HF Modulation Analysis
Contents
1. When External Arm is selected you will be able to select a channel and edge to be used to arm the measurement Once armed the SIA 3000 measures edges after the Arm Delay has elapsed Arm Number When External Arm is selected Arm Number allows you to choose a channel to be used as the Arm FA Arming Edge Choose the edge type to arm the measurement This is only available with External gt Arm Back Returns to the previous menu HIGH FREQUENCY MODULATION ANALYSIS PAGE 7 OF 12 WAVECREST CORPORATION 2002 Voltages Menu View 1 Sigma Threshold Voltage auto Channel Voltage Arm Voltage Back Threshold Voltage When set to AUTO sets start and stop threshold reference voltages based on the minimum and maximum level found on each channel from Pulsefind The threshold is automatically set to the 50 point The voltages are shown in the voltage display boxes after a pulsefind is completed Select USER VOLTS to manually enter threshold voltages in the voltage display boxes A pulsefind cannot be performed with this setting Channel Voltage When Threshold Voltage is set to AUTO use the channel control to view the threshold voltages derived from PULSEFIND Voltages are displayed under Channel Voltage In USER VOLTS the voltage can be set here Arm Voltage When Threshold Voltage is set to AUTO use the channel control to view the threshold voltages derived from PULSEFIND Voltag2. 1e ste valleys vi 0001y V2 001v For Help press F1 a al Figure 7 PLL response with modulation Also note that any rising curve that Senne lal settles at a high level will be seen by an elmls o Slelpieis of a 21 a HiFreqMod3 2 olx FFT as a low frequency component see Figure 8 The FFT treats the rising curve as the first part of an entire High Frequency Modulation FFT N clk Edge to Measure 0dB modulation curve The FFT will show a very low frequency spike which is an F j tee artifact Any real frequency nh a R components will remain constant in 83408 Tooo frequency and amplitude as 3db Freq Eni is decreased to view more cycles of the clock but any artifact will move down in frequency Passes to Avg FFT 99 668 8 1 OHz Page 1 of 4 7 503MHz 15 007MHz High Frequency Modulation Summary Mean Minimum Maximum Pk Pk 1 Sigma 114001ps 8607ps 162 958ps 154 352ps Figure 8 shows the FFT of the 1 sigma n fan mor Sae 1 Clock Peak 22 377ps 65 14MHz 3 927ps data from Figure 7 The low frequency Noeeh in mhe se shoulder is an artifact caused by the ine CN Mee in rising curve of the 1 Sigma plot Notice maaan i that the frequency spike to the right of the artifact is the true deterministic jitter Figure 8 The FFT o
3. CONTACT WAVECREST CORPORATION 7626 GOLDEN TRIANGLE DR EDEN PRAIRIE MN 55344 WWW WAVECREST COM 1 952 646 0111 HIGH FREQUENCY MODULATION ANALYSIS WAVECREST CORPORATION 2002 Rev09 11 02_ tag PAGE 12 OF 12 200015 01 RI EV A
4. H a expand zoom box over area of interest k Foo i Zoom out double click left mouse kon 8 Status Bar Cursor coordinates Displayed in the ah box at lower right portion of the VISI E Page 1 of 4 High Frequency Modulation Summary M Mi Pk Pk Panel Units are same as those in plot sona ee Minimum gm PERK Pk Pk 389251ps 25 921ps 0 526ns PJ Freq RJ r F gt lack Peak 16647ps 2502MHz 4 704ps CI cf 7212 Statistics View aaa ane ne PE Contains measured data from each For Help press F1 K 832 539Periods Y 311 444ps f S Figure 2 High Frequency Modulation Tool view plot provided by the tool Making your measurement e Open the High Frequency Modulation tool in VISI e Connect your source to a measurement channel Go to Acquire Options and use Add Del Channel to select that channel e On the Front Panel or Tool Bar press Pulse Find amp Verify the voltage levels and close out the pulse find box Press the Single Acquire button LI Device under Test Figure 3 Instrument Setup Understanding the Results All of the measurements and data required Tool Edt Action Display View Help to calculate and display the different views mjsmjs 0 alelmiela el sical 2 or plots available in the tool are gathered i D E simultaneously on one pa
5. case the RJ RMS value measured may not contain the same spectral content as did the measurement with Fmax set to 1 The reported RJ RMS value would be lower with the Fmax set to 2 Passes to Avg FFT Selects the number of passes to average for the FFT output Averaging will generally reduce the noise floor of the FFT Back Returns to the previous menu PAGE 6 OF 12 WAVECREST CORPORATION 2002 Arming Menu View Arm Delay 19 21ns isim The arm delay sets the minimum time from an arm event to the first measurement edge Figure 3 shows there is a user selectable 19 to 21 ns delay from the Arm event to the first measurement Arm Delay 19 21ns PERIOD MEASUREMENT EXAMPLE CHI ARM ON START Arming Mode arm On Stop on ar Tree CH2 EXTERNAL ARM Atm Number Figure 3 Typical Arm Setup time The Pattern Marker is the External Arm Arming Edge Ri ing Arming Mode An arm is required to make every measurement e The Arm on Stop selection will use an edge from the currently chosen measurement channel Stop refers to using an edge type rising or falling that begins the e The Arm on Start selection will use an edge from the currently chosen measurement channel Start refers to using an edge type rising or falling that begins the measurement e The External Arm selection will use an edge from a channel different than the one s chosen to make the measurement s
6. may not contain the same spectral content as did the measurement with Fmax set to 1 The reported RJ RMS value would be lower with the Fmax set to 2 Low Pass Freq MHz Upper frequency limit for the window over which RJ and Pu is calculated Default is the Nyquist frequency Back Returns to the previous menu PAGE 9 OF 12 WAVECREST CORPORATION 2002 View Options Menu View 1 Sigma FFT Window Triangular X Padding Multiplier 16 Alpha Factor X axis Span Periods Back FFT Window To reduce spectral information distortion of FFTs the time domain signal is multiplied by a window weighting function before the transform is performed The choice of window will determine which spectral components will be isolated or separated from the dominant frequency s Each window function has advantages disadvantages over other windows FFT Padding Multiplier Padding increases the frequency resolution of the FFT Generally a higher padding value will increase transformation processing time FFT Alpha Factor Varying the Alpha Factor illustrates the inverse proportionality relationship between the spectral peak width and the sidelobe rejection of the Kaiser Bessel window As the Alpha Factor increases the spectral peak widens and the side lobes shrink As the Alpha Factor decreases the spectral peak narrows and the side lobes increase in amplitude X axis Select the horizontal plot axis uni
7. 0s N Clack Pk Pk 1197ns 100 704kHz 117 913ps low frequencies Therefore Low frequency 7 vi 0 025V v2 jitter components are lower in amplitude Ee Figure 6 Shows the N clk FFT A 1MHz peak is present HIGH FREQUENCY MODULATION ANALYSIS PAGE 3 OF 12 WAVECREST CORPORATION 2002 PLL Measurements This section covers the measurement of EE BOE Phase Locked Loop PLL devices using Aelel 0 slejpisia ol 1 21x this tool When analyzing PLL s the 1 High Frequency Modulation 1 Sigma vs Span Edge to Mesure Sigma Plot responses produce a Ops era characteristic curve which indicates the ch noise response and bandwidth of the PLL s feedback loop cs 34B Freq kHz Figure 7 shows the analysis of a PLL In enc EX the displayed 1 Sigma vs Span view Fn ie note the characteristic loop response of the PLL over N periods The 1 sigma f jitter values increase over a small E E number of periods x axis This relates se S S UAN 1 1 A DPeriods 600Periods 1200Periods to high frequencies Ata certain number Pisoni SO of periods or frequency the jitter stops lise M tmps dome eme tte mn Pk Pk 316 453ps 25 33ps 464 752ps 439 422ps accumulating and the jitter or PJ Fe Al P Clack Peak 22377ps 6514MHz 3 927ps modulation effects are seen as hills and Mami Tite
8. HIGH FREQUENCY MODULATION FOR THE SIA 3000 Applications of High Frequency Modulation Analysis tool e View accumulated jitter in the modulation domain the signal causing the jitter e Look at the frequency components of jitter using an FFT e View how the jitter changes over time or output cycles Introduction The focus of this guide is to familiarize the user with the High Frequency Modulation tool allowing quick and easy measurements and interpretation of results Refer to the SIA 3000 User s Manual and the VISI help files for more information Theory of Operation The analysis of measurements gathered with the High Frequency Modulation tool allows the user to see jitter accumulation This tool automatically creates histograms over an increasing range of periods The process is shown in Figure 1 The SIA 3000 builds a histogram of time measurements for a single period then increments and makes another histogram of two periods repeating this process until enough periods are spanned to cover the 3dB frequency Measurement Schedule Histograms of measurements Clock Count 2 J 7 1 1 X Samples 2 2 2 d Count 3 J v a v a I 1 X Samples mi 1 2 3 1 2 3 1 2 3 Count 4 J Mf ETOT re 44 X Samples a 154 1 2 a 1 2 3 4 1 2 A ount n anan f
9. es are displayed under Arm Voltage In USER VOLTS the voltage can be set here Back Returns to the previous menu HIGH FREQUENCY MODULATION ANALYSIS PAGE 8 OF 12 WAVECREST CORPORATION 2002 RJ PJ Filters Menu View 1 Sigma RJ PJ HPF Natural Rolloff High Pass Freg MHz RJ PJ LPF Nyquist Low Pass Freq MHz Back HIGH FREQUENCY MODULATION ANALYSIS RJ PJ HPF Natural Rolloff Uses the 3dB frequency as the Natural Rolloff This is a Natural Rolloff because the amount of data acquired by the tool naturally limits the lowest frequency seen Brickwall Brick wall cuts off any frequencies below this value Note that the 3dB frequency should be lower than the brick wall cutoff or the natural rolloff will have a filtering effect NOTE Depending on the frequency of the clock being measured it is possible that the Nyquist Frequency of the clock 1 2 the clock rate could intrude into or be below the region chosen as the First Order filter frequency Additionally if the Fmax Divider is set at any value other than one this will eliminate any frequency content above divided Fmax For example if measuring a 400MHz clock the Fmax or Nyquist is 200MHz Therefore a First Order filter set to 2OMHz would be 20dB down at 200MHz But if the Fmax divider is set to 2 then the nyquist or Fmax would be 100MHz and would cut into the first order filter In this case the RJ RMS
10. f J F 7 an 1 1 J F x Samples 1 Sigma values from histograms are plotted FFT of the autocorrelation of the variance of 1 sigmas a periodic jitter component oe lo rp 1 lox e D 2 h j a Nha AA i UM Y 15 Y N y Event Frequency The 1 Sigma and Peak to Peak values from each histogram can then be plotted relative to the Figure 1 High Frequency Modulation Analysis data acquisition number of periods spanned These plots allow you to see how jitter changes or accumulates over increasing numbers of periods In doing so we are able to see the Jitter Modulation This time domain data can be transformed to the frequency domain by using an FFT of the autocorrelation of the variance of the 1 sigma values Frequency vs power can then be plotted HIGH FREQUENCY MODULATION ANALYSIS PAGE OF 2 WAVECREST CORPORATION 2002 200015 01 REV A High Frequency Modulation Tool Visi HiFreqMod_clkiD Tool Edit Action Display View Help Dialog Bar U U Alzua BO elella ol Ha 2e View Access to Acquire FFT algorithm a High Frequency Modulation Sigma ve Span ETS Arming and Filter settings A Chee Hits Per Measure Plot area En Zoom in hold down left mouse button fon 160ps
11. f the data from Fig 8 component at 1MHz HIGH FREQUENCY MODULATION ANALYSIS PAGE 4 OF 12 WAVECREST CORPORATION 2002 High Frequency Modulation Main Menu View View Mina W The View pull down menu provides the user several different ways to see the acquired measurement data depending on the active tool Acquire Options Acquire Options Opens the Acquire Options menu Arming Opens the Arming menu Arming Voltages Opens the Voltages menu Voltages RJ PJ Filters Opens the RJ PJ Filters menu RJ PJ Filters View Options Opens the View Options menu View Options FEEL HIGH FREQUENCY MODULATION ANALYSIS PAGE 5 OF 12 WAVECREST CORPORATION 2002 Acquire Options Menu View 1 Sigma Channel Hits Per Measure 300 Edge to Measure Rising HPF 3dB Freq kHz 100 000000 Fmax Divider Passes to Avg FFT Back i HIGH FREQUENCY MODULATION ANALYSIS Channel Opens the channel selection menu Use the keypad to select one measurement channel Hits Per Measure Determines the number of time measurements that will be made for each edge or point on the plot Edge to Measure Select Rising or Falling edge to measure HPF 3dB Freq kHz The HPF 3dB low rolloff frequency is the frequency of the half power point of the 20dB decade knee The choice of this frequency will determine the low frequencies visible on the FFT The 3dB Frequency kHz is used to determine the maximu
12. lues were increasing A further reduction in the 3db Freq value would measure over coins 1737 Pers CRE LE more periods to show this modulation PEREN Minimum Maximum PK PK 1 Sigma O52ins 11127ps O853ns 0 842ns Pk Pk O766ns 34106ps 124 ns 1 207ns Fre After viewing the data provided by the 1 1 Pen op hte ER sigma vs span view Figures 4 amp 5 use the me keyboard Page up or Page down keys to see For Help press F1 K 370 097Periods Y 0 857ns F the frequency and amplitude of the periodic p jitter components in the FFT views Figure Figure 5 Shows the jitter modulation over more 6 periods y Kh Too Help These views are FFTs of the Z mS ao eenvh ekina zki Ea autocorrelation of the variance of the 1 High Freguenoy Modulation FFT N ek a sigma values in the 1 sigma vs span plot cer Two FFT views are provided FFT N clk and _ 1 clk The N clk view shows every jitter frequency as having equal effects on a on period of the carrier frequency The 1 clk view shows how the jitter ene frequency components and amplitude affect i a single clock period Low Frequency jitter rss ne has a smaller effect on the jitter of a single Ho Fer elton Sry Minimum Maximum Pk Pk clock period Compared to the N clk view rs me Je oane Vite Pk Pk 34106ps the 1 clk view has a 20 db decade roll off for c eu Simas True 539
13. m measurement interval to be used in sampling and is entered in kHz A lower 3dB Frequency extends the time required to acquire the measurement set because histograms over many more periods must be acquired Below the 3dB Frequency a natural roll off of approximately 20dB per decade is observed The default value is 100kHz 3dB Frequency affects how much data is acquired and therefore the choice of this value also affects the test time Fmax Divider Allows scaling of the FFT by dividing the upper frequency end of the FFT Default is 1 which shows frequencies of jitter up to 1 2 the clock rate Changing this value allows faster analysis of lower frequency information by skipping edges and ignoring high frequency effects However changing this value without changing the HPF 3dB frequency will reduce the number of edges measured NOTE Depending on the frequency of the clock being measured it is possible that the Nyquist of the clock 1 2 the clock rate could intrude into or is below the region chosen as the First Order filter frequency Additionally if the Fmax Divider is set at any value other than one this will eliminate any frequency content above divided Fmax For example if measuring a 400MHz clock the Fmax or Nyquist is 200MHz Therefore a First Order filter set to 20MHz would be 20dB down at 200MHz But if the Fmax divider is set to 2 then the Nyquist or Fmax would be 100MHz and would cut into the first order filter In this
14. ss or acquisition firs Fre Before investigating the FFT start by a viewing the 1 Sigma vs span plot view 2 eat Figure 4 Try to get at least two full cycles Les ne of the modulation of interest Note that z Figure 4 shows increasing jitter Further data mE needs to be gathered over enough periods to determine if this is modulation or if it stops de Pis OPeriods 150Periods 300Periods accumulating This is done by decreasing ee the 3db Freq value in the Dialog bar menu te aime peP Be aware that test times increase when ey ee ie ee looking for lower freq modulation since the Camara MH tool will need to look over many more Vi 4m vwe uw porgue oF ihe sigialunderitest Figure 4 Shows increasing jitter not enough information to determine if modulation is present HIGH FREQUENCY MODULATION ANALYSIS PAGE 2 OF 12 WAVECREST CORPORATION 2002 In Figure 5 3db Freq value on the Dialog n e ty ver ter Bar has been decreased to 15 0 KHz from AmA Au aeei ee Ss 100KHz in figure 4 The periodic nature of mm a the plot data clearly shows that there is a modulation present The jitter increases to a a maximum of about 851ps over 370 Periods Pe and decreases to a minimum over 748 n aon periods The minimum 1 sigma values are poet constant A lower frequency modulation would be indicated if the minimum 1 Sigma ae va
15. t of measure Delay Time Shows time from Arm Delay Periods Shows number of periods from Arm Elapsed Time Shows elapsed acquisition time Event Take a measurement for each event from START event to STOP event and plot these values versus the count Hit Number Low Frequency Modulation Hit number is integer value assigned to each measurement as it is made Measurement Shows number of total measurements Span Periods Shows number of periods measured Span Time Shows units of time measured Time Take a measurement and plot the value versus time UI Spans Shows number of Ul spans measured Back Returns to the previous menu HIGH FREQUENCY MODULATION ANALYSIS PAGE IOOF 12 WAVECREST CORPORATION 2002 Summary The High Frequency Modulation Analysis tool capitalizes on the capabilities of the SIA 3000 s internal architecture specifically the Nth event counters The ability of the counters to span multiple edges or periods allows the SIA 3000 to perform measurements allowing it to view jitter modulation of the signal under test This data is plotted on either a 1 sigma vs span view or an FFT view Performing an FFT of the autocorrelation of the variance from the 1 sigma information will give a frequency vs power plot allowing the user to see the frequency components of the jitter modulation and their respective amplitudes HIGH FREQUENCY MODULATION ANALYSIS PAGE II OF 12 WAVECREST CORPORATION 2002 FOR MORE INFORMATION
16. value measured may not contain the same spectral content as did the measurement with Fmax set to 1 The reported RJ RMS value would be lower with the Fmax set to 2 High Pass Freq MHz Lower frequency limit for the window over which RJ and PJ is calculated Default is Corner Frequency This setting cannot be set lower than the corner frequency RJ PJ Low Pass Filter LPF Nyquist This is the default The highest frequency that the tool can measure is 1 2 the clock rate or the Nyquist First Order Enter a 3dB frequency This filter does not eliminate all the jitter energy above this frequency it is not a brick wall filter The frequency components or spectral content above this frequency will still contribute to the RJ RMS as displayed from the 1 sigma plot see figure NOTE Depending on the frequency of the clock being measured it is possible that the Nyquist of the clock 1 2 the clock rate could intrude into or be below the region chosen as the First Order filter frequency Additionally if the Fmax Divider is set at any value other than one this will eliminate any frequency content above divided Fmax For example if measuring a 400MHz clock the Fmax or Nyquist is 200MHz Therefore a First Order filter set to 20MHZ would be 20dB down at 200MHz But if the Fmax Divider is set to 2 then the Nyquist or Fmax would be 100MHz and would cut into the first order filter In this case the RJ RMS value measured
Download Pdf Manuals
Related Search