Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[HKJ]

I do generally like locking the front panel, probably because I often use TestController for automatic logging and testing.
Using a handheld meter with TestController will usually avoid any locking the user interface and TC works just as well with a handheld meter* as with a bench meter.
*Mostly meters with a serial computer connection, HID interface is problematic.

[Giuss]

How the dual display feature works? Is possible fo example to measure voltage and current at the same time? The manual doesn't explain too much

[oz2cpu]

How the dual display feature works? Is possible fo example to measure voltage and current at the same time? The manual doesn't explain too much

DUAL switches between the two modes, try to activate it you hear it,

[tautech]

How the dual display feature works? Is possible fo example to measure voltage and current at the same time? The manual doesn't explain too much

Simple overview of Dual measurement capability here:
https://www.siglenteu.com/operating-tip/dual-measurement-connections-for-voltage-period-frequency-and-current-measurements-for-sdm3000-series-dmms/

[bateau020]

Be aware, on a SDM3055, latest firmware (1.01.01.25), the bug mentioned in https://www.eevblog.com/forum/testgear/siglent-new-bench-dmm-sdm3055/msg3779651/#msg3779651 is present. Use manual range settings for DCV/DCI, then it works correctly.

[Tjuurko]

SDM3065X (firmware 3.01.01.10)
5 years of waiting are over: it became possible to work with a PT100/PT1000 temperature sensor using a 4-wire circuit.
It remains to find out the secret SCPI-command to enable this mode.

[oz2cpu]

did anyone address the odd random voltage drops seen on SDM3055 ?
see the attached curve..

it is NOT the input voltage, it is not a temperature thing (verified by other instruments in parallel)
blue raw data from SDM3055
red 10 times avarage
data fetch every 3 sec via ethernet,
SDM3055 been powered on at least one week constantly,
I have seen this ALWAYS, happens about one time pr day, sometimes it looks a bit different.
software issue ??

[Kleinstein]

A 50 µV drop is about 10 x more than to expect from a LM399 reference. The time scale is also quite slow with the lower values for some 10-20 minutes. So I don't think this would be some normal popcorn noise - it is unusually  slow and also rather large amplitude for this. The 3055 uses a bandgap reference, but I would not expect that kind of noise from a bandgap ref.  A pure SW issue would be strange, as it happens so rarely and than for a long time with still some variations  - maybe more like some EMI / interference from WLAN or similar that is doing a new channel assignment / reset about daily.

Still 50 µV on a 20 V range would be still below the resolution of the 5.5 digit meter - so on the display those jumps would hardly be visible.

It this type of jump only happening when measuring a significant votlage or is it also happening when measuring zero volts ? This information could help to see if the error is more an additive or multiplicative.

[tautech]

did anyone address the odd random voltage drops seen on SDM3055 ?
see the attached curve..

it is NOT the input voltage, it is not a temperature thing (verified by other instruments in parallel)
blue raw data from SDM3055
red 10 times avarage
data fetch every 3 sec via ethernet,
SDM3055 been powered on at least one week constantly,
I have seen this ALWAYS, happens about one time pr day, sometimes it looks a bit different.
software issue ??

V1.01.01.25 firmware ?

[oz2cpu]

>V1.01.01.25 firmware ?

YES SIR

[oz2cpu]

to kleinstein, I have been running tests with multible meters in parallel,
it is only a "thing" on my SDM3055
in a few hrs one more SDM3055 show up here, then i prove if this also happens on that one,
AND more important ?? is it in sync ? or just as random, time will tell,
it is also using same fw

any one also who own SDM3055, also own a nice clean 10V supply to test with ?
perform same zoom ? curve ? need to be running a few days

[oz2cpu]

it is too erly to come with big conclutions,
however see the two new curves,
both units been running 24hrs, before i started this measurement.

blue is a brand new SDM3055
red is my one year old SDM3055

same FW and same SW as reported by the screen,
the input is the SAME voltage LM399
the only modification done is offset so the curves are nicely placed in the 0.5mV full screen range
data fetch every 3 sec.

tomorrow i can go pickup another DMM and run all 3 in parallel with the same input
and reset the test.

[CDN_Torsten]

Interesting findings...the step looks very 'binary'. 
Can we assume that both DMMs are located in the near-exact same physical location and are orientated the same? (in case this voltage step is being caused by some external influence)

[oz2cpu]

thanks for your reply CDN_Torsten,
they are located directly on top of each other, and powered with the same Y mains cable, nothing else nearby is turned on,
i have been running this test for weeks now, and it looks about the same, the binary drops are NOT in sync with day/night
or ambient temperature, if it was anything from an external source, i would belive we should see the same, on the other unit
at the very same time. ?
I plan to keep this running a few days,
(i go nuts if this is not going to happen on the other SDM3055, this means mine is defective,
and since this is way under spec, noise or accurasy or resolution, no one from Siglent is going to fix or repair or replace it..)

[mawyatt]

This is very interesting as we see this same type "jump" on our SDM3065X!

Although not as high, more like 5~10uV jumps. The setup is also similar with multiple DMMs (dual KS34465A, DMM6500) reading 7V or 10V from a LM399 (or LTZ1000) reference. This was noted some time ago and we began to leave the 3065X powered ON hoping to reduce this effect thinking it was due to the internal LM399 reference and attributed as "popcorn noise", thus aging the LM399 should help. The frequency of the jumps seems to be lower now after a few months of powered ON, and hopefully will get better over time.

Frankly I'm disappointed in the performance of the 3065A, and why we've selected the DMM6500 and another KS3465A as our DMM needs grow. Even much prefer the pair of old HP/AG 34401As we have to the 3065A.

We first observed this "jumping" when doing long Trend Charts on multiple DMMs, only the 3065X showed this behavior mid-last year.

Understand the 3065X uses a LM399, but the 3055 uses another reference? If, so then maybe something else common to both DMMs is causing this jump??

Best,

[oz2cpu]

both my curves are SDM3055 units,
when i had an SDM3065, i am so sorry i did not perform this test
yes the SDM3055 uses MAX6325 2.5V 1ppm ref. and AD7190 ADC 24 bit

[Kleinstein]

For a LM399 reference it is relatively common to see jumps in the 4 µV range. With 10 V signal this would be some 6 µV.
I would not habe too much hope the LM399 popcorn noise to improve with aging. It may get better, but it may as well get worse.

I would be surprised to to see jumps in the reference votlage to see some 50 µV at 10 V reference - so some 12 µV for a 2.5 V reference.
The reference is a bit lower grade, but it should not have that much noise. The weakness is more with the TC and thermal hysteresis.

[mawyatt]

You might consider the excellent DMM6500 and KS34465A. After you use these awhile in some real work situations then you'll get a good perspective. IMO these are in a different DMM league than the 3065X for actual lab (research) use, but YMMV.

Best,

[mawyatt]

Since both DMMs are experiencing this "jump", although at different levels, and both have different references, maybe something else is contributing?

Best,

[oz2cpu]

mike ? how do you conclude this :

>Since both DMMs are experiencing this "jump", although at different levels, and both have different references, maybe something else is contributing?

by looking at this attached result, of two SDM3055 in parallel,
I say : it is proven, that the jump is NOT from the DC input voltage,
right ??

PS: yes yes, i know, there is also a KS34465A showing up tomorrow :-)
but that is not what this thread is about

[Kleinstein]

With the 2 parallel readings it is clear the error is not from the reference souce.  It looks like only one of the 3055 is effected and not both. Somehow the jumps are quite similar (same size and short / long pulse) for a pure analog effect and still too variable for a software problem.

The 50 µV jump hight is rather large for normal popcorn noise, even though the 20 V range starts with a divider. So at the actual input amplifier and ADC input it would be only 5 µV of a jump. From teardown pictures it looks like they use AD8629 OPs for the amplifier, an this AZ OP should not show popcorn noise of any significant level.
To narrow the source down, maybe da a test with either shorted input (no more effect from the meters reference) or a smaller reference in the 2 V range (no more divider at the input).
Another point may be to run a few small amplitude temperature cycle to see if this can provoke similar jumps.

[mawyatt]

mike ? how do you conclude this :

>Since both DMMs are experiencing this "jump", although at different levels, and both have different references, maybe something else is contributing?

by looking at this attached result, of two SDM3055 in parallel,
I say : it is proven, that the jump is NOT from the DC input voltage,
right ??

PS: yes yes, i know, there is also a KS34465A showing up tomorrow :-)
but that is not what this thread is about

Was referring to our SDM3065X and your SDM3055.

Best,

[oz2cpu]

Kleinstein, our luck, in my unluck, just checked my invoice and it is 5 days out of waranty
so this means i am free to open it, and figure this out, once and for all :-)
but i wait a few days until the two units in parallel test is over.
have we any schematics ? or is all know-how based on close up pictures of the pcb ?
the idea about shorting the input, and see if it is still there is good, maybe it is a 20V range issue ?
so i should stay in this range, for the first test to make sense ?

Mike , Was referring to our SDM3065X and your SDM3055.
ok, thanks, yes now it makes sense :-)

[Kleinstein]

My knowledge about the HW is to a large part from the close up pictures shown - there are not that many parts and even without seeing the connections one can get a good idea on how the input stage is about build. The SDM3055 should still be relatively simple - the 3065 seems to be a bit more tricky, getting 20 V with high impedance.
Some more information is from the alternative / hacked firmware version some-one from France developed. This gave the hint that there seems to be no amplification before the ADC and the lower range gain is supposed to be from using the ADC internal gain steps. This makes absolutely sense as this gain is quite stable and already part of the ADC.

If it is an issue with the input divider part / 20 V range, chances are the problem may not show up with a short. A shorted input would not show a change in the divider ratio. The shorted input test would mainly check if the error is more like an additive (kind of offset error) or a multiplicative error.

When doing a test with a short, one may as well do alternating readings in the 20 V and 2 V range (e.g. 1 minute interval). At least so far the jumps are for quite some time.  So one could test the 2 ranges simlultaneously. Chances are the input switching could use a mechanical relay, and this would also be a candidate for the error.  Mechanical switching a low level signal can be tricky with oxidized contacts. So one may even consider alternating between the 20 V and 200 V range and still have the 10 V test signal. Some excercise may help to clean the relay contacts.

The 20 V range is with 10 M input impedance and thus in theory contact resistance at the input can be an issue too.

[oz2cpu]

thanks a lot kleinstein, you give me new hope,
i allready plan to open it, and run the test while open, and then knock on the relays while looking at the curves,
i do think it is very random, and i feel it is getting worse, not better,
i can also solder a wire from the ADC input, and from the REF and perform parallel measurements
to reveal where this odd jump come in, if i cant reveal it on ADC input or REF, it can only be the ADC chip i need to replace
since it contain all the gain stuff too, i think i give it a high value of possibility, even before i have performed any measurements,
wrong assumption ??
hey : we can even play a game : who can guess where it is, before we start to dig :-)
lets see who is the lucky ? or the most clever ?