How to get a really good measurement of a sharp square wave on a scope?

[Markw4]

Some jitter wisdom from another thread (where I was talking about jitter sensitivity of various dacs including 1-bit):

I said:

Clock jitter can do it. Enough to mess with ITD localization, sometimes enough to collapse a sound stage. Wrong power filtering for a clock oscillator can be one cause.

An interesting point in one response:

...Normal clock jitter cannot do it. It would have to be so bad as to be non-functional as a clock. The minuet differences people talk about simply do not matter. I know, because I have measured clock stability using an HP 5372A many times with all kinds of digital systems. Any functional crystal oscillator without major problems is good enough. Ceramic resonators are most times good enough...

I still disagree on that point. Sorry, but I do.

Now, I am back out. Good day to all the good and well-meaning folks here.

 

[Mark Tillotson]

but is it too much to ask for to have perfect magical test equipment that ignores the laws of physics and gives you ideal images of what's there?!

No, 'scope vendors can sell you very expensive active probes that will do this - they have absurdly low capacitance and high bandwidth and a price that will bring out a cold sweat, but its doable!

 

[anatech]

Hi Mark Tillotson,
Exactly right - if it is that critical, this is the cost of knowing with less uncertainty.

Hi Markw4,
No problem with disagreement - if you can back it up with real evidence. Tested under controlled situations by people who can replicate and set up a valid experiment. This is the basis of knowledge.

There are certain things we know by looking at the magnitude of the event (timing in this case) and a human's ability to differentiate between them. Magic aside, there are limits bound by the physical world and our bodies.

Snipe and disappear. Back out you go until the next opportunity.

 

[Markw4]

Okay, I guess I am called upon to defend myself. I do have a question with respect to the recent/present discussion regarding the OP an his interest in 50fs phase jitter crystals. My question is this: would any clock be good enough, likely even an ceramic resonator, for his purposes? If the answer is to the effect of maybe not, then a follow up question would be, what symptoms would be considered as valid evidence a lower phase noise clock is needed, only a nonlinear distortion FFT? Or, if distortion looks okay but there is some problem with ITD localization, how would that be measured for verification purposes with the best (and possibly most expensive) test equipment available today?

 

[MarcelvdG]

Well worth investigating. But the edges will be filtered out, other factors may change with edge shape.

It isn't the actual edge shape that is an issue. You're throwing that out in the filter.

In the case of a non-return-to-zero single-bit DAC, when the falling edge is steeper or less steep than the rising edge, the average value of 01100 is different from 01010. There is nothing a low-pass filter can do about that.

One solution is to go for return to zero.
01100 then becomes 0010100000
01010 then becomes 0010001000
The number of rising edges then becomes the same for both patterns, and the same holds for the number of falling edges.

 

[APEXHiFi]

True, and my design is a return-to- zero design. It also uses a negative complementary signal fed into an inverting side of the analogue filter as well to further reduce nonlinearities, but that doesn't stop me from playing with ideas and seeing what results I get!