Help: My New Rigol DS1054Z Shows a Weird Double Trace

[frozenfrogz]

Software: 00.04.04.03.02
Board: 0.1.4
Boot: 0.0.1.4
Firmware: 0.2.3.11
CPLD: 1.1

I can not reproduce the double trace via the steps posted on reddit.

[JohnPen]

These are the results following checks on my upgraded Rigol 1054Z.   Pre conditions:  No input signal, a 500usec TB for best visibilty and 1mV range ( X1 sensitivity).

 AC/DC input selected.  Vector Display - Visual amplitude of noise ~1mV  (filled in noise)   Dots amplitude of noise  ~1mV (2 noisy traces ~1mV apart
 Channel GND selected.  Vector Display - Visual amplitude of noise  ~500uV  (filled in noise)  Dots amplitude of noise ~500uV ( 2 noisy traces ~500uV apart)

In both cases this is the basic noise level of the scope but the Dots mode displays it in a somewhat more confusing way.  In either case Vector or Dots with a low level signal in mV you are unlikely to see anything useful unless it is is several mV above the scopes basic noise level.   Having an external low noise amplifier could be useful in some applications to raise the signal level and improve the signal to noise ratio on the scope. 

Edit_ It is worth noting that this noise level also impacts on the trace triggering unless your input signal is quite a lot higher than the noise level.

[Porcine Porcupine]

I have to say this forum is fantastic. All the great responses I've gotten here have been very helpful.

single shot. are the multiple traces still in the display?

and do an FFT of the signal, make it use you can see the frequencies up to...
what are the AM and FM frequency bands in your area?

Yes, they're still there with single shot. I don't really see anything notable in the FFT corresponding to the AM band (535-1605 kHz), but there is some obvious stuff in the FM (88 to 108 MHz) band.

Seems like everybody else has it, too so you won't gain anything. Take comfort from the fact that nobody here even knew about it. Whatever the cause of the problem it has zero impact in real life.

I think you've got me nearly convinced to forget about it. One concern I still have is whether I'm seeing a normal amount of noise for this scope. The whole reason I switched to dots and found the double trace is because the calibration signal trace looked too fat to me. You can see in the image below I'm seeing about around 100 mVpp with the vertical scale set to 500 mV/div. I'm assuming the signal itself is relatively noise free of course, but does 20% of the vertical setting seem reasonable here for this scope? If it does, I think I'm finished worrying about it.

Software: 00.04.04.03.02
Board: 0.1.4
Boot: 0.0.1.4
Firmware: 0.2.3.11
CPLD: 1.1

I can not reproduce the double trace via the steps posted on reddit.

I just found out how to get the extended system information. It looks like we have an identical combination of hardware and software.



BTW, here's a forum post about getting the extended information if anyone is wondering.

https://www.eevblog.com/forum/testgear/new-rigol-ds1054z-oscilloscope/msg1101808/#msg1101808

[cybermaus]

Grounding should be good. The house I'm in was built about 30 years ago in compliance with the codes of the time. All the wall sockets have dedicated ground conductors. The ground and neutral wires are bonded only in the main breaker panel as they should be per the NEC, and the ground is connected to a copper ground rod driven into the ground near the main panel.

I verified continuity between the oscilloscope's ground and the power cord's ground prong. I tried plugging the oscilloscope into a surge protector that provides filtering and directly into several wall sockets on different circuits, but the double trace persisted.

Silly suggestion, but did you check continuity between power cord ground and actual grounding pin and/or said neutral bonding?

When I bought my house actually had the screw linking to the ground pin loose. Not just a little due to thermal expansion cycling, but like several turns loose. Like the code inspector never did his job. But it can be in other wiring places as well.

[Fungus]

I think you've got me nearly convinced to forget about it. One concern I still have is whether I'm seeing a normal amount of noise for this scope. The whole reason I switched to dots and found the double trace is because the calibration signal trace looked too fat to me. You can see in the image below I'm seeing about around 100 mVpp with the vertical scale set to 500 mV/div.

It looks identical to mine.

I'm assuming the signal itself is relatively noise free of course

FWIW you can turn on averaging mode. Averaging mode can get rid of noise on repetitive signals by finding complete waves in the incoming signal and averaging them together (see manual for details).

This is the same signal with averaging turned on - the line is 1 pixel wide:



I also tried averaging in dots mode and it cures the "double line" problem! 

This is dots mode with averaging turned on - no double lines:


Conclusion: Your hardware is fine, IMHO. It's all about the way the 'scope copies noisy data to the screen.

We need to learn all the display options and when to use each one.

[JPortici]

there is some obvious stuff in the FM (88 to 108 MHz) band.

and it shouldn't be there.
i still have this scope among other equipment. at home if i'm not extra careful the signal is swamped in FM picked up by the scope (not by the probe) so that 10 mV/div is the best i can do at full bandwidth.
at the work lab instead the scope is very quiet.

i think cybermaus is onto something

[Porcine Porcupine]

Silly suggestion, but did you check continuity between power cord ground and actual grounding pin and/or said neutral bonding?

When I bought my house actually had the screw linking to the ground pin loose. Not just a little due to thermal expansion cycling, but like several turns loose. Like the code inspector never did his job. But it can be in other wiring places as well.

I didn't check that, but I'll check it out when I get a chance since you mentioned it.

It looks identical to mine.

Cool, I think I'm satisfied now that my scope isn't defective.

and it shouldn't be there.
i still have this scope among other equipment. at home if i'm not extra careful the signal is swamped in FM picked up by the scope (not by the probe) so that 10 mV/div is the best i can do at full bandwidth.
at the work lab instead the scope is very quiet.

i think cybermaus is onto something

That's interesting. The stuff I see in the FM band goes away when I set the channel's input coupling to ground. I'm not sure if that means what I'm seeing is getting in through the probe or not.

[Old Printer]

When the discussion comes as to which scope to buy, some will offer the opinion that the large number of 1054Z users, particularly on this forum, is a major plus. It is instances like this that give those statements credence. Imagine trying to sort out something like this with a more obscure scope. You might well wind up sending it off to the factory concerned it is part of a larger fault. All of these low end scopes are going to have some issues, the question is what to do about it. In this case, probably nothing, but how to come to that conclusion on your own.  Dave

[IAmBack]

In my scope I can replicate this even with "coupling" set to GND. I simply see two yellow lines, 4mV one from other (10x probe set in menu).
When I set time base to 200ms and press run/stop (it lit in red), then go to 1ms, those lines are getting close one to each other making "noisy line". but with two thin black lines...

When I turn on rest of the channels, I see it in each of them. Channel 1 and 3 shows little more distance between lined. It looks like 8-channel scope...

The distance between lines is changing with the "mem depth" setting - increasing mem depth increases distance between "doubled" lines. Non-GND coupling also makes the distance wider...

Maybe it is some kind of pattern noise being "uncovered" by the decimation before displaying them?

[Fungus]

Maybe it is some kind of pattern noise being "uncovered" by the decimation before displaying them?

Yes, definitely this. You can see it happening if you go into zoom mode (see my screenshots).

[IAmBack]

I like this thread.
It explains why I need new Siglent 4ch scope

[Fungus]

I like this thread.
It explains why I need new Siglent 4ch scope

Yeah, because that one will be perfect. 

Better the devil you know...

[IAmBack]

I like this thread.
It explains why I need new Siglent 4ch scope

Yeah, because that one will be perfect. 

Better the devil you know...

I guess, that when all such "features" of the new scope will be discovered, then a newer one will apear in the market. This makes GAS fresh and vivid. As John Cleese said - there is no illness that can't be prolonged with an expensive cure .

[IAmBack]

I think you've got me nearly convinced to forget about it. One concern I still have is whether I'm seeing a normal amount of noise for this scope. The whole reason I switched to dots and found the double trace is because the calibration signal trace looked too fat to me. You can see in the image below I'm seeing about around 100 mVpp with the vertical scale set to 500 mV/div.

It looks identical to mine.

Mine has BY-AY less than 60 mV.

[JPortici]

That's interesting. The stuff I see in the FM band goes away when I set the channel's input coupling to ground. I'm not sure if that means what I'm seeing is getting in through the probe or not.

just so you know

-1054Z always do everything from screen data, which is not the acquired data. there are countless topics on that
-1054Z doesn't a true GND mode. In the 1054Z when you choose GND coupling what it actually does is draw a line at zero on screen data so every function returns zero. Or maybe somebody can explain how it can do GND mode without phisically shorting the input to ground, as every other scope i've ever seen do.
-if you terminate the input with a terminator you may see no added noise, just the front end noise which is adequate

[StillTrying]

Does anyone want to see what the bottom of the Siglent CML+'s 1kHz looks like at 100mv/div.

edit: I'll take that as a no, so here it is anyway! It looks much better, the probe is on X10, there's no control over the cml+'s much lower 5MSa/s sampling rate.

[IAmBack]

-1054Z doesn't a true GND mode. In the 1054Z when you choose GND coupling what it actually does is draw a line at zero on screen data so every function returns zero. Or maybe somebody can explain how it can do GND mode without phisically shorting the input to ground, as every other scope i've ever seen do.

I can see "double line" in "GND mode"... So no zero.

[Fungus]

-1054Z doesn't a true GND mode.... maybe somebody can explain how it can do GND mode without phisically shorting the input to ground

Why would it need one? GND mode is a throwback to analog 'scopes where you often lost the trace and needed a way to get it back on screen so you could figure out how to reset the horizontal/vertical controls or refocus the beam (remember when oscilloscopes had a "focus" knob? That's where the GND button comes from...)

These days we can just push the vertical position knob instead(!) 

I guess the fact that the haters always target the DS1054Z really just shows how good/popular it really is. 

I can see "double line" in "GND mode"... So no zero.

Right. Just something that's close enough.

[Fungus]

It looks identical to mine.

Mine has BY-AY less than 60 mV.

The base noise will be from radio transmitters in the area, etc., not from inside the 'scope.

[IAmBack]

It looks identical to mine.

Mine has BY-AY less than 60 mV.

The base noise will be from radio transmitters in the area, etc., not from inside the 'scope.

Probably, but I'm not sure if my scope is in "quiet" place. Anyway, not a big difference.

[jsantoro]

I've retired and got back into electronics as a hobby a while ago. I keep thinking I need to update some of my old equipment but after seeing posts like this I'm hesitant to replace my Tek 2247 with something new. Been looking at all these new DSO's and it seems they all have some sort of quirk. I want one but don't know why. Is it a case of I don't know what I'm missing? They do take up a whole lot less bench space.

[tggzzz]

I've retired and got back into electronics as a hobby a while ago. I keep thinking I need to update some of my old equipment but after seeing posts like this I'm hesitant to replace my Tek 2247 with something new. Been looking at all these new DSO's and it seems they all have some sort of quirk. I want one but don't know why. Is it a case of I don't know what I'm missing? They do take up a whole lot less bench space.

All equipment has its quirks.

Don't replace your Tek, but do augment it with a DSO. They have complementary advantages.

Do consider what you need from new equipment. If you can tolerate 10MHz bandwidth, then consider a Digilent Analog Discovery. It is a 14 bit scope (good for its spectrum analyser) with built-in AWG (so you can do bode plots of filters etc), digital pattern generator and logic analyser.

[Porcine Porcupine]

After having this oscilloscope awhile, I've become convinced the problem is a firmware bug causing it to be stuck in peak detect mode. I've posted it as a bug in the DS1000Z bug thread. Here's a link for future reference:

https://www.eevblog.com/forum/testgear/rigol-ds1000z-series-buglist-continued-(from-fw-00-04-04-03-02)/msg1442914/#msg1442914

[emocabras]

After having this oscilloscope awhile, I've become convinced the problem is a firmware bug causing it to be stuck in peak detect mode. I've posted it as a bug in the DS1000Z bug thread. Here's a link for future reference:

https://www.eevblog.com/forum/testgear/rigol-ds1000z-series-buglist-continued-(from-fw-00-04-04-03-02)/msg1442914/#msg1442914

I tested different frequencies and waveforms with the signal generator in my rigol ds1054z I can not see double line.

check you have power line problems, try to raise oscilloscope from the earth to do the test, to diagnose your local earth reference send strange frequencies.









5.00ns Dots

[Fungus]

Does anyone want to see what the bottom of the Siglent CML+'s 1kHz looks like at 100mv/div.

Yes!

edit: I'll take that as a no, so here it is anyway! It looks much better, the probe is on X10, there's no control over the cml+'s much lower 5MSa/s sampling rate.

"Much better" is relative, the point is it has the exact same problem. Thanks for confirming it!