Re: [SI-LIST] : Jitter measurement

About this list Date view Thread view Subject view Author view

From: David Instone (
Date: Wed Mar 21 2001 - 01:40:57 PST

   I guess I'll take a crack explaining that. I'm going to have to make
a few assumptions about how the scope actually works but I think it will
be close enough to show the mechanisms at work.
 Take the 8 Gs mode. At trigger the scope could determine the time
interval between trigger and next sampling pulse and thus determine
where on the screen to place its samples,( or it could simply commence
sampling). The accuracy with which the time at which the waveform
actually crossed the voltage level at which you are measuring the jitter
then depends on 1. The accuracy of that trigger to sample delay
measurement (or the trigger to samply delay repeatabilty)and 2. The
algorythm used to 'join the dots' and thus determine where the 'line'
crossed the jitter measurement voltage level.

Now in the RIS mode the start of sampling can be delayed by a random
number of 25Gs/s periods before commencing the actual sampling which
still takes place at 8Gs/s, thus filling in the spaces between previous
samples. The accuracy now depends on the accuracy of those delays and
again on the accuracy of the dot joining algorythm. Now however that
algorythm has 3 times as any dots to work on so it can more acurately
predict the wave shape, thus it can more accurately predict where the
crossing point was.

Now for a couple of points on the acctual measurment. If you have any
relativly low frequency FM, say due to bad LF powersupply decoupling,
then the scope will not detect this. You have to make a whole series of
measurments at different delays from trigger to determine this. At best
it's very time consuming. This is where the spectrum analyser comes
into its own. LF FM will show up as repetative peaks spaced by
multiples of the modulation frequency either side of the oscillator
frequency. eg 60khz FM will show up as peaks spaced at -180K, -120K,
-60K, +60K, +120K and +180K etc from the center frequency. You will
need the maths Paul spoke of to determine the amplitude (ie amount) of
deviation but at these low frequencies they should not be there, the
presence of such spikes means that your decoupling of the oscillator
supply needs to be improved.

Finally a word on p-p jitter values. If the jitter has a gaussian
distribution then the p-p value is without limit (ie +/- infinity),
scopes normally limit somewhere around +/- 3 sigma, Your scope looks
like it's using about +/- 5 sigma. I'd be willing to bet that whatever
jitter measurment you made the p-p would always show up as about
10*sigma. You mentioned that you are using this for a Gigabit serdes.
If it's Fibre channel or Gigabit ethernet then use 14 * sigma to obtain
the p-p value. 14 sigma includes all but 10^-12 of the gaussian

Sunil Kumar wrote:
> Hello Paul and Mike..
> Thanks a lot for your valuable inputs.
> Let me add some more information. I am measuring jitter of 62.5 MHz
> crystal oscillator (LVPECL). This clock is passing through a clock tree
> and finally is given to a Gigabit Transceiver (SERDES). My aim is to find
> the jitter at the clock input of the transceiver, and I am looking
> for few picoseconds sigma jitter. The real time oscilloscope
> (Lecroy) which I am using, is having bandwidth of 1GHz and sampling rate
> of 8Gs/s. I have found out that in addition to single shot (real time), it
> supports Random Interleaved Sampling (RIS). It performs RIS at
> 25Gs/s. Hence sampling step is 125ps in single shot and it is 40ps
> in RIS. Now the jitter measurement results for oscillator output are:
> sigma peak-to-peak
> RIS 7ps 65ps
> Single shot 11ps 100ps
> So, I am getting more jitter in single shot in comparision of RIS. Does
> higher sampling rate of RIS give better results, or something is wrong?
> Thanks..
> Sunil Kumar


Dave Instone. Compliance Engineer Storage Systems Development, MP24/22 Xyratex, Langstone Rd., Havant, Hampshire, P09 1SA, UK. Tel: +44 (0)23-92-496862 (direct line) Fax: +44 (0)23-92-496014 Tel: +44 (0)23-92-496000

**** To unsubscribe from si-list or si-list-digest: send e-mail to In the BODY of message put: UNSUBSCRIBE si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP. si-list archives are accessible at ****

About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Thu Jun 21 2001 - 10:11:17 PDT