DMM6500 "Glitch" on DCV

[mawyatt]

We just received a new DMM6500 to support an ongoing very complex and involved development project. We've noticed a seemingly random "glitch" propagating thur the system while observing 32 channels (system will have at least 128 channels) the system output waveforms with a MSO (SDS2104X+). At first we thought that this was due to something in the system causing the glitch, where the waveform would take a small "dip" for a short period and then return to normal. After some time carefully evaluating the system and tracing the source of the glitch which does not appear with the other DMMs involved (HP34401A, AG34401A, KS34465A, SDM3065X), it's caused by the DMM6500 while in DCV mode, and it attributed to the DCV Auto Ranging Function.

Has anyone experienced anything like this with the DMM6500?

It's interesting that none of the other 4 precision DMMs exhibit this behavior (yes, we are using all 4 DMMs simultaneously, actually we need a minimum of 12 more!).

Think what is going on is the auto range function is causing a DMM input impedance or a charge injection as "seen" from the DMM input to momentarily change and disrupting the waveform being monitored by the DSO.

Of course the work around is to turn off the Auto Ranging, but need to assess this quickly if it's a defective DMM6500.

Any responses from those knowledgable that have the DMM6500 is greatly appreciated.

Best,



Best,

[HighVoltage]

What is the simplest setup to repeat this glitch?
What does the waveform look like?
Picture of the glitch?

I have a DMM6500 and have not noticed anything in this regard.

[mawyatt]

Created a very simple waveform from the system to show the "glitch". It's just a Bi-Phase modulated precise DC voltage, thus creating a precision squarewave.

The glitch is very random and takes awhile to detect, but after a few thousand single traces finally captured one. Then moved the DMM lead to another system output and again after many single traces was able to capture another. The lines & outputs are not properly terminated so ignore the ringing, but you can see how the waveform has a small droop glitch using the "Zoom" DSO function. You can see the droop on the magenta trace just before the negative going transition on one channel, and then on the green trace after the transition on another channel (DMM moved to green trace).

Going to try and setup something "outside" the System and see if it also shows up.

Thanks for the help.

Best,

[SilverSolder]

What is being shown on the different scope channels?

[mawyatt]

Added a note above regarding traces.

If we add a 10K resistor in series with the DMM + lead the glitch doesn't appear, a 1K or lower and the glitch still appears randomly.

This seems to indicate that the DMM Auto Range is inter-acting in some unique way, and maybe going thru ranges where the input saturates and this is disturbing the DMM input in such a way that's propagating out the DMM input, likely in a subtle charge, capacitance or impedance change.

Here's another trace at slow 50Hz BiPhase Modulation we got lucky to capture, note the glitch droop in the Zoomed in yellow trace.

One possible future solution is to include a monitoring buffer pickoff from the driver waveforms that includes a series resistor, may include this in a breakout PCB for the DMM and scope monitoring points.

One of the reasons we're concerned, is others will become involved with this advanced research project, and utilizing multiple DMM6500s to carefully monitor the channels. These DMMs could cause a disruption in the advanced MMW beam steering system which the multiple waveforms control, and that would be a nightmare to trace back to the DMM as the source.

Could not duplicate this result using a AWG squarewave, even including a series resistance.

Anyway, thanks for help.

Best,

[tautech]

Here's another trace at slow 50Hz BiPhase Modulation we got lucky to capture, note the glitch droop in the Zoomed in yellow trace.

Hardly surprising when just using a standard edge trigger.

Try Interval and set it to trigger whenever the high time is less than what it should be and it will catch it reliably. 

[mawyatt]

The SDS2000x+ has been very helpful with this development. Here's what the setup looks like now, it will grow quite a bit just for the Phased Array Beam Steering Controller aspect, it's only ~25% now. Note the 32 precision pots, couldn't get the precision resistors & ICs required so had to revert to old school pots

When we get to 128 or 256 channels, ouch my head hurts

Siglent gear has been critical in the development of this system, as you can see. Looking forward to the new SDS6000 DSO, but at 8 bits we'll likely recommend a 12 bit Lecroy. We had to "park" the one SDP330X-E with the blue label on far right behind the Fluke DMM, too risky with current limit offset. The lab linear dual HV supply sits on top the dual SPD3303X-E (switcher HV supplies on far right), not in use since we need to carefully monitor the supply currents in the HV section and just using +-32V with the SDP3303X-E instead of the +-80V required full up, is good enough for now.

Best,

[mawyatt]

Ok, think we've figured this out.

The HP34401A introduces a random tiny glitch when we drop the Bi-Phase waveform modulation rate to low rates, can hear the relays clicking like crazy since it's trying to figure out what the range should be and what polarity the input is.

Best,

[tautech]

Snip....
using +-32V with the SDP3303X-E instead of the +-80V required full up, is good enough for now.

You can series multiple units of these PSU's up to their rated max above GND of 500V.
This spec is sorta hidden in the datasheet however it's an official feature of these PSU's although not widely known or publicized.

I believe SDS6000A beta testers have the Chinese version (SDS6000 Pro) which are not 8 bit however the units we will see in the west offer oversampling to 16 bit and will get a SDS6204A in a few months to replace my SDS5054X .....

[mawyatt]

Snip....
using +-32V with the SDP3303X-E instead of the +-80V required full up, is good enough for now.

You can series multiple units of these PSU's up to their rated max above GND of 500V.
This spec is sorta hidden in the datasheet however it's an official feature of these PSU's although not widely known or publicized.

I believe SDS6000A beta testers have the Chinese version (SDS6000 Pro) which are not 8 bit however the units we will see in the west offer oversampling to 16 bit and will get a SDS6204A in a few months to replace my SDS5054X .....

Yeah, we used a couple SDP3303s for creating +-~60V and that worked, but we still needed +-80V eventually. However, we also needed another 3303 to supply the required other voltages and after hosing up the one 3303 current offset, we decided to get a supply directly capable of +-120V. Outside the one 3303 I hosed up by improper cal procedure, the others are working well and providing great benefit.

Without some verifiable ENOB, I'm not sure I would trust a core 8bit ADC to be extended to 16 bits...that's an expansion of 256 and my experience with ADCs indicates some serious limitations, unless a vigorous self calibration procedure is involved. With a core ADC of 12 bits the single shot captures, just like the "glitch" we were chasing, are better served, as well as getting 16 times closer to 16 bits with some enhancing

Eventually extensive FFTs with dual tone IMD will revel the true performance, as this is getting into the signal processing and viewing domain of spectral analysis, where dynamic range is paramount. 

Not saying they can't pull an ENOB of well over 14bits from an 8bit ADC, but that will require a stellar 8 bitter for starters and extensive calibration to keep a reasonable bandwidth. Guess we'll soon find out

Best, 

[2N3055]

Ok, think we've figured this out.

The HP34401A introduces a random tiny glitch when we drop the Bi-Phase waveform modulation rate to low rates, can hear the relays clicking like crazy since it's trying to figure out what the range should be and what polarity the input is.

Best,

If you enable HI-Z on 34401A it will have GOhms of input resistance on ranges up to 10V, on 100V and up it will change to 10MOhm input. If signal is high impedance, that would be visible. Also it would change input capacitance..

[mawyatt]

If you enable HI-Z on 34401A it will have GOhms of input resistance on ranges up to 10V, on 100V and up it will change to 10MOhm input. If signal is high impedance, that would be visible. Also it would change input capacitance..

The signal is from an high voltage (180VPP) op-amp operated closed loop (G=16) with a series 1K resistor outside the loop, the 34401A was in standard 10Meg input mode I recall.

Since we now have 2 confirmed DMMs doing this, although not at the same Bi-Phase modulation rates, going to consider the effect as charge, or some type impedance change related to Auto Range and input polarity reversal.

Best,