EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: jacekowski on January 17, 2022, 04:41:47 pm
-
Hi All,
I have recently purchased brand new MSO5104 which exhibits very strange behaviour with CH1 and CH3 enabled at the same time (it also happens with just CH1 but it is less obvious),
Pictures show the problem quite nicely, both channels are fed with the same 120MHz or 130MHz sine wave, however CH1 shows a "spike" or "artifact" happening randomly (can be quite easily detected by setting trigger level so it only just triggers).
My local distributor replaced the scope once with a unit where the issue was almost not noticeable at 120MHz, but it is still there and noticeable at 130MHz.
Has anybody noticed a similar issue on their unit, also, does it happen on higher bandwidth units as well? (i don't want to spend money on upgrading bandwidth to test it on potentially faulty scope and at the same time i don't want to end up with warranty issues if i upgrade it via alternative means).
-
Are you 100% sure that you signal generator is producing totally clean sine waves?
Not saying it isn't but ...
D.
-
I don't have an explanation, just some observations/questions:
a) It looks to me like a problem with the signal to me. Is it the exact same signal/cable connected to both inputs? What happens if you swap the connections around, does it happen on channel 3?
b) Upgrading the bandwidth for testing is really, really cheap. Just find the MSO5000 hacking thread.
c) Those traces look really thin. I've been assured that Rigol MSO5000s are unusable due to ADC noise so what gives? Anybody...?
-
Pendrix:
I'm sure signal generator is ok (i'm feeding same signal into both ch1 and ch3 and issue disappears when ch2 or 4 is also enabled).
Fungus:
a) issue stays on ch1
b) i don't want to void my warranty yet
c) single capture so no persistence.
-
Can you run the self calibration (if possible / present)? If the problem dissapears with more channels enabled, it could be a calibration problem when 2 ADCs are used in parallel. The sine waves also look distorted.
-
Fungus:
a) issue stays on ch1
It doesn't make sense that it would be only on some peaks and not all of them. What percentage of peaks show this? (approx)
Does it do it at lower frequencies or only when you go past the bandwidth limit?
-
I don't have an explanation, just some observations/questions:
a) It looks to me like a problem with the signal to me. Is it the exact same signal/cable connected to both inputs? What happens if you swap the connections around, does it happen on channel 3?
b) Upgrading the bandwidth for testing is really, really cheap. Just find the MSO5000 hacking thread.
c) Those traces look really thin. I've been assured that Rigol MSO5000s are unusable due to ADC noise so what gives? Anybody...?
It's in Stop mode Fungus but the trigger point is finding the noise.
-
Can you run the self calibration (if possible / present)? If the problem dissapears with more channels enabled, it could be a calibration problem when 2 ADCs are used in parallel. The sine waves also look distorted.
This.
Rigol uses parallel overlapped ADCs and this looks exactly like it does when ADCs are not properly calibrated and scaled. Run autocal and check again.
-
c) Those traces look really thin. I've been assured that Rigol MSO5000s are unusable due to ADC noise so what gives? Anybody...?
It's in Stop mode Fungus but the trigger point is finding the noise.
So it's in averaging mode and the glitch is repeating periodically enough for it to smooth out the signal...?
-
Pictures show the problem quite nicely, both channels are fed with the same 120MHz or 130MHz sine wave, however CH1 shows a "spike" or "artifact" happening randomly (can be quite easily detected by setting trigger level so it only just triggers).
If you are feeding them the exact same signal--by this I would think you mean one channel from the signal generator and a cable with a tee--then how can they be horizontally displaced in the first picture?
Can you specify exactly how the input is set up--type of sig gen, cables, etc, perhaps a photo?
Is it possible to ever see this issue if you trigger from CH3? Do the glitches always appear at the trigger point or can you catch it happening on a non-trigger cycle?
-
I forgot to ask:
Do you have cables 50Ohm terminated at the scope side??
-
So it's in averaging mode and the glitch is repeating periodically enough for it to smooth out the signal...?
Presumably this is a single-shot and the trace will only be one sample wide regardless of how much noise there is. Look at how terrible the sine shape is--if that is due to noise, imagine a large numbers of them with different distortions all laid over one another.
-
So it's in averaging mode and the glitch is repeating periodically enough for it to smooth out the signal...?
Presumably this is a single-shot and the trace will only be one sample wide regardless of how much noise there is. Look at how terrible the sine shape is--if that is due to noise, imagine a large numbers of them with different distortions all laid over one another.
I was looking at the first image. The flat spots at the bottom of the waves should be noisy, right?
But let's not derail the thread with that until we find out what a self-cal does and/or where the glitch is coming from.
If it's only on CH1 then we can suspect the 'scope.
OTOH OP's story of "exact same signal" doesn't add up because as you rightly pointed out there's a big phase difference between the two channels.
-
nctnico: 2N3055:
Calibration does nothing to fix it.
Fungus:
Approximately less than 10% more than 1%.
It does it at lower frequencies but it is a lot harder to trigger on it reliably and find it (and it is a lot less obvious).
Shift was caused by daisy chaining the channels, most other tests were done with them being fed via equal lengths cables from same T (see attached photo).
Also, you will probably not see any significant noise due to the time base and sample rate (it's only 20 samples per division)
bdunham7:
I've been using siglent SDG2122X up to 120MHz and for signals above 120Mhz i've been using HackRF.
I've tried two different setups, with T coming out of signal generator into bnc/coax into 50ohm terminators at scope end (as shown in attached photo) OR generator into BNC/coax into T connected to ch1 (or ch3) and into another BNC/coax to other channel (which caused shift between the channels).
It is possible to catch the glitches without using trigger but it takes longer to find a nice one.
It seems like i might be looking at the same issue as here:
https://www.eevblog.com/forum/testgear/review-rigol-mso5000-tests-bugs-questions/msg3187486/#msg3187486 (https://www.eevblog.com/forum/testgear/review-rigol-mso5000-tests-bugs-questions/msg3187486/#msg3187486)
Also, i have a tektronix mso3054 here that does not show any issues with the signal.
-
Can you use the same setup as in the photo, set the trigger at about the zero volt level, offset the channels by 1 division and then let it run in the 'normal' mode and then post a screenshot?
-
Just adding one more thing here, the glitch happens on the bottom side of the sinewave as well (see attached).
You can see that there is something happening around the bottom of ch1 (yellow) trace, while pink(purple?) has nothing around it.
On the last one i've also enabled colour grading as it seems to make the issue more visible.
-
That looks like noise of some sort to me. Can you do all three of those again but triggering on CH3?
-
Triggering from CH3.
-
nctnico: 2N3055:
Calibration does nothing to fix it.
Also, i have a tektronix mso3054 here that does not show any issues with the signal.
As a last ditch effort I'd try to swap the cables and use 50 Ohm in line terminators. However if a different DSO shows the signal correctly (using single shot acquisition without averaging / persistence enabled), then I think it is time to return the Rigol MSO5000 and ask for a refund.
-
Your setup looks good and I think you said you physically swapped the connectors at the scope and the problem stays with CH1. It just looks like CH1 is noisy or has some strange distortion or reaction to the input signal. Was the first scope also noisy on CH1 and was it more or less than this?
-
nctnico:
I'm already using 50Ohm terminators but those don't do anything to the issue.
bdunham7:
1st scope was slightly worse than 2nd one.
-
The only thing that seems a bit baffling is why it goes away when you select CH2 or CH4. I wonder if the 'noise' might be an instability in the ADC timing, as others previously referred to when the asked you to autocal the scope.
-
The only thing that seems a bit baffling is why it goes away when you select CH2 or CH4. I wonder if the 'noise' might be an instability in the ADC timing, as others previously referred to when the asked you to autocal the scope.
Or the path between the two ADCs is picking up noise. I can imagine that there is a path (PCB trace) that brings the signal from channel 1 to the ADC of channel 2. If that is routed badly, then it could pick up noise.
-
nctnico:
On the first scope i've managed to catch something that looks like maybe the samples from ADCs are arriving out of order. Maybe it is still the same thing, but interpolation is somehow making it appear a lot worse.
-
This looks more like the samples are not aligned in time properly. I notice you are using vector mode which is a linear interpolation method; it basically connects the dots with straight lines so I don't expect any interpolation artefacts to occur. But the image above shows a slightly different effect than when 2 channels are enabled; then you also see peaks beyond the signal amplitude. Typically a self calibration cycle should fix these kind of issues. If not then something is severely broken.
-
The only thing that seems a bit baffling is why it goes away when you select CH2 or CH4. I wonder if the 'noise' might be an instability in the ADC timing, as others previously referred to when the asked you to autocal the scope.
Or the path between the two ADCs is picking up noise. I can imagine that there is a path (PCB trace) that brings the signal from channel 1 to the ADC of channel 2. If that is routed badly, then it could pick up noise.
There are no two or whatever ADC chips inside scope. It has single ADC chip (Ankaa SP ASIC) that has 1 2 and 4 paths. Inside ADC are several (at least 4) interleaved ADCs. They went Keysight way of inside that chip can actually having 64 slower interleaved ADC.. We don't know for sure. There is a patent, they are using parallel ADC inside chip.
That Ankaa chip is also said to have oscilloscope specific DSP functions. That could be some corrections or what not.
And with massively parallel ADC, adjusting them all together to achieve low noise and monotonic performance is hard..
-
Just for another reference: SDG2042X, 2Vpp Sin @ 120 MHz from channel 1 via a bnc “T” on the SDG output, set to 50 ohms, in to channels 1 & 3 of each of the scopes (MSO5000 and RTB2000), which are each set to 1:1 with a 50 ohm terminator on each channel input, via a 50 ohm impedance cable (RG316), bandwidth limiting off.
Guess we need to send our RTBs back too, haha. Actually not very worried about this - there are quite a few variables that could contribute, I would guess. Though it's an interesting phenomenon, likely related to the effects of the sampling mechanisms/relatively low bit-depths/sig gen jitter/trigger mechanisms, I don't see it as a major issue, though I certainly could be wrong, and am not afraid to admit it.
This kind of reminds me of the (over?) obsession about input noise levels, though I can certainly understand the OP wanting to understand exactly what’s going on.
In my experience I actually don't mind the slightly higher noise floor on the Rigol, compared to the RTB. Its waveform update rate is many times more than that of the RTB, esp. at low signal levels/small time scales, and it seems like it's actually easier to see some low-level signals on the Rigol, maybe because it's not filtering so aggressively, and likely other factors, too. I'd rather see a bit more noise, rather than filter out more of the signal.
Here, the Rigols get used more frequently than the RTBs simply because they have numerous features the RTB scopes lack (Nth-edge triggering, at least 10 Mbps UART decoding, zone triggering, etc.) and the generous array of options on the Rigol scopes allow a wide variety of tasks to just get done, every day. In my opinion, for the money especially, they are bad-ass powerful tools that can bring in the cash -- as long as you're aware of the limitations -- just like _any_ tool.
I do personally like much about the RTB scopes, though – mainly the web interface, which is the best I’ve seen for a scope so far. And, they have been adding sorely missed features … very slowly. The FFT is also very nicely done and looks beautiful, too. But still, the Rigols are the go-to tools for many jobs where the added features really come in handy – crude though they may sometimes be, they get the job done.
Don’t really like touch screens on scopes at all, and am very thankful these instruments work so well with a wireless mouse! This is another area where the Rigol wins over the RTB – no need to use the touch screen, ever.
In the end, as I'm sure you all know, it's what you accomplish using the tools, rather then the obsession over the instruments themselves -- that's what's important (to me, at least...)
Would love to try Siglent scopes sometime -- if they ever decide to give the user full control over zoom functions. As they are now, they're a bit like a manual mode on a DSLR that decides it will set certain parameters for you -- whether you want it to or not. No thanks!
Good luck on any continued investigation of this phenomenon, and hope the RTB image will provide some added perspective. Will watch from afar. All the best!
[attachimg=1]
[attachimg=2]
-
Would love to try Siglent scopes sometime -- if they ever decide to give the user full control over zoom functions. As they are now, they're a bit like a manual mode on a DSLR that decides it will set certain parameters for you -- whether you want it to or not. No thanks!
Seems you won't have to wait much longer as alternative memory management options are to be implemented in the new 12 bit HD model.
https://www.eevblog.com/forum/testgear/siglent-sds2000x-hd-12bit-(published-for-chinese-domestic-market-only)/ (https://www.eevblog.com/forum/testgear/siglent-sds2000x-hd-12bit-(published-for-chinese-domestic-market-only)/)
-
Seems you won't have to wait much longer as alternative memory management options are to be implemented in the new 12 bit HD model.
https://www.eevblog.com/forum/testgear/siglent-sds2000x-hd-12bit-(published-for-chinese-domestic-market-only)/ (https://www.eevblog.com/forum/testgear/siglent-sds2000x-hd-12bit-(published-for-chinese-domestic-market-only)/)
Been wondering whether the new model might have addressed that -- excellent news! Would be great if it could filter down the line with a f/w update at some point.
Will be keeping an eye out -- thanks for the info!
-
Seems you won't have to wait much longer as alternative memory management options are to be implemented in the new 12 bit HD model.
https://www.eevblog.com/forum/testgear/siglent-sds2000x-hd-12bit-(published-for-chinese-domestic-market-only)/ (https://www.eevblog.com/forum/testgear/siglent-sds2000x-hd-12bit-(published-for-chinese-domestic-market-only)/)
Been wondering whether the new model might have addressed that -- excellent news! Would be great if it could filter down the line with a f/w update at some point.
Will be keeping an eye out -- thanks for the info!
I have a hunch this is in the new SDS6000A too but until I get mine in a month or so....no comment. ;)
There was talk of implementing it in existing models but don't know where that's gone as if they do they sure don't want to break any existing functionality. :scared:
-
Seems you won't have to wait much longer as alternative memory management options are to be implemented in the new 12 bit HD model.
https://www.eevblog.com/forum/testgear/siglent-sds2000x-hd-12bit-(published-for-chinese-domestic-market-only)/ (https://www.eevblog.com/forum/testgear/siglent-sds2000x-hd-12bit-(published-for-chinese-domestic-market-only)/)
Been wondering whether the new model might have addressed that -- excellent news! Would be great if it could filter down the line with a f/w update at some point.
Will be keeping an eye out -- thanks for the info!
I have a hunch this is in the new SDS6000A too but until I get mine in a month or so....no comment. ;)
There was talk of implementing it in existing models but don't know where that's gone as if they do they sure don't want to break any existing functionality. :scared:
"I have a hunch this is in..."
No need to hunch neither guess.
Simply: It is there in SDS6kA: Memory management. Normal auto and then if user really need for some rare or weird purposes there is also available fixed sample rate and fixed memory.
SDS6000A Series Digital Oscilloscope User Manual (https://www.siglenteu.com/download/11336/) version EN01A, page 90: Memory Management
SDS6000A Quick Start (https://www.siglenteu.com/download/11339/) version EN01A, page 25: (D) Select the Memory Management mode (Auto, Fixed Sample Rate, and Fixed Memory)
-
Don’t really like touch screens on scopes at all, and am very thankful these instruments work so well with a wireless mouse! This is another area where the Rigol wins over the RTB – no need to use the touch screen, ever.
Example Siglent SDS6000A UI is made so that if user want he do not need use touch panel or physical control panel at all. It can well use alone with wired or wireless mouse. (and naturally also with all control methods mixed)
-
Guess we need to send our RTBs back too, haha. Actually not very worried about this - there are quite a few variables that could contribute, I would guess. Though it's an interesting phenomenon, likely related to the effects of the sampling mechanisms/relatively low bit-depths/sig gen jitter/trigger mechanisms, I don't see it as a major issue, though I certainly could be wrong, and am not afraid to admit it.
This kind of reminds me of the (over?) obsession about input noise levels, though I can certainly understand the OP wanting to understand exactly what’s going on.
Neither of your photos looks unusual and certainly doesn't show any issues like the ones the OP is reporting.
-
Example Siglent SDS6000A UI is made so that if user want he do not need use touch panel or physical control panel at all. It can well use alone with wired or wireless mouse. (and naturally also with all control methods mixed)
Great to hear! Have appreciated your posts over the years -- this is another good one. Haha!
Siglent continues to impress. It's exciting to watch...
-
Simply: It is there in SDS6kA: Memory management. Normal auto and then if user really need for some rare or weird purposes there is also available fixed sample rate and fixed memory.
SDS6000A Series Digital Oscilloscope User Manual (https://www.siglenteu.com/download/11336/) version EN01A, page 90: Memory Management
SDS6000A Quick Start (https://www.siglenteu.com/download/11339/) version EN01A, page 25: (D) Select the Memory Management mode (Auto, Fixed Sample Rate, and Fixed Memory)
But decoding, math, measurements, etc are still limited to what is on screen. So it is still useless :palm:
But this is getting wildly offtopic here.
-
Here Rigol MSO5074 CH1 and CH3, , 2 x 50 ohm cables with 50 ohm terminators.
Siglent SDG2042X, 120Mhz sinus, 2Vpp, 50 ohm output.
It seems there are no problems. I don't see any peaks like yours.
Just in case, check the installed firmware version.
v00.01.03.00.03 2021/10/18
- Optimized waveform display in XY mode.
- Optimized the DC gain calibration algorithm.
- The La channel is decoded in parallel, which solved the problem of decoding error in negative polarity.
Any firmware optimisation regarding ADC calibration can help.
-
Please use the same time scale the OP uses; likely this is key to reproducing the issue. Also, the issue may not show always so it could be missed. Also try 130MHz instead of 120MHz.
-
Please use the same time scale the OP uses; likely this is key to reproducing the issue. Also, the issue may not show always so it could be missed. Also try 130MHz instead of 120MHz.
I have used the same time scale and 500mV/div, 120MHz. I don't have a 130MHz signal.
Almost no problem. The maximum peak I can find it's like in the attached picture. But it can be just noise, my environment is not clean. Near the oscilloscope it's a powerfull desktop and in the same room a wi-fi router.
I have attached another 100mV/div capture, with the maximum peak I can find.
-
Can you change trigger slope and set trigger voltage as low as it goes and check what you see there?
-
Please use the same time scale the OP uses; likely this is key to reproducing the issue. Also, the issue may not show always so it could be missed. Also try 130MHz instead of 120MHz.
I have used the same time scale and 500mV/div, 120MHz. I don't have a 130MHz signal.
Almost no problem. The maximum peak I can find it's like in the attached picture. But it can be just noise, my environment is not clean. Near the oscilloscope it's a powerfull desktop and in the same room a wi-fi router.
I have attached another 100mV/div capture, with the maximum peak I can find.
If it is external noise, then I would expect both traces to show the exact same signal disturbance. However, that doesn't seem to be happening.
-
If i've well understood, as it's specified in the first post, the problem on one scope was at 120mhz, and on a replacement unit "moved" at 130mhz, so maybe it's not on all the scopes at the same frequency (so in yours at 120mhz can be fine, but show it at 130mhz).
Maybe the artifact/noise can move a bit.
-
Can you change trigger slope and set trigger voltage as low as it goes and check what you see there?
I have done for CH1 falling slope trigg, CH2 also, and CH3. There are no differences. See captures.
I have tried freq down to 100MHz, no changes.
If you want to check something else, just let me know. Limited to 120Mhz.
-
Please use the same time scale the OP uses; likely this is key to reproducing the issue. Also, the issue may not show always so it could be missed. Also try 130MHz instead of 120MHz.
I have used the same time scale and 500mV/div, 120MHz. I don't have a 130MHz signal.
Almost no problem. The maximum peak I can find it's like in the attached picture. But it can be just noise, my environment is not clean. Near the oscilloscope it's a powerfull desktop and in the same room a wi-fi router.
I have attached another 100mV/div capture, with the maximum peak I can find.
If it is external noise, then I would expect both traces to show the exact same signal disturbance. However, that doesn't seem to be happening.
Indeed. So it can be the internal noise / ADC artifacts.
-
The MSO5000 has a reputation of being noisy. Down on the 50mV/div region you're bound to experience some of it. Not sure if that setting is actually analog gain or just zoom in on 100mV.
-
The MSO5000 has a reputation of being noisy. Down on the 50mV/div region you're bound to experience some of it. Not sure if that setting is actually analog gain or just zoom in on 100mV.
I have this MSO5074 and Siglent SDS-1104X-E. Down to 20mV/div the noise is about the same. At 10mV/div I can see clear the difference in favor of Siglent.
But the signal's refresh rate is much more faster on Rigol. On Siglent is "frozen",the equivalent of a 4x average on Rigol. I believe it can be a different way to process the signal, apart from the much faster sample rate of Rigol.
Under 5mV/div there is no dubt noise advantage in favor of Siglent. The lower you go, the bigger the noise difference.
For me both of them are very useful, I like both.
Important for me is to know their strengths and weaknesses and use them accordingly.
-
Most likely you have a faulty scope from the same batch as the first scope. But you are in a small group as most persons would have hacked the scope to the full bandwidth, so it could be related to the low pass filter on ch1 which limits it to 100M.
Try the new firmware as that has a calibration update, check FFT of CH 1 vs CH 3 and see if anything is noticeable different between the two.
-
But you are in a small group as most persons would have hacked the scope to the full bandwidth, so it could be related to the low pass filter on ch1 which limits it to 100M.
Try the new firmware as that has a calibration update, check FFT of CH 1 vs CH 3 and see if anything is noticeable different between the two.
I would have hacked it, but i wanted to wait with voiding my warranty until i'm sure the thing is 100% working.
Latest firmware did nothing (it was the first thing i did).
Rigol went back to the distributor for a refund so i can't do any more testing on it.
-
I would have hacked it, but i wanted to wait with voiding my warranty until i'm sure the thing is 100% working.
It wouldn't void your warranty, it's just the same as installing an official upgrade.
Plus it can be reset to factory state just as easily as it can be hacked.
-
As far as i know installing unofficial "upgrades" voids the warranty, wherever they can prove it is another thing (but then me admitting that on a public forum wouldn't be a good idea).