Author Topic: Siglent SDM3055 VAC RMS bug or strange behavior  (Read 4734 times)

0 Members and 1 Guest are viewing this topic.

Offline gwidemanTopic starter

  • Contributor
  • Posts: 36
  • Country: us
Siglent SDM3055 VAC RMS bug or strange behavior
« on: July 28, 2022, 09:50:53 am »
I have an SDM 3055 with latest firmware (1.01.01.25), and it appears to have a bug in the 200mV AC range.

The issue occurs when measuring waveforms that have a low "duty cycle", in which case the value on the 2V AC scale is reasonable, but switch to the 200mV AC and it's wildly off.

I first encountered this bug when measuring chopped sine waves of the sort you would see from a triac or SCR power control circuit, when set to a low ON-angle. But I have since reproduced it with a plain old square wave with low duty cycle.

Example waveform:


Here's the correct VAC value on the 2V range. It agrees with two other connected meters, and the RMS value on the scope display.


And here's the display if you switch to 200mV range.


This occurs whether you change the range manually, or allow the meter to autorange.  The amount of error on the 200mV range varies, seemingly worst at the top of that range.  It does not appear for ordinary sine waves. I have not explored other waveforms, but as noted above, it may be particular to waveforms with low duty cycle.

I'm pretty sure this is a bug, unless someone can tell me why this is legit behavior!

Graham
« Last Edit: July 28, 2022, 09:55:46 am by gwideman »
 

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 3180
  • Country: us
Re: Siglent SDM3055 VAC RMS bug or strange behavior
« Reply #1 on: July 28, 2022, 12:22:22 pm »
These these DMMs utilize an analog RMS chip and likely at the lower 200mv range the analog chip is saturating on the high peak values of the narrow pulse. This is a high crest factor waveform and thus subject to this type of waveform "limiting".

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 
The following users thanked this post: gwideman, 2N3055, ch_scr

Offline alm

  • Super Contributor
  • ***
  • Posts: 2837
  • Country: 00
Re: Siglent SDM3055 VAC RMS bug or strange behavior
« Reply #2 on: July 28, 2022, 12:38:10 pm »
According the SDM 3055 spec sheet the maximum crest factor of signals for ACV measurements (at full scale) is 3. What is the crest factor of your signal?
 
The following users thanked this post: gwideman

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14017
  • Country: de
Re: Siglent SDM3055 VAC RMS bug or strange behavior
« Reply #3 on: July 28, 2022, 01:43:34 pm »
The crest factor for this signal should be around 5 (a little over 2 V peak to peak and with AC coupling likely some 1.8 V peak and 360 mV RMS).

So the signal is too high in crest factor and should not use the full range. So if a high crest factor is possible, the user should not use the AC ranges up to 100% FS, but more like only up to 50% or 25% and if in doubt also check the next higher range. The reduced range than may not leave much headroom to the point where the accuracy is reduced (the specs are for > 5% of FS).

The point to complain about is that the meter does no recognize the limiting and indicates an error message. Ideally the meter should detect the internal clipping and show it.
 
The following users thanked this post: gwideman

Offline gwidemanTopic starter

  • Contributor
  • Posts: 36
  • Country: us
Re: Siglent SDM3055 VAC RMS bug or strange behavior
« Reply #4 on: July 29, 2022, 10:21:28 am »
Thanks to everyone who replied on this thread -- very informative and on-point answers which make me glad I asked.

I do see the point of view that this measurement misapplies the meter because a signal with this high of a crest factor exceeds an item in the meter's specification.

However, I'd argue that one of the main purposes for the AC RMS measurement features is in mains power measurement, where control of power by way of SCR-switching is commonplace, not to mention other MOSFET-switched  methods, all of which result in chopped sine waves.  And where power is adjustable is just the sort of place you'd want to measure it, and its RMS value.

It would seem fairly inevitable that a user would be drawn into using the meter in exactly the circumstance where the chopped sine wave exhibits a high crest factor, without even realizing it, and get a quite bogus result.  I agree with @Kleinstein that we would expect the meter to alert us to the high crest factor, and if set to autorange, just switch to a higher range.

I'm not sure whether it's feasible for Siglent to address this in firmware, but if possible, I hope they do. In my observation, it's the only meter, out of four that I had on this task, that had this behavior. (I suppose the others might have also hit a problem, just not at the exact same values. )
 

Offline mawyatt

  • Super Contributor
  • ***
  • Posts: 3180
  • Country: us
Re: Siglent SDM3055 VAC RMS bug or strange behavior
« Reply #5 on: July 29, 2022, 11:35:15 am »
I'm not sure whether it's feasible for Siglent to address this in firmware, but if possible, I hope they do. In my observation, it's the only meter, out of four that I had on this task, that had this behavior. (I suppose the others might have also hit a problem, just not at the exact same values. )

Don't think this is possible with the DMMs that utilize the analog RMS chips since they have no way of knowing what the input Crest Factor is. Meters that utilize the Digital Method to extract the RMS from the DMM input, like the KS34465 and DMM6500, may have this ability and preform better with high Crest Factor waveforms.

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14017
  • Country: de
Re: Siglent SDM3055 VAC RMS bug or strange behavior
« Reply #6 on: July 29, 2022, 12:38:15 pm »
I'm not sure whether it's feasible for Siglent to address this in firmware, but if possible, I hope they do. In my observation, it's the only meter, out of four that I had on this task, that had this behavior. (I suppose the others might have also hit a problem, just not at the exact same values. )

There is likely not much that can be done in firmware. It would need some extra hardware, like extra comparators to detect when the signal exceeds the clipping limits or even better hwrdware to also measure the positive / negative extremes. AFAIK a few meters like Keithley2001 /2002 also measure the peak voltage - though not sure if the firmware actually checks the peak values when doing an RMS measurement.

A possible way would be to measure the same signal also in the next higher range and check if the results are consistent. However due to the relatively slow settling of the RMS converter this is not really practical for the firmware.  So it would be the users reposibility to do this check if a high crest factor is suspected and more than some 25% of the range are used.
With digital RMS things are easy, and the autoranging can use the peak voltages.
 
The following users thanked this post: gwideman


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf