Bench DMM Noise & Stability Comparisons with TestController

[KungFuJosh]

It's too late to do that now; I've already loaded the saved table of measurement values.
But you can still see that the reference isn't the worst.

You can't run a new 2 minute and 5 minute test with devices you still own?

[Martin72]

Of course I could have done that, but I just didn't do it earlier—for heaven's sake.

[KungFuJosh]

    

[TylerPeppy]

Oh man, I need to get my uDR1399 rev 2 board up and running, to try and produce an even more stable 1399 based reference.

Hoping to toss more stability measurements into the mix with this... But Josh has my last aged working ADR1399, so I might need to do some accelerated aging in a thermal chamber with some new pieces 

Tyler

[KungFuJosh]

Oh man, I need to get my uDR1399 rev 2 board up and running, to try and produce an even more stable 1399 based reference.

Hoping to toss more stability measurements into the mix with this... But Josh has my last aged working ADR1399, so I might need to do some accelerated aging in a thermal chamber with some new pieces 

Well, it was working before you gave it to me. 

I'll probably order a couple parts and fix it. Labels are our friends!

[TylerPeppy]

Oh man, I need to get my uDR1399 rev 2 board up and running, to try and produce an even more stable 1399 based reference.

Hoping to toss more stability measurements into the mix with this... But Josh has my last aged working ADR1399, so I might need to do some accelerated aging in a thermal chamber with some new pieces 

Well, it was working before you gave it to me. 

I'll probably order a couple parts and fix it. Labels are our friends!

I was messing with the idea of using BNC or triax cables on the outputs of my next module  to simplify cabling. I will probably stick with my Pomona plugs for now, but use colors that don't cause confusion 

[KungFuJosh]

I was messing with the idea of using BNC or triax cables on the outputs of my next module  to simplify cabling. I will probably stick with my Pomona plugs for now, but use colors that don't cause confusion 

Looking at the attached photo of the two references, tell me which one I'm going to destroy immediately, and which is probably safe?

Label it!!! Please? 

[TylerPeppy]

Confirmed, rev2 might include some input labels to differentiate the power inputs from dut outputs

In all seriousness, I am hoping to maybe use the intended Wago terminals I fitted last time for power input. Dut output remains undetermined.

[KungFuJosh]

Confirmed, rev2 might include some input labels to differentiate the power inputs from dut outputs

In all seriousness, I am hoping to maybe use the intended Wago terminals I fitted last time for power input. Dut output remains undetermined.

I get wanting to use those for power input, but I would recommend power input and 10V output both be clearly labeled, and look nothing alike. I do look forward to trying it though.

The first rule in product design is: make it difficult to destroy the product or the user.

[Salient666]

Please do test that! I'm thinking about building one or two of those.

Just to confirm, when you open the Setup box for the DMM6500, it shows filter off and sample count 1? If so, that's good to know. I know my DAQ6510 is noisier than it should be. Having something to compare is helpful.

Thanks,
Josh

I've done the test for the AliExpress both on the 10V output as well as the direct Zener voltage output.
The results are a little surprising because the difference between the two meters are larger than in my previous test with the DMMCheck Plus.
One thing is clear though, the zener output is much more stable so I think the components used to make the 10V are sub-par on this device.

[TylerPeppy]

Please do test that! I'm thinking about building one or two of those.

Just to confirm, when you open the Setup box for the DMM6500, it shows filter off and sample count 1? If so, that's good to know. I know my DAQ6510 is noisier than it should be. Having something to compare is helpful.

Thanks,
Josh

I've done the test for the AliExpress both on the 10V output as well as the direct Zener voltage output.
The results are a little surprising because the difference between the two meters are larger than in my previous test with the DMMCheck Plus.
One thing is clear though, the zener output is much more stable so I think the components used to make the 10V are sub-par on this device.

I can't fully tell just by looking at the design, but I am assuming this is the application note schematic from the LM399 datasheet? I can't tell if this is bootstrapped or not. If it is, I would be surprised by the results, as the bootstrapped designs typically have low noise from what I have seen.

The board design is also a bit wonky; I would have placed the LM399 on an isolated island with zener power coming in one side, and zener output on the other side. I would have also raised it and placed the LM399 in a foam block to avoid air currents, and provide resistance to temperature fluctuations. The LM7815 being close to other components as a main source of heat could also cause some drift to the gain set resistors and the op-amp I reckon.

Also, no knowing if these are "fell into a reject bin at the factory" type parts either, there could be a whole host of other things too. Just my two cents though.

I am highly surprised that the DMM6500 is so much noisier than the 34465A. That is quite concerning...

Edit: Looking at your data tiles again, does anyone notice the slow rise and fall of the average output voltage pattern looking sinusoidal? Or is that just me seeing things?

Tyler

[TUMEMBER]

What's more puzzling is what's shown in the "standard deviation." This isn't normal. Both meters operate on the same range (10V and 7V), which is still the same range. But on the KS SD at 7V, the reading dropped to less than 1 micron (5 times better) than on the DMM6500. At 10V K34465, it was only twice as good—what's going on there?

[KungFuJosh]

I am highly surprised that the DMM6500 is so much noisier than the 34465A. That is quite concerning...

My DAQ6510 and 34465A have a similar relationship. I like the DAQ better, but the noise seems to be higher. Adding a filter can help, but that wouldn't be appropriate for these games.

The AliEx refs are cheap toys; you shouldn't overthink their design.

[Martin72]

   

Here you go, just finished it....
(Auto range, 10 PLC, 200µV span, appx 90min warmup time DMMs+Reference before start)

[Salient666]

I did a shorted input test to see what happens then.
I guess my wiring/external influences are a major part here?

With shorted inputs I get 1ppm for the Keysight and 0,7ppm at the Keithley (60 minutes test).

[TylerPeppy]

I am highly surprised that the DMM6500 is so much noisier than the 34465A. That is quite concerning...

My DAQ6510 and 34465A have a similar relationship. I like the DAQ better, but the noise seems to be higher. Adding a filter can help, but that wouldn't be appropriate for these games.

The AliEx refs are cheap toys; you shouldn't overthink their design.

But this is odd, in the pictures Salient posted, the DMM6500 was nearly 5x higher SD compared to the Keysight...

[Kleinstein]

The test with the shorted input and auto range would test the 100 mV range and this is usually dominated by the protection and amplifier noise, while the reference noise has essentially no effect.
The test with 10 V has quite some effect from the external reference, the meter internal reference and some from the ADC, while the amplifier noise is usually negligible.
The test with manual 10 V range and a short would mainly test the ADC with only a little effect of the amplifier and protection.
So this are different tests and one meter can be good with one aspect and be poor with another.

There is also the length of the test that can have an effect: the reference noise of the LM399 found in most meters here has popcorn noise that does jumps from time to time. This may happen roughly every few minutes and there can well be 2 min intervals with no such jump. This way on can have 2 min intervalls with pp noise of maybe 1 µV (without a jump) or 7 µV (with 1 or more jumps). So the results can be quite random.
Another point is that residual noise of the meters (ADC and amplifier) can be white noise - typical for HP/KS meters or with some extra LF component (0.01-2 Hz)  - typical for Keithley meters. This can make the Keithley meters look good for very short runs (e.g. 5 seconds) and poor for longer ones.

[KungFuJosh]

   

Here you go, just finished it....
(Auto range, 10 PLC, 200µV span, appx 90min warmup time DMMs+Reference before start)

Thank you! Can you please select SI7 for the SDM4065A on the Current values tab? Then the info boxes will relate better.

Thanks,
Josh

[KungFuJosh]

I did some short tests last night. Here's the DAQ6510. Auto (100mV) and 10V ranges.

Thanks,
Josh

[Salient666]

I did some short tests last night. Here's the DAQ6510. Auto (100mV) and 10V ranges.

Thanks,
Josh

I have now done the same tests with my DMM6500

Edit: It seems indeed that with AutoZero OFF the noise is less.

[KungFuJosh]

It looks like with a short our meters are much closer. However, your AZ noise is more significant than mine for some reason. 🤷

10V range we both get about 15uVpp with AZ off. With AZ on, I was about 16uVpp, and yours around 20uVpp. That's not huge on the grand scheme of things, but it is curious.

[TylerPeppy]

It looks like with a short our meters are much closer. However, your AZ noise is more significant than mine for some reason. 🤷

10V range we both get about 15uVpp with AZ off. With AZ on, I was about 16uVpp, and yours around 20uVpp. That's not huge on the grand scheme of things, but it is curious.

I haven't followed super closely, are you guys using panel mount 4W shorts? If not, I can do some 2W shorts and see how they compare.

[Salient666]

It looks like with a short our meters are much closer. However, your AZ noise is more significant than mine for some reason. 🤷

10V range we both get about 15uVpp with AZ off. With AZ on, I was about 16uVpp, and yours around 20uVpp. That's not huge on the grand scheme of things, but it is curious.

It’s close enough for me.
I still cannot explain though why on one evening I get certain results from my dmmcheck plus with both meters and the next day with the same cabling completely different (worse) results. One measurement was late in the evening and the other the next morning. There must be some sort of external interference?? I know my “Lab” is very noisy anyway, I cannot use any NCV device there for instance, the moment I pick it up it starts to beep immediately, no matter where I point it at.
Also the manual calibration of my scope fails there, I have to take it to another room to make that successful.

[KungFuJosh]

It looks like with a short our meters are much closer. However, your AZ noise is more significant than mine for some reason. 🤷

10V range we both get about 15uVpp with AZ off. With AZ on, I was about 16uVpp, and yours around 20uVpp. That's not huge on the grand scheme of things, but it is curious.

I haven't followed super closely, are you guys using panel mount 4W shorts? If not, I can do some 2W shorts and see how they compare.

Yes, 4W shorts. If you don't have some, get some.

This is the one I was using: https://www.aliexpress.us/item/3256808269105832.html
I also have this one: https://www.aliexpress.us/item/3256804393374448.html

Both are good enough / the same "brand."

[Martin72]

Thank you! Can you please select SI7 for the SDM4065A on the Current values tab? Then the info boxes will relate better.

I'll do that next time around, and I'll use battery power again, too...

Martin