How to measure deterministic jitter from eye-diagram by using command

Do any of the measurement capabilities on the eye diagram tool not help here? Probably you want the "Random Jitter" measurement (either left or right) which will give you a standard deviation of the two transition regions. You can then send these measurements to ADE. Since you're getting the standard deviation, for peak to peak, you'd typically multiply the standard deviation by 6.

Note these were added in an IC617 ISR - I can't remember exactly when they were added.

I see from your other post you're using abEyeCross. That should be OK, but nowadays most of the needed functionality is built-in to the eye diagram assistant.

Regards,

Andrew.

Dear Andrew,

Thanks a lot for your reply.

I was simulating a simple test setup. I am feeding a PRBS sequence of bitrate (1UI) of 31.25 pS to a chain of buffers. I am applying a sine wave of frequency 11 MHz and amplitude of 50 mV on the supply of the buffer chain. Now I am seeing the eye of the buffer chain output.

The jitter is deterministic here as I am not adding any Random noise and it is a simple transient analysis.

I checked the eye-Assistant. It shows both deterministic jitter and Random jitter numbers.

I have following doubts:-

a) I wonder why it is showing Random jitter number ?

b) Which jitter number I need to consider for calculating my peak-to-peak deterministic jitter number.

Kind Regards,

The trouble is that it's quite hard to tell by looking at the waveform data alone whether it is deterministic or random jitter. In this case you'd just see a single spread of data, whereas if you read the documentation, it explains that for the deterministic component it is looking for two peaks around each edge - so there would be some distribution due to the deterministic part, and then a random distribution around that. 

So I'd suggest taking a look at the documentation - it does explain what it does (you can hit the "?" icon on the eye diagram assistant to take you to the right document).

Andrew.

The trouble is that it's quite hard to tell by looking at the waveform data alone whether it is deterministic or random jitter. In this case you'd just see a single spread of data, whereas if you read the documentation, it explains that for the deterministic component it is looking for two peaks around each edge - so there would be some distribution due to the deterministic part, and then a random distribution around that. 

So I'd suggest taking a look at the documentation - it does explain what it does (you can hit the "?" icon on the eye diagram assistant to take you to the right document).

Andrew.

But I am not adding any noise (there is no external noise source) and it is a simple transient run,  then I wonder why I should get random jitter in the first place. This is a very basic doubt.

The measurement doesn't really know whether it's random or not. It is merely looking at the variation of the edge positions. There's some heuristics in the way the calculations are performed to try to distinguish the random part from the deterministic part. If you plotted a histogram of the variation of the threshold crossing when the variation is a deterministic sine wave, it won't look vastly different from if it had been random.

It's just statistics...

Andrew.