 4 PSK TRUNKS For the students to generate and demodulate their own QPSK signal would ideally require four MULTIPLIER modules per TIMS 301 bay A simplification is for each group of students to generate a QPSK two MULTIPLIER modules and to model only one channel of the QPSK receiver one MULTIPLIER Instructor s Manual 63 If you have only two MUL
90. th message frequency for a fixed peak phase deviation But here there is a change of peak phase deviation with no change of message frequency The relationship between bandwidth number of significant sideband pairs and peak phase deviation is not a linear one Q8 for the 2 5 kHz message 20 2 5 8 On the 1 significant sideband criterion the number of sidebands required is 11 and the bandwidth 55 kHz With an increase of message frequency by a factor of 4 B reduces to 2 the sidebands spread out but less are needed On the 1 criterion the number required is now changed to 4 so the new bandwidth 80 kHz Q no change Introduction to FM using a VCO TRUNKS No signals required Instructor s Manual 45 46 comments For the two tone message combine an AUDIO OSCILLATOR and the 2 kHz message from MASTER SIGNALS module in ADDER Use a BUFFER at the ADDER output to vary the two tone amplitude answers to tutorial questions Q1 outside of a communications environment for example For low power short range situations such as non critical remote control telemetry Q2 sidebands will be spaced 1 kHz apart either side of the central 100 kHz carrier The relative amplitudes are calculated from Bessel tables knowing 5 1 5 Instructor s Manual Thus freq rel ampl o Q3 sideband are spaced apart by 1 kHz Relative to the amplitude of the unmodulated carrier 1 on the 10 signif
91. y means of a slotband definition than a passband width See the definition of slotband and other filter terminology in Appendix A to Volume Al of Communications Systems modelling with TIMS Q3 there are occasional traces which fall within the eye They are more likely to be observed with a storage than a more conventional type of oscilloscope Q4 it exercises more data patterns Q5 the transitions between top and bottom of the eye at the corners of the eye will be misplaced The noisy channel model 18 TRUNKS No signals required answers to tutorial questions Q1 decibel is a normalised scale which simplifies amplitude response comparisons It allows responses of filters in cascade to be added It preserves relative shapes for different scale ranges Logarithmic frequency offers a useful compression of the scale an octave covers the same width at all frequencies For small frequency ranges say less than an octave it is less important although it will show symmetry in the case for example of a second order bandpass filter A disadvantage is that the absolute level of measurement is not shown explicitly and this can be important information Q2 the sinewave will not appear distorted if the channel is narrow band thus not passing distortion components see the Part I experiment entitled Amplifier overload The single voltmeter reading alone would not reveal the existence of overload Two or more measurements should reveal the

Instructor`s Manual Communication Systems Modelling

Contents

Download Pdf Manuals

Related Search

Related Contents