EEVblog #683 - Rigol DS1000Z & DS2000 Oscilloscope Jitter Problems

[EEVblog]

It's also interesting I'm not getting exactly 68kHz from every plot, which further points to PLL instability.  The next spur over seems to be more consistent at 125kHz.

I think I got about 65KHz after zooming in with 500KHz span, so that's probably more accurate for mine. It's in the video at the lab, can't confirm right now.

[orin]

Before we go any further, I have to ask how Dave measured that.  Was it with an EM probe or directly connected to the PLL output?  If it was an EM probe, the 68KHz could be coming from somewhere else.  Note that the bad plots with 100KHz modulation also had 68KHz spurs.

On which post did you see 68kHz with the 100kHz modulation?  I'd like to take a closer look.  On some of the pre-beta plots I'm getting 62.5kHz.

I was eyeballing the before and after plots that Dave posted.

[orin]

I did put the final release version of the new firmware on my scope and I've not seen the 5us jitter since.  I'll let it cool down overnight and test again in the morning.

No jitter at 5 us immediately after turning on this morning either.  Temp about 15 deg C.

[EEVblog]

I think I got about 65KHz after zooming in with 500KHz span, so that's probably more accurate for mine. It's in the video at the lab, can't confirm right now.

Nope, make that 67.5KHz

[aargee]

Waiting another few days and I'll update my 1054z with the latest firmware.

I don't really care for the internal operational details of my cheap little DSO (well, maybe I do - but not for these reasons).
It is the performance to specification that really should matter, and even with that, I have in mind the price point of this device and how much I'm willing to relax my expectations accordingly.

There still seems to be a lot of people searching for something here? What if Rigol had potted the PLL circuit, issued the fix and no one could reverse engineer it, a quiet thread this would become!

My two cents... 

[Bud]

Sure we do make some noise, aargee, this is an engineering, not a consumer/foolery forum. What a boring blog #683 would have become if we did not. Also bets are many readers have learned some technical things here.

[MarkL]

Before we go any further, I have to ask how Dave measured that.  Was it with an EM probe or directly connected to the PLL output?  If it was an EM probe, the 68KHz could be coming from somewhere else.  Note that the bad plots with 100KHz modulation also had 68KHz spurs.

On which post did you see 68kHz with the 100kHz modulation?  I'd like to take a closer look.  On some of the pre-beta plots I'm getting 62.5kHz.

I was eyeballing the before and after plots that Dave posted.

Hmm... You're right.  The 68kHz and 120kHz peaks do match up pretty well.

I still think it's a result of the PLL not being locked and component variations in the loop filter are in play from scope to scope, but it's worthwhile poking at the power pins to make sure there's not a strong 68kHz or 120kHz ripple that might be driving the PLL nuts.

I'm pretty sure I looked for this very early on while checking for external 100kHz influences, but it never hurts to verify.

[orin]

A couple of FFTs from Alessandro's program.

Source is an 8640B, 1MHz, about +13dBm.  First is the unmodulated signal.  Then I added a few %AM just to see what it looked like.  I can't persuade it to do 60KHz deviation FM at 1MHz.  I'll experiment some more to get an example with FM.

Anyway, the FFT of the unmodulated 1MHz looks good to me and the 60KHz AM shows nicely when turned on.

Edit: typo - it was 1MHz carrier.

[poida_pie]

Sure we do make some noise, aargee, this is an engineering, not a consumer/foolery forum. What a boring blog #683 would have become if we did not. Also bets are many readers have learned some technical things here.

I have to express my gratitude to you, Bud, for the time taken to show how all this PLL stuff works.
4 weeks ago I had no idea of contemporary PLL clock design. Wasn't phase locked loops used in analog TVs?
That's about all I knew - nothing.
But now after downloading and reading and rereading AD pll chip product info, AD pll design white papers, reading yours and MarkL's posts and more
I can keep up with what you are saying. A bit of the spectral world is still a mystery, i.e. why is the 65kHz tone so important...but that will come clear to me soon enough.
Having a broken DS1054Z PLL clock was the driving force to learn about something new (pll clocks) and discover they are used just about in all things radio including mobile phones.
As a result I felt confident to open the DSO, and put in a better designed loop filter, after sniffing the SPI data sent to the pll chip to determine all the programmed parameters.
 
Thanks again.

 

[MarkL]

A couple of FFTs from Alessandro's program.

Source is an 8640B, 1MHz, about +13dBm.  First is the unmodulated signal.  Then I added a few %AM just to see what it looked like.  I can't persuade it to do 60KHz deviation FM at 1MHz.  I'll experiment some more to get an example with FM.

Anyway, the FFT of the unmodulated 1MHz looks good to me and the 60KHz AM shows nicely when turned on.

Edit: typo - it was 1MHz carrier.

Just testing?

While you're at it, want to try a 1.2Mpt FFT of a 100MHz unmodulated signal?  Zoom in to a span of 500kHz.  See if you get those (approx) 68kHz and 120kHz peaks.

Edit: It might also be interesting to heat or cool the PLL and surrounding components and repeat the capture.  I'm going to bet the peaks move.

[Bud]

I have to express my gratitude to you, Bud, for the time taken to show how all this PLL stuff works.
... now after downloading and reading and rereading AD pll chip product info, AD pll design white papers, reading yours and MarkL's posts and more I can keep up with what you are saying.

Thank you and kudos to you for not missing an opportunity to learn.

A bit of the spectral world is still a mystery, i.e. why is the 65kHz tone so importantThanks again.

The absolute value is not important at all, it most likely will change with changing the loop filter. What is important is it should NOT be there. Look at the last attachment pics I posted yesterday, the artefacts (anything other than the carrier) present there are tiny, the PLL output is clean. This is what you want when building a PLL.

[orin]

A couple of FFTs from Alessandro's program.

Source is an 8640B, 1MHz, about +13dBm.  First is the unmodulated signal.  Then I added a few %AM just to see what it looked like.  I can't persuade it to do 60KHz deviation FM at 1MHz.  I'll experiment some more to get an example with FM.

Anyway, the FFT of the unmodulated 1MHz looks good to me and the 60KHz AM shows nicely when turned on.

Edit: typo - it was 1MHz carrier.

Just testing?

While you're at it, want to try a 1.2Mpt FFT of a 100MHz unmodulated signal?  Zoom in to a span of 500kHz.  See if you get those (approx) 68kHz and 120kHz peaks.

Edit: It might also be interesting to heat or cool the PLL and surrounding components and repeat the capture.  I'm going to bet the peaks move.

100MHz?  Not fair.  Mine is an unmolested 1054Z.  Anyway, here it is, along with a real 60KHz FM modulated 16MHz signal.

Edit: the FM signal was produced by feeding the 8640B output to the 8568A SA and fiddling with the 8640B FM controls until I got a spectrum with -35dBc or so sidebands.  Then I fed it to the scope through a 50 ohm terminator.

Edit #2: It was 12 Mpts.

[MarkL]

While you're at it, want to try a 1.2Mpt FFT of a 100MHz unmodulated signal?  Zoom in to a span of 500kHz.  See if you get those (approx) 68kHz and 120kHz peaks.

Edit: It might also be interesting to heat or cool the PLL and surrounding components and repeat the capture.  I'm going to bet the peaks move.

100MHz?  Not fair.  Mine is an unmolested 1054Z.  Anyway, here it is, along with a real 60KHz FM modulated 16MHz signal.

Sorry - I thought it was a new unit with the 100MHz eval license still in effect.  But if not, enough of the 100MHz signal is getting through anyway to see the shape of the ADC clock.

Your peaks are at 68kHz and 121Khz, as best I can tell by interpolating points on the picture.  As Bud described, these shouldn't be there at all and in all likelihood are a result of the PLL not being locked.  The noise floor should be mostly flat.

Is this the newly released firmware?  Or the beta?

Edit: the FM signal was produced by feeding the 8640B output to the 8568A SA and fiddling with the 8640B FM controls until I got a spectrum with -35dBc or so sidebands.  Then I fed it to the scope through a 50 ohm terminator.

Same characteristic shape (of the noise floor).  And it should be 20*log10(100M/16M) = 16dBc less than the 100MHz capture, which it is.  This confirms it's the ADC clock.

Was there an additional point in using an FM signal to test?

Edit #2: It was 12 Mpts.

I can tell.  The points are densely packed.

Did you say you were going to try to do some direct SA measurements on the PLL out?

[orin]

While you're at it, want to try a 1.2Mpt FFT of a 100MHz unmodulated signal?  Zoom in to a span of 500kHz.  See if you get those (approx) 68kHz and 120kHz peaks.

Edit: It might also be interesting to heat or cool the PLL and surrounding components and repeat the capture.  I'm going to bet the peaks move.

100MHz?  Not fair.  Mine is an unmolested 1054Z.  Anyway, here it is, along with a real 60KHz FM modulated 16MHz signal.

Sorry - I thought it was a new unit with the 100MHz eval license still in effect.  But if not, enough of the 100MHz signal is getting through anyway to see the shape of the ADC clock.

There isn't a 100MHz eval.  Using rise time, it was measuring about 78MHz out of the box with all evals in effect.

Your peaks are at 68kHz and 121Khz, as best I can tell by interpolating points on the picture.  As Bud described, these shouldn't be there at all and in all likelihood are a result of the PLL not being locked.  The noise floor should be mostly flat.

Is this the newly released firmware?  Or the beta?

The PLL is certainly a lot better than Dave's 'before' plot where from previous posts, the PLL was clearly in a startup mode, bouncing off its limits and producing the 100KHz spaced spurs.

The FW is the latest.  As posted by Chris at Rigol.

Edit: the FM signal was produced by feeding the 8640B output to the 8568A SA and fiddling with the 8640B FM controls until I got a spectrum with -35dBc or so sidebands.  Then I fed it to the scope through a 50 ohm terminator.

Same characteristic shape (of the noise floor).  And it should be 20*log10(100M/16M) = 16dBc less than the 100MHz capture, which it is.  This confirms it's the ADC clock.

Was there an additional point in using an FM signal to test?

Both the AM and FM tests were to compare a real signal to effects from the ADC clock.

Did you say you were going to try to do some direct SA measurements on the PLL out?

I have the ability, but I'm not doing anything yet.

[TMM]

before:


after:


can't complain with that 

edit: for the sake of completeness, here is at 7.41us which is where we would expect to see an issue from a 67.5KHz delta spur


and here at 33.3us for what i estimate from Dave's vid are the large spurs at a delta of ~15KHz. By 66us which is where we expect a corresponding null the jitter is no better nor significantly worse.

 

[Chipguy]

I did not expect anyone to so badly confuse AC trigger coupling with AC input coupling. I'm pretty sure I showed I was suing the trigger menu more than once in the video, and mentioned it was trigger coupling.

I confused it because I didn't listen closely at the first time.
Got a girlfriend AND 2 cats meowing around. That often results in input overload

[AlessandroAU]

I'm trying to run AlessandroAU's software (want to do a video on it), but it does not detect either the 32bit or 64 labview Runtime driver.
What am I doing wrong?

Dave, did you get it working?

Installing Labview runtime killed LAN capability on my UltraSigma and RUU. USB still works.

I lost LAN capability too, but I thought it was due to having Agilent's VISA installed when I installed the NI runtime.

I ended up installing the NI 32 bit runtime, then UltraSigma, then whatever it wanted to update.  USB is working.  I've not tried LAN since.

All the screenshots I've posted have been taken by Alessadro's program - it's working for me over USB.

However: Alessandro, I use a relatively low resolution laptop for instrument control and the FFT is mostly offscreen at the bottom.  I'd like scrollbars or the FFT in a separate tab.

Edit: fixed typo

Indeed, even on my 1680x1050 screen the program takes all space. The last version even starts on my second screen - not sure Alessandro if you are using a huge monitor or two of them.
 
Would be good to have resize ability if possible.

I've got a couple of large monitors in the lab, sorry for making it so large to those with smaller resolutions. Resize support is abit of a pain, no promises when I can make that happen but I'll give it a go when I'm back in the lab. Primary goal is getting it working for various scopes/firmwares as there are lots of bugs still.

I can't do much about the LAN incompatibility unfortunately

[EEVblog]

I'm trying to run AlessandroAU's software (want to do a video on it), but it does not detect either the 32bit or 64 labview Runtime driver.
What am I doing wrong?

Dave, did you get it working?

Nope, on two machines.

[AlessandroAU]

I'm trying to run AlessandroAU's software (want to do a video on it), but it does not detect either the 32bit or 64 labview Runtime driver.
What am I doing wrong?

Dave, did you get it working?

Nope, on two machines.

Odd, are you using USB, on windows 7 and have installed the 32bit runtime?

[EEVblog]

Odd, are you using USB, on windows 7 and have installed the 32bit runtime?

Tried 32bit and 64bit version. When I run the program says NI driver is not installed.
Yes, win 7, neither USB or ethernet yet, program doesn't even run.

[AlessandroAU]

I'm trying to run AlessandroAU's software (want to do a video on it), but it does not detect either the 32bit or 64 labview Runtime driver.
What am I doing wrong?

Dave, did you get it working?

Nope, on two machines.

Odd, are you using USB, on windows 7 and have installed the 32bit runtime?

Can you talk to the scope through the bundled ultrasigma suite?

[Owen]

If i remember correctly NI's Labview Runtime Driver ist just for execuiting your appliction build... If you're using instrument control elements (e.g. NI-VISA) you have to install the NI Device Drivers, too.

[sergey]

Hey guys,

First of all congratulations to all DS1000z series owners

Second, i'm a bit confused by information about new firmware for DS2000 series. Especially with that new thread new thread started recently. Did someone upgraded without any regressions (things which became worse in new fw)?

[sergey]

To be fair Sergey my new thread shows very rare errors and may only be on the faster DS2000, until someone tests on the DS1000z, which has different FW .  Like one error only occurs with fast pulse of 2nsec rise time which never happens on a DS1000Z.

Well, i've got DS2072 which i'm not sure will be a good idea to update now (working on some project now and don'w want to run into situation with untrusted scope...) That's my concern atm

Are there other issues discovered than the two in the first message in your topic? If not, they seems to not be critical for me and i'll be able to give it a shot with update next week.

[beNative]

Today I applied the new firmware and both trigger jitter issues seem to be resolved, although I must admit that the jitter issue on my unit (DS1054z) was not quite as serious as compared to some serious cases that were reported by other users. I hope it fixes the issue on those units as well.
At first glance it seems to me they also addressed the 'frequency measurement issue' where in some cases the scope was not able to calculate the frequency even when two or more periods of a waveform were shown (Dave also mentioned this in his video).
Can Rigol confirm that this was also addressed in this firmware update or am I experiencing here a Placebo effect?