Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[skander36]

Ok. I'll repeat. Again and again ... for you 
The problem is that if I leave it to warm-up it is out of specs. No it will not stabilize. After 3 hours see below pic (Ref Voltage is 1.000007), it still drift even if more slowly.
And the fact that your meter does work ok and stabilize after warm-up just show that mine has an issue. This is why I'm saying about the gambling.
That wasn't fixed or even signaled to me by Siglent when it was to them.
Datasheet: Yes you can compare the datasheet from other brands when the values claimed to be the same or better in case of Siglent.

[coromonadalix]

well if you put if far from the others, does it stabilize ?

some of them may or have a fan on the rear or on the side ... it could negate the warm up to some extend,  i was never found of stacking things like this

each item of mine has a separated box / enclosure for them, they don't share air volume or affect the other    if i may say

[KungFuJosh]

Ok. I'll repeat. Again and again ... for you 
The problem is that if I leave it to warm-up it is out of specs. No it will not stabilize. After 3 hours see below pic (Ref Voltage is 1.000007), it still drift even if more slowly.
And the fact that your meter does work ok and stabilize after warm-up just show that mine has an issue. This is why I'm saying about the gambling.
That wasn't fixed or even signaled to me by Siglent when it was to them.
Datasheet: Yes you can compare the datasheet from other brands when the values claimed to be the same or better in case of Siglent.

Thank you. 😉

Here's the thing though, that's a single picture. It doesn't tell me anything about drift. It tells me that the measurement of one meter is different than the other meters.

It not being the same shows there's a difference in accuracy, but a single measurement doesn't tell me anything about drift.

I'll assume your reference is perfectly accurate and stable at 1.000007. I agree, your Siglent has the worst accuracy of the 3 meters based on that assumption. Is that out of spec though? Since I was clearly bad at calculating that previously, let's quote:

100.0000% - .0040% = 99.9960% or 0.999960.  4 nines.  The lower threshold would be 10V x 0.999960 = 9.99960V.

10.000007 x 0.99996 = 9.99960699972. Your post-warmup test is within spec, despite being the least accurate of the 3 meters.


After 3 hours warmup, does that less accurate meter stabilize within stated specs, or does it continue to drift? That's the remaining question.

Thanks,
Josh

[KungFuJosh]

each item of mine has a separated box / enclosure for them, they don't share air volume or affect the other    if i may say

I fully agree with that in theory, and keep all my meters separated, but I think the Siglent is probably the least affected by this as they all use side fans. I don't know which side the DMM6500 exhausts, but if it exhausts to the left, maybe that's affecting the 34465A. 🤷

The Siglent exhausts to the right.

[skander36]

well if you put if far from the others, does it stabilize ?

some of them may or have a fan on the rear or on the side ... it could negate the warm up to some extend,  i was never found of stacking things like this

each item of mine has a separated box / enclosure for them, they don't share air volume or affect the other    if i may say

No, it won't stabilize. There was years where it was my single benchmeter and it was in different rooms / places.

[KungFuJosh]

No, it won't stabilize. There was years where it was my single benchmeter and it was in different rooms / places.

Well, if you're ever bored enough, I'd love to see a drift test chart in TestController (and the advanced readout) after it's had a few hours warmup.

[skander36]

Ok. I'll repeat. Again and again ... for you 
The problem is that if I leave it to warm-up it is out of specs. No it will not stabilize. After 3 hours see below pic (Ref Voltage is 1.000007), it still drift even if more slowly.
And the fact that your meter does work ok and stabilize after warm-up just show that mine has an issue. This is why I'm saying about the gambling.
That wasn't fixed or even signaled to me by Siglent when it was to them.
Datasheet: Yes you can compare the datasheet from other brands when the values claimed to be the same or better in case of Siglent.

Thank you. 😉

Here's the thing though, that's a single picture. It doesn't tell me anything about drift. It tells me that the measurement of one meter is different than the other meters.

It not being the same shows there's a difference in accuracy, but a single measurement doesn't tell me anything about drift.

I'll assume your reference is perfectly accurate and stable at 1.000007. I agree, your Siglent has the worst accuracy of the 3 meters based on that assumption. Is that out of spec though? Since I was clearly bad at calculating that previously, let's quote:

100.0000% - .0040% = 99.9960% or 0.999960.  4 nines.  The lower threshold would be 10V x 0.999960 = 9.99960V.

10.000007 x 0.99996 = 9.99960699972. Your post-warmup test is within spec, despite being the least accurate of the 3 meters.


After 3 hours warmup, does that less accurate meter stabilize within stated specs, or does it continue to drift? That's the remaining question.

Thanks,
Josh

About 3-4 hours it seem to stop drifting.
Yes a picture can't tell about drift but will be more clear if I tell you that it started from 1.000002 and reach 0.999979.
Also I am not sure if this math is correct because the range of Siglent is 200 mV and it has 2200000 counts.

[mawyatt]

We are guilty of stacking instruments one on top the other and side by side, not good practice but we have very limited space available.

Talk about warm up time, the old HP/AG34401s take many hours. We just replaced the fading display in the old HP34401 (type with all red terminals) and waited a couple hours for the AG34401A and HP34401A to agree. These don't have fans which has good and bad connotations 

Best,

[skander36]

...
Well, if you're ever bored enough, I'd love to see a drift test chart in TestController (and the advanced readout) after it's had a few hours warmup.

There is no warm-up time specified in manual. In service they calibrated the dmm after 90 min.
In that interval the dmm continue to drift. I didn't see anywhere a 3-4 hour period for warm-up. This will be stupid.
Again the fact that your dmm stabilize after a resonable period show that my dmm has an issue.

[skander36]

We are guilty of stacking instruments one on top the other and side by side, not good practice but we have very limited space available.

Talk about warm up time, the old HP/AG34401s take many hours. We just replaced the fading display in the old HP34401 (type with all red terminals) and waited a couple hours for the AG34401A and HP34401A to agree. These don't have fans which has good and bad connotations 

Best,

Nice! A real lab!

[KungFuJosh]

About 3-4 hours it seem to stop drifting.
Yes a picture can't tell about drift but will be more clear if I tell you that it started from 1.000002 and reach 0.999979.
Also I am not sure if this math is correct because the range of Siglent is 200 mV and it has 2200000 counts.

IIRC, I think it was defpom that said 4 hours was best for the 3065.

The math is correct, each range has a specific accuracy stated in the datasheet.

...
Well, if you're ever bored enough, I'd love to see a drift test chart in TestController (and the advanced readout) after it's had a few hours warmup.

There is no warm-up time specified in manual. In service they calibrated the dmm after 90 min.
In that interval the dmm continue to drift. I didn't see anywhere a 3-4 hour period for warm-up. This will be stupid.
Again the fact that your dmm stabilize after a resonable period show that my dmm has an issue.

The datasheet states 90-minute warmup time.

Maybe waiting 4 hours is stupid, but if it works, it works. Many labs don't shut their bench DMMs off at all.

I agree, my 3065X is likely better than yours, but yours appears to meet accuracy specs. Only a drift test after a minimum of 90 minutes will show whether it meets drift requirements.

For the sake of experimenting, it would be nice to see your potentially faulty meter with a drift test performed after 90 minutes, and again after 4 hours warmup.

Thanks,
Josh

[mawyatt]

We left our SDM3065X powered on for weeks and months 24/7 when we got it, sometimes still leave it on for long periods. This is/was to help "age" the un-aged LM399 used as a reference. KS/AG/HP and Keithley used aged references which helps with the better initial long term stability.

Best

[skander36]

We left our SDM3065X powered on for weeks and months 24/7 when we got it, sometimes still leave it on for long periods. This is/was to help "age" the un-aged LM399 used as a reference. KS/AG/HP and Keithley used aged references which helps with the better initial long term stability.

Best

I've done in the first year, but mine even after 6 years still has a drift that cancel his precision.

[mawyatt]

We left our SDM3065X powered on for weeks and months 24/7 when we got it, sometimes still leave it on for long periods. This is/was to help "age" the un-aged LM399 used as a reference. KS/AG/HP and Keithley used aged references which helps with the better initial long term stability.

Best

I've done in the first year, but mine even after 6 years still has a drift that cancel his precision.

Well that does seem defective. If you have to "keep it" and without warranty support a "shot in the dark" would be to replace the LM399.

For ~$20 might be worth a gamble and maybe pre-age the new LM399 before installing. With Scotts calibration routine this maybe possible, however please take all this with a grain of salt, as we have no idea what this entails

Best

[skander36]

...
Well that does seem defective. If you have to "keep it" and without warranty support a "shot in the dark" would be to replace the LM399.

For ~$20 might be worth a gamble and maybe pre-age the new LM399 before installing. With Scotts calibration routine this maybe possible, however please take all this with a grain of salt, as we have no idea what this entails

Best

Thank you mawyatt for your advice!
As I have the other two meters that works well, I think that Siglent will be available for this.
But I have to find good references,because the pre-aged LM399 that I found was pulled components that have pop-corn noise. Maybe is better to buy new from Digikey or Mouser and keep them powered for about 1 year.
Thanks!

[Kleinstein]

LM399 essentially all have popcorn noise:usually 2 levels some 3-4 µV aprart and jumping between them roughly on minute scale.
Unless there is more popcorn noise the pulled reference would be OK. It is likely already aged quite a bit.

If one wants better, one could consider an ADR1399.  It needs a little more reference current (e.g. ~2 mA instead of ~ 1mA) and an RC element in parallel to the output.

The LM399 usually has a quite low TC. The main point here is the quality of the temperature regulation and thus not dependent on a trim. So the drift warm drift shown before is very likely not from the reference but more from other parts in the DMM: the divider from 7 to some 5 V for the ADC ref. and the divider from some 20 V to 5 V for the ADC.
With the SDM3065 the 2 V range may actually be the more stable one. It may be worth to check the warm up drift also for the 2 V range.
One could also bring out the 7 V ref. voltage and just check its drift directly.

[Korchahin]

While I'm waiting for the ADR1399 demo board, a question arose regarding self-calibration. If you take an ADR1399 board and, using a resistive divider and stable resistors, measure the voltages and write them down. Then manipulate the meter. And calibrate and measure using the same board using preliminary data. Is it possible to do this?

[TurboTom]

...
For ~$20 might be worth a gamble and maybe pre-age the new LM399 before installing. With Scotts calibration routine this maybe possible, however please take all this with a grain of salt, as we have no idea what this entails
...

Since the LM399 isn't single-handedly responsible for generating the reference voltage -- usually, there's at least one OPAMP and several low TC resistors involved as well -- I would take a closer look at the whole reference section before taking invasive action.

Using one of your more stable meters in high input impedance DCV mode, you may want to have a look at the voltage directly across the LM339 to find out if this is already drifting. You may also want to check a series resistor (if there is any) to look for stability of the reference current. As far as I experienced, the LM399's are quite reliable components -- either they work or they don't (rarely). The problem may also relate from anywhere between the LM399 and the ADC. Maybe even a component in the signal path can lead to the observed problem. The trouble is that the measurements take so long everytime, so a proper diagnostics takes a lot of time and discipline. Hence, it's questionable if going through this effort makes sense.

Anyway, Siglent sould have taken your complaint more seriously and just replaced your instrument. Actually, in a case like this, and knowing the way those companies' service departments usually work, I might have felt the urge to introduce some "more obvious" problem in the instrument during the warranty time in order to "convince" the manufacturer that a replacement is the best option... 

[skander36]

...
For ~$20 might be worth a gamble and maybe pre-age the new LM399 before installing. With Scotts calibration routine this maybe possible, however please take all this with a grain of salt, as we have no idea what this entails
...

Since the LM399 isn't single-handedly responsible for generating the reference voltage -- usually, there's at least one OPAMP and several low TC resistors involved as well -- I would take a closer look at the whole reference section before taking invasive action.

Using one of your more stable meters in high input impedance DCV mode, you may want to have a look at the voltage directly across the LM339 to find out if this is already drifting. You may also want to check a series resistor (if there is any) to look for stability of the reference current. As far as I experienced, the LM399's are quite reliable components -- either they work or they don't (rarely). The problem may also relate from anywhere between the LM399 and the ADC. Maybe even a component in the signal path can lead to the observed problem. The trouble is that the measurements take so long everytime, so a proper diagnostics takes a lot of time and discipline. Hence, it's questionable if going through this effort makes sense.

Anyway, Siglent sould have taken your complaint more seriously and just replaced your instrument. Actually, in a case like this, and knowing the way those companies' service departments usually work, I might have felt the urge to introduce some "more obvious" problem in the instrument during the warranty time in order to "convince" the manufacturer that a replacement is the best option... 

Thank you! Indeed I'm aware of the complexity of the reference stage and without pick-up the schematic  from the board and analyze before I wouldn't start.
About the Siglent, it was nice if they at least tell me that it has an issue with some drift observed by them.

[skander36]

While I'm waiting for the ADR1399 demo board, a question arose regarding self-calibration. If you take an ADR1399 board and, using a resistive divider and stable resistors, measure the voltages and write them down. Then manipulate the meter. And calibrate and measure using the same board using preliminary data. Is it possible to do this?

You said that just acquired the meter. If it's new it should not need any calibration. Do you have any suspicion about accuracy?
Even if you use an ADR1399 board(or any other non labeled/calibrated board) it must be calibrated with a calibrated meter first (7.5-8.5 digit recomended).
To check for drift you can use TestController.
To check for the accuracy you can use the Excell calculator from R&S(attached).

Regards,

[Martin72]

About the Siglent, it was nice if they at least tell me that it has an issue with some drift observed by them.

If they had noticed it.
After the warm-up phase, the Fluke calibrator is connected and if the values are within the tolerance range, the device is fine.
How long will it take, an hour, two hours...
If you have noticed a drift over a longer period of time that falls outside the tolerance range, then I would complain about the device accordingly.
However, if the meter stays within the range, unfortunately it's bad luck, then yours just drifts more than others.

[skander36]

About the Siglent, it was nice if they at least tell me that it has an issue with some drift observed by them.

If they had noticed it.
After the warm-up phase, the Fluke calibrator is connected and if the values are within the tolerance range, the device is fine.
How long will it take, an hour, two hours...
If you have noticed a drift over a longer period of time that falls outside the tolerance range, then I would complain about the device accordingly.
However, if the meter stays within the range, unfortunately it's bad luck, then yours just drifts more than others.

Also if I had known that is an automatic procedure, I would have asked specifically, but i believed that they will check given the request : calibration and adjustment. 

[Korchahin]

Stability test video - SDM3065 and other

[KungFuJosh]

For the price of the Keithley, you could buy 2 SDM3065Xs, one of them being the SDM3065X-SC. That's why I don't have a Keithley.

I definitely want a DMM with a scanner card. 2 bench DMMs isn't enough for me, and I don't think 3 will be enough either. 😉

[skander36]

With Keythley DMM 6500 you can can have a scan card for about 50-80 $.
This is mine.