Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[rolfdegen]

Fan comparison

Delta Fan from Siglent



Quiet Noctua fan

[KungFuJosh]

In other words, rough match the airflow, and get 10dB quieter; ~41% noise reduction.

[mawyatt]

Do folks with the SDS3065X see random "jumps" as shown here connected to a 10V reference (Vref10) which is also connected to our newer KS34465A as shown (same scales). The time span is 1hr, and vertical scale is 30uv total range, so roughly ~15uv "jump" or ~ 1.5ppm. 

Know this is most likely from the LM399 SDS internal reference and we've seen this many times before but curious about what other folks "see" with their SDM3065X DMMs?

Best,

[Kleinstein]

The jumps of a LM399 ref. are usually on the order of 3-4 µV or 0.5 ppm. So 15 µV would be about 3 times to large for the LM399 typical jumps.

With 2 meters in parallel, there is the possibilty that one meter sees the disturband from AZ switching of the other meter.  It may be worth checking if the problem also exists without the 2nd meter.

[mawyatt]

As mentioned we've seen this often, and without anything connected to the SDM3065X except the reference and with other references, so it's definitely coming from the SDM3065X!! Don't see this behavior with any of our other seven total 6 1/2 digit DMMs, just the SDM3065X.

Agree, 1/2 ppm not too bad, 1.5 ppm is concerning and at some point may consider changing the LM399.

Hopefully some folks will report what they "see" with their SDM3065X DMMs.

Best,

[rolfdegen]

Could it be the VREF10 modul ? In the range of 70-75°F the voltage difference is approximately 15uV.

VREF10-001 Typical Temperature Graph

[mawyatt]

Could it be the VREF10 modul ? In the range of 70-75°F the voltage difference is approximately 15uV.

VREF10-001 Typical Temperature Graph

No it's not the VREF10 Reference, if it was this would show up on both the SDM and KS34465A. Here's a plot with just the SDM showing another "jump", so also not the parallel connection of both meters. WE've also seen this with other References we have, 2 based upon LM399 and LTZ1000, all with over 2 years continuous aging.

Also, temperature in "The Labs" doesn't typically vary more than +-2F with set point at 77F and SDM shows a relative "quick 15uv jump" which could not be caused by a temperature change and it changes up and down. In fact it's jumped up and down ~15uV a couple times in the past 3 hours and the temperature is 77.5F and hasn't changed more than 0.5F.

Best,

[KungFuJosh]

I just powered on my meter and a ref. Not sure my ref is good enough to bother, but I'll report back later with the results.

[Kleinstein]

The resolution looks rather limited. At what speed is the DMM running. The scaling is also not telling much just as the KS original. So it is a bit hard to judge on the noise and step size.

For the 4 lm399 refs I have measured and the reports found so far the popcorn noise jumps are pretty consistant. Much larger jumps could as well be from other parts then just the reference.
As a test one could have a longer time series with a stable resistor (e.g. 20 K) and see if there are comparable jumps somehow. This would test slightly different parts of the meter, eliminating the reference.

[mawyatt]

As mentioned the vertical scale on both the SDS and KS are 30uv full scale, the horizontal scale on both is 1 hr, so we were seeing ~15uV "jumps" in both directions. Just like the dreaded "Popcorn" noise the level would jump up by ~15uv stay awhile then jump back down by the ~ same amount.

The resistor is a good idea, we have some precision Caddock's we'll try!!

Edit, maybe not such a good idea see post below, however may still prove useful info wrt to various sections of the DMM and which might be the major noise contributor.

Best,

[Nanitamuscen]

Popcorn may not be visible when measuring resistances, because the same voltage source is the reference voltage for the ADC and the reference voltage of the current source.

[KungFuJosh]

Here's my SDM3065X set at 100PLC connected to an AD584 ref.

[mawyatt]

Popcorn may not be visible when measuring resistances, because the same voltage source is the reference voltage for the ADC and the reference voltage of the current source.

Good catch!!! Caught me when I wasn't thinking

Best,

[mawyatt]

Here's my SDM3065X set at 100PLC connected to an AD584 ref.

Thanks, if you can let it run for a day or so and see if you spot any of what we've shown.

Best,

[Kleinstein]

Popcorn may not be visible when measuring resistances, because the same voltage source is the reference voltage for the ADC and the reference voltage of the current source.

Good catch!!! Caught me when I wasn't thinking

Best,

The resistance test would not see popcorn noise from the reference, but it could still see popcorn noise from other parts of the meter. So if similar jumps are still there in the resistance mode, they are very likely not from the LM399. The reference is not not only possible source for popcorn noise.

[mawyatt]

The resistance test would not see popcorn noise from the reference, but it could still see popcorn noise from other parts of the meter. So if similar jumps are still there in the resistance mode, they are very likely not from the LM399. The reference is not not only possible source for popcorn noise.

Just as Nanitamuscen indicated, and know that the Reference isn't the only potential source for popcorn but the most likely!!

So far our original assumption of the LM399 seems to be proving out, no apparent popcorn noise on the resistance measurement.

Best,

[rolfdegen]

Hello everyone. I have now finished my modified Siglent SDM3055. I didn't use the various low noise adapters for the Noctua fan. They had a resistance of 50 ohms and 147 ohms. Instead I soldered a 75 ohm resistor to the fan's 12V connection. He is very quiet. I also gave the processor a heat sink.

[mawyatt]

No signs of this popcorn type noise measuring a precision resistor. Returned to the 10V reference and this behavior is back again, so pretty much points to SDM3065X LM399!!

Best,

[KungFuJosh]

Here's 7 hours at 10V, 10PLC. Reminder that it's a aliexpress AD584 ref. 😉





What PLC is your meter set to?

[KungFuJosh]

This is the ref I was using. I think it did pretty well for what it is. The lower voltages certainly appear more stable; though I've never run the other voltages as long as this test.

[rolfdegen]

I also use an AD584KH with 10V reference voltage. I only have an SDM3055. Here are my measurements...

AD584KH 10V => 10.0001V (SDM3055)


SDM3055 Long-term measurement cold start. Measured voltage 10.0001V


SDM3055 Long-term measurement warm start. SDM3055 was turned on an hour before. Measured voltage 10.0001V

[rolfdegen]

My AD584-M has a LiPo battery and I charged it before taking the measurement. Charging caused the battery to heat up and the reference voltage had a different value. When cold, the 10V setting had a voltage of 10.0000 - 10.0001V. Immediately after charging the battery, the voltage rose to 10,0004 volts. I waited an hour and the voltage was back to 10,0000V.

AD584-M circuit diagram


https://www.markhennessy.co.uk/ad584_references/

[mawyatt]

Here's 7 hours at 10V, 10PLC. Reminder that it's a aliexpress AD584 ref. 😉





What PLC is your meter set to?

It is 10PLC and a sliding 1hr window. Here's result just taken with SDM and DMM6500, both similar settings.

Points too SDM LM399 internal reference!!

Edit: Site flipped images over, had to reload again

Best

[KungFuJosh]

It is 10PLC and a sliding 1hr window. Here's result just taken with SDM and DMM6500, both similar settings.

Points too SDM LM399 internal reference!!

Pretty weird! I obviously don't have that issue with mine, maybe you should contact Siglent? Is it still under warranty?

[alm]

Popcorn noise is a known issue that some fraction of the LM399s exhibit. References can be selected for it, which as far as I know most big manufacturers do. But since it was previously discussed here that Siglent may not be doing any burn-in on their references before shipping the unit, it may well be they don't do any selection either, so a fair fraction of their meters might be expected to exhibit popcorn noise.