scope jitter

Here are two rising edges from a digital delay generator, programmed

5.5 ns apart. The scope is set for 10 second persistance.

formatting link

The scope is triggering on the ch 1 input, halfway up. The sweep speed is maxed at 2 ns per division.

Digital scopes used to have a full sample of p-p jitter. This one shows no visible jitter on the ch1 trace, and hardly any on ch2.

I'm guessing that the trigger is really software, based on the reconstructed (software lowpass filtered) waveform. It's impressive, however they do it.

If I move the ch1 input to the AUX connector and trigger external, I see a full ns of p-p jitter on the ch2 rising edge. That means that the external trigger input is probably a classic comparator, and it's not synchronized to the 1 GHz sampling clock, and there's no attempt at interpolation.

Other scopes may do the external trigger thing better.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin
Loading thread data ...

You have to be careful with these software driven scopes. We have a few Tek DPO scopes of different generations. The old ones didn't do too much fudging, and you could tell. You got what came out of the ADC.

The later ones do a fair bit of fudging. One thing that really hit me was to take a sharp square wave with no ringing, and view it at a fairly fast time/div. Looks sharp. Then, take it slower, ie. greater time/div and freeze it (stop acq). Then, expand the time with the time/div dial (not magnification) and you get all these wavelet-like artifacts on each edge. I'm not sure what is going on there, whether changing the time/div without changing the actual sample rate screws up how they get more bandwidth out of the ADC, or what. Anyway, ISTR that it would do this at any time base, so it wasn't only doing the wavelet thing to get the last couple tens of ps out of the ADC. I never see this when the scope is just running "live".

Jon

Reply to
Jon Elson

... And that's why analog storage scopes still rule. :->

Reply to
Julian Barnes

What would come out of the ADC in this case would be 20 dots on the screen, jumping all over the place. Some reconstruction is necessary.

I am disappointed that they didn't do the external trigger better.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

In your opinion, which display exhibits a scope-induced artifact?

RL

Reply to
legg

Apples and Oranges. When you speed up the sample rate you get more information so the reconstruction is very accurate. When you capture a display and then magnify it, you are trying to see information that wasn't there in the buffer. So you can see the reconstruction artifacts.

I don't see how this is worse than an analog scope. The problem with an analog scope is triggering it. If you can trigger it on what you want to see, the digital scope shows you the same thing. It's when you

*can't* trigger on the detail you wish to see that the digital scope rules. You can zoom in on the detail and get something rather than not being able to see it at ALL.
--

Rick
Reply to
rickman

Obviously, 1 ns of p-p jitter is not real. For this scope, it's best to trigger through one of the four "vertical" inputs, not the external trigger path, when jitter matters.

This scope will not measure jitter, which is interesting. Software-dominated scopes do have a lot of places to play marketing games. And to have bugs.

Doesn't anybody else have a digital scope to try this on?

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Also when there is only one or very few events in a long time and even with the curtains drawn and the neighbor's porch lights off you can't see it.

Last week I had yet another case where the analog scope was outshining the digital one big time. A statistically very dispersed signal with some unwanted noise in there. IOW I had to see unwanted fuzz inside a lot of other "legit" fuzz, neither of which could be categorized a priori.

The only problem is that you can't easily obtain fast analog scopes anymore, except for one Japanese series that is very expensive. So I am hoping that the old Tek 7704A is going to hang in there until I am mostly retired. Its power supply had a few bzzzt episodes but at that age it's probably like the occasional little cardiac event in an older person.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

Digital scopes emulate that now.

--

Rick
Reply to
rickman

When you want a beautiful picture for a manual or a data sheet, you can just keep firing single acquisitions until you get the picture (and the numbers!) that you like.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

I have four that can do a factor of 8 to 20 better than that--doesn't everybody?

Cheers

Phil Hobbs

Reply to
Phil Hobbs

The $350 Rigol scopes have 1 Gs/s ADCs and sweep 2 ns/div.

Any decent rising edge can be tee connected to a channel input and then into another channel or the external trigger.

It's impressive how many people want to talk and not actually do things.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Right! And, some scopes make this CLEAR, by either going to a point by point display (which looks really sparse and ugly), or a piecewise-linear interpolation, which looks better, but usually is quite clear where the points are.

But, when they try to fudge it, BADLY, and ADD all these wavelet curlicues that can't possibly be there, then it makes me wonder what other stuff they are trying to cover up with signal processing software.

Mostly, I am strongly in the camp of liking digital scopes, and love being able to see what happened BEFORE a non-repetitive trigger, and similar things that can only be done with them. But, I want to have some feel of what tricks they are employing.

Jon

Reply to
Jon Elson

You mean one that can actually record sub-nanosecond-spaced waveforms?

Perhaps the signal generator might also be slightly specialized.

RL

Reply to
legg

That slowly but surely becomes a problem. The Iwatsu ones were always very pricey and now they've discontinued the whole analog series. The fastest analog one (1 GHz TS-81000) got discontinued some 3 years ago, and recently the rest of the 300 and 470 MHz ones followed suit.

The 7704A is a big monster, but keep yours working while you can :)

Now you can only get analog scopes from China, and apart from some of the expensive rarities, which are becoming rarer and more expensive every year, they are mostly designed for use in school teaching, so they only ever go to either 20 or 40 MHz. 60 is rare, 100 is really slim pickens, and as for 150, 200 and 300 MHz, AFAIK there is only one company that sort-of still makes them (they used to make them for the Chinese government and military back in the communist days, and as long as the state is still their major shareholder they are trying to keep the production of those models alive where possible).

Besides, your 7704A, being Tek from its "golden years", can probably exceed what its nameplate says by some margin. If you buy something that says "200 MHz" on its nameplate today (no matter from whom), most likely it will go to about 180 - 190 MHz and then suddenly and horribly tank (including all sorts of weird measurement artifacts imaginable).

When you hit the limits, digital does not tend to degrade gracefully, especially when it's cheap. As for analog, here's an example of the analog "equivalent" (one of the rarities, not of the teaching ones):

formatting link

The signal in the picture is a linear sweep from an RF signal generator (actually from a frequency characteristics analyzer, not in the photo, but here it's just being used as a signal source). The signal starts very low (below 1 MHz, as good as DC) and sweeps all the way up past the analog scope's bandwidth. The signal amplitude stays constant over the whole range of frequencies, so you actually get to see the scope's vertical amplifier characteristics vs. frequency. The X timebase on the scope is set so that the sweep rate would "just fit" from 0 to 500 MHz on the screen, so the signal frequency is "50 MHz/div" on the X-axis.

Note that the XJ4362A's nameplate bandwidth rating is 100 MHz, this is the first two horizontal divisions in the picture. At 100 MHz the gain of the scope input peaks, then at 150 MHz it gets "back" and crosses the nominal level, and after 200 MHz it starts to drop really fast. Around

225 MHz there seems to be some resonance somewhere, and after some 270 MHz the signal finally disappears away into the zero line.

Regards Dimitrij

Reply to
Dimitrij Klingbeil

Yes. In this thread I was getting 30ps p-p

Old (high end) Agilent scope > Rigol scope.

This one has four channels but no separate trigger input.

Allan

Reply to
Allan Herriman

I tried it on a TDS 694C and an Agilent MSO6054A, using a BNC tee and about 10 ns worth of RG58 to delay the edge between inputs. The Agilent had no visible jitter regardless of whether the other channel was Ch2 or Ext Trig. Standard deviation of Ch1-Ch2 delay was reported at 12.5 ps (i.e., too sm all to resolve at 1 ns/div.)

On the TDS 694C I saw about the same amount of jitter on Ch1-Ch2 as I see w hen I follow the directions in the manual for a (histogram-based) trigger j itter test, about 30-40 ps peak to peak. The trigger jitter test reports a std dev of about 7 ps on this scope, and the spec limit is 8 ps.

With Ch1 triggered externally there was more jitter, maybe 100 ps pk-pk. I don't think the specs are supposed to be the same. The rear panel trigger input Z is ">= 1.5K", for one thing, rather than 50R like the Ch1-Ch4 in puts.

-- john, KE5FX

Reply to
John Miles, KE5FX

They make some fine stuff.

Even with probe timing adjustments available in increments of 100pS, wouldn't it be wasted on a signal displaying dots at 1nS intervals?

Perhaps Rigol or Owon users have an eye out for the limitations of their hardware or are just more sensitive to potential ridicule. Better trust is only developed when it can be compared directly to more elaborate gear, which isn't always available or possible.

To be truly relevent, the measurement would need an external trigger, which the most popular Rigols (4ch) don't even have......so disallow the majority of your potential helpers in this query.

I wouldn't trust the timing on the Owons here with a ten foot pole, but use them daily for recording tasks. I won't apologize for this.

RL

Reply to
legg

The max sample rate is 10 Gs/s, namely 100 ps. So it could be that it behaves like my DPO, namely does software internal triggering (essentially zero jitter) and non-interpolated external triggering, with 1 clock p-p jitter.

It would cost maybe 15 dollars to add external trigger clock interpolation.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Can you view the trigger waveform? If not, it may not be sampled with another ADC. I can't see adding another ADC and front end for a trigger circuit and not making it a fifth channel.

How would your trigger interpolation circuit work?

--

Rick
Reply to
rickman

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.