How to determine jitter with a scope

[miguelito]

I have a budget digital scope (Rigol DS1054Z - upgraded to 100MHz bandwidth). I was thinking of a way to produce an eye diagram with it. I only care about 16/44 at the moment, so we are talking a bit rate of 1.44 MHz which isn’t all that much.

 

The few articles/videos I found use more sophisticated digital scopes with clock recovery features, but I am thinking there might be a way to do this regardless: If I trigger on a transition and then look at transitions, say, ten bit-lengths later, I should be able to see something akin to an eye diagram. I don’t really care about absolute measures as much as I care to see, for example, what mods/cables/connectors/cable lengths reduce jitter. And more specifically I am looking at simple mods to a CD transport (Cambridge CXC) to reduce such jitter.

 

So two basic questions:

1- Does this make sense?

2- What signal would you play on the transport to get the maximum number of transitions? White noise should not be such a bad choice.

 

Thx.

[mansr]

Have you seen this application note?

https://www.sitime.com/api/gated/AN10007-Jitter-and-measurement.pdf

[miguelito]

31 minutes ago, mansr said:

Have you seen this application note?

https://www.sitime.com/api/gated/AN10007-Jitter-and-measurement.pdf

This is fabulous. Thx @mansr

[miguelito]

On 10/1/2019 at 5:32 PM, mansr said:

Have you seen this application note?

https://www.sitime.com/api/gated/AN10007-Jitter-and-measurement.pdf

@mansr

 

I got this working although the math part not yet - scope config unclear, would like to do the math on the persisted data in the scope. I could also capture and move the data to the computer and use python to analyze. Thank you.

 

Question: Is the SPDIF input supposed to be terminated? I do have one pass-through BNC 50 Ohm terminator, but not a 75 Ohm. Closer to 75 than 1M obviously. But is it supposed to be terminated? I would have guessed yes... Thx.

[mansr]

4 hours ago, miguelito said:

Question: Is the SPDIF input supposed to be terminated? I do have one pass-through BNC 50 Ohm terminator, but not a 75 Ohm. Closer to 75 than 1M obviously. But is it supposed to be terminated?

What are you measuring? If you have an S/PDIF output simply connected to the scope, you should set the scope input to 75 Ω termination.

[miguelito]

1 hour ago, mansr said:

What are you measuring? If you have an S/PDIF output simply connected to the scope, you should set the scope input to 75 Ω termination.

That was my question, thx. My scope cannot be set to anything but the 1M Ohm impedance termination. I have 50 Ohm terminators but no 75 Ohm terminators (easy enough to make one).

[miguelito]

I am trying to measure jitter by triggering on an edge and then measuring n-th edge std dev. That's all I really care about. 

[mansr]

1 hour ago, miguelito said:

That was my question, thx. My scope cannot be set to anything but the 1M Ohm impedance termination.

How odd. It's a pretty basic feature.

[miguelito]

10 hours ago, mansr said:

How odd. It's a pretty basic feature.

It's a $349 scope... Frankly it is such a simple feature to replicate with external terminators... Plus it is so easy to blow out a terminator with such a little voltage (eg shouldn't put a voltage higher than 10V on a 1W terminator) that I rather use an external termination than internal... For example:

 

https://www.amazon.com/gp/product/B06XHFGVXM/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1

 

and 

 

https://www.amazon.com/gp/product/B01ASF5UHY/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1

 

Easy and safe.

[miguelito]

Interesting simple measurement: I see a rise time of about 8ns with a Blue Jeans coax cable ($18)  vs 12.6ns with an Audioquest Cinnamon ($69). All properly terminated.

 

Thoughts?

[mansr]

10 minutes ago, miguelito said:

Interesting simple measurement: I see a rise time of about 8ns with a Blue Jeans coax cable ($18)  vs 12.6ns with an Audioquest Cinnamon ($69). All properly terminated.

 

Thoughts?

The AQ cable has higher capacitance and/or inductance. Both figures are well within spec.

[miguelito]

15 hours ago, mansr said:

The AQ cable has higher capacitance and/or inductance. Both figures are well within spec.

Yes, I get that. Now consider the 5ns slope difference, and consider a crossing voltage noise of ~1mV (in the source or the receiver, just back of the envelope). So give or take this means the AQ cable could inherently have jitter that is higher by 5ps. I guess this is not much...

[mansr]

56 minutes ago, miguelito said:

Yes, I get that. Now consider the 5ns slope difference, and consider a crossing voltage noise of ~1mV (in the source or the receiver, just back of the envelope). So give or take this means the AQ cable could inherently have jitter that is higher by 5ps. I guess this is not much...

The clock recovered by typical S/PDIF receivers has much higher jitter anyway, the best specs I've seen being 50 ps rms.

[miguelito]

2 hours ago, mansr said:

The clock recovered by typical S/PDIF receivers has much higher jitter anyway, the best specs I've seen being 50 ps rms.

Ok... Compared to specs from high end DACs this looks awful!

[Arpiben]

In case you don't have it already this J.Dunn document may help you:

http://www2.electron.frba.utn.edu.ar/~jcecconi/Bibliografia/13 - Medicion de Amplificadores/Documentos/AudioPrecision_AN5_DigitalAudioMeasurement.pdf