Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[mawyatt]

The readings provided are just for comparison between the two 34401As, the KS34465A DMMs, and the SDM3065, so no claim for absolute accuracy should be considered, only relative measurements.

Hmmm.  How closely do the three HPAK models agree, how old are they and when/where were they last calibrated?

One HP34401A is very old (thus HP nomenclature), the other is Agilent 34401A thus later. These were disassembled and repaired, then ad-hoc in-house calibrated to the newer KS34465A. So not attempt at absolute accuracy is claimed, just relative and drift comparisons.  Likely we'll be getting another new KS34465A or KS34461A for our on-going project, when/if they become available.

One thing I was not happy about with the newly acquired SDM3065X was the cal date is listed as 2/2020, and per the recommended cal cycle was already over twice the recommended "storage time". Here's Siglent Calibration/Certification statement, shown in red, then see blue.

SIGLENT has determined that the factory calibration of this instrument is not significantly affected by storage of up to 180 days before first-time use. Cal Interval should start at the time the unit is placed in service or 180 days past the “Date of Calibration” on the certificate received with the unit.

So this states that when our "new" SDM3065X was placed into service in 8/2021, factory calibrated 2/2020, and it was already over 1 year out of calibration per Siglent's recommendations when we open the box

So we can't place much "faith" in absolute SDM3065A readings at this time since it's technically out of calibration, so only as a comparison with the KS34465A and the 34401As. We haven't raised a ruckus over this yet, since we've been totally consumed with many other "family" issues over the past few weeks

Edit: Didn't answer your question. All three, the two 34401As and the KS34465A agree within +-10uv when using the 10V reference. Also agree very well elsewhere.

Best,

[bdunham7]

Edit: Didn't answer your question. All three, the two 34401As and the KS34465A agree within +-10uv when using the 10V reference. Also agree very well elsewhere.

Well there is really only one calibration point there since the others were adjusted to the 34465A.  What I look for is meters that haven't been calibrated in a long time and are not interrelated in any way.  When those agree, I think that's probably even more reliable than most calibrations and you can get a better sense of how these things work long term.

I'm a bit surprised that Siglent would ship a meter that far after the cal date, but it probably won't matter as far as accuracy.  If you needed the paper, OTOH, they would need to correct that issue--and it sounds like they would if you needed them to.

[mawyatt]

Well there is really only one calibration point there since the others were adjusted to the 34465A.  What I look for is meters that haven't been calibrated in a long time and are not interrelated in any way.  When those agree, I think that's probably even more reliable than most calibrations and you can get a better sense of how these things work long term.

That's why we used the term relative rather than absolute, accuracy can only be inferred with the KS34465A, everything else is relative. If the SDM3065A had been within the calibration cycle, then it could be used as well. Also why getting another KS34465A or 34451A was mentioned.

I'm a bit surprised that Siglent would ship a meter that far after the cal date, but it probably won't matter as far as accuracy.  If you needed the paper, OTOH, they would need to correct that issue--and it sounds like they would if you needed them to.

I was told by Saelig when we ordered that this was the last they had in stock, we had requested they ship the latest inventory. However, we did not realize the cal date until much later, when reviewing things that was caused by the other SDM3065A thread regarding drift, and thinking it should have been well within this year at worst!

Best,

[Kleinstein]

With the short supply it looks like they really go to the bottom of the inventory and include old ones that collected dust.

Chances are that in few months there is no more problem with units that have old cal - more like finding units calibrated just before shipment.

With so much initial drift, the initial calibration does not seem to very valuable anyway, unless you need the papers.

[klausES]

...Don't understand, what's the 9V block??...

[Bad_Driver]

My SDM3065 is now running for nearly 300 h. Source is as recommended a 9 volts (battery) block in a styropors box.

Today I fiddled with my HP 34401 to build a wifi-rs232 bridge because my always running server is too far away for
a rs232 cable.

I was very surprised to see that both meters now show exactly the same value - ongoing for hours.
See attached photo.

Yes, it‘s clear that this is a lucky coincidence but nice to see.

I let it now run up to 500 hours. This should be enough.

[Bad_Driver]

Some first measurements after keeping the SDM3065 running for now 480h.

But first a disclaimer to avoid unnecessary discussions:
- It's my hobby, 35 years ago I was involved in measurement topics at university but this is long, long ago
- a real measurement needs at least a 7.5 digit DMM fully calibrated, I can only compare relatively to my HP34401a (20y+ old)
- for some days I got the strong desire for appropriate Voltage references but reading thru the threads here convinced me that it needs
  a lot of effort and funding to get something good enough for 6.5 digit DMMs and again access to a much better calibrated DMM
- all measurements were done with the help of HKJs great software TestController https://www.eevblog.com/forum/testgear/program-that-can-log-from-many-multimeters/
- temperature was always 23...24 C in my lab.
- all readings done with Hi-Z (10Gohms) and 100 PLC

My observations:

First surprise: The readout of the SDM is 3 digits more than displayed see photo but the last 2 make no sense to me
May be software is still prepared for a upcoming SDM3085X  (joke)

In photo 2 you see the warming up of the HP34401a against the running SDM3065 for a 13h period. Voltage source is the 9 volts battery (block) in a double styropors box.
I was surprised how sensitive at this microVolts level the battery is for mechanical influence, sitting on my bench and do some soldering was enough for drifts. 

In photo 3 the same data normalized for the first 2 hours. I've done some mathematical voodoo in Excel to keep the noise levels of both devices visible.
It took 2 hours for the HP34401a to warm up and as we will later see there is a little movement during the next hours.
Plz be aware that the readings are 7.5 digit from both devices but it's clear that the SDM is a little bit noisier.

Last picture shows the measured difference of the (HP-SDM) at 9.5 volts starting at 120 min till 720min.
It's obvious that there is another "settlement" of the HP during the next 3 hours. 
As mentioned before the SDM is now for 20 days 30...40 Microvolts (0.03...0.04 mV) in the 20 V range lower than the HP (in 10 Volt range). There was no additional drift
during this "always on" period.

Looking into the specs of both devices and doing some rough calcs (assuming 10 volts reading):
SDM: In 20 V DV range the drift in 90 days can be 0.0030% of measured value + 0.004% of range = 1.1 mV
HP:   In 10 V DV range the drift in 1y (there is no number for 25 years ) can be 0.0050% of reading and 0.0030% of range = 0.85 mV

Assuming both devices are on the lower limit of this span the HP shows (for 9.5 volts) 9.499,15 volts and the SDM 9.498,9 volts. (250 uV difference)
The measured difference of about 30...40 uV is much better.

I'm pretty happy with this results but this is of course "amateurish", further measurement in other ranges and of linearity can be made.
But I have now enough confidence in the SDM and can use it for my projects and I'll avoid the "volt nuts" rabbit hole.
Next week I will do the measurement of the SDM's warming up against the running HP.

[tonyalbus]

Thank you very much for your tests, very interesting! 

[mawyatt]

@ Bad_Driver,

Nice work, thanks for the effort. Your results are similar to our SDM3065X, it's quite stable and seems to have a lower time constant to a stable reading than our KS34465A, although shows more "noise" in the readings.

Do you know if the LM399 is kept powered on when the SDM3065X is in the OFF state but plugged in to the mains?

Best,

[bdunham7]

Looking into the specs of both devices and doing some rough calcs (assuming 10 volts reading):
SDM: In 20 V DV range the drift in 90 days can be 0.0030% of measured value + 0.004% of range = 1.1 mV
HP:   In 10 V DV range the drift in 1y (there is no number for 25 years ) can be 0.0050% of reading and 0.0030% of range = 0.85 mV

Surely those specs are wrong?  The 34401A 1-year spec @10V is .0035% + .0005% or 400uV/40ppm.  I presume the Siglent is in the sam ballpark.

[Bad_Driver]

Yes, you are right, I was in a wrong row. My mistake, See attached for all specs in DC. HP above, SDM below.

Recalc:
HP in 10 volts range @ 10 V value: max. (1y) error: 0.4 mV
Siglent in 20 volts range @ 10 V: max. (90d): 0.38 mV  - both in the same ball park.

This night I rethought the noise observation. I can't claim so easy that the SDM is noisier.

I found out that the last transmitted (relevant) digit (#7) changes in 3 uV steps but from the HP in 1uV steps (in range 20 volts/10volts). You can clearly see this in the histogram of both (attached) and in the readings.

That means, that the SDM in the 20 volts (going up to 22 volts) range quantize about 7.3 Mio steps
and the HP about 10 Mio steps - do you think my assumption is right? Someone discovered what ADC is used in the SDM?

[mawyatt]

Just did a quick test between some other work. These were Cs and Rs in fixtures made to directly plug into the 4 banana jacks on the DMMs. The Cs are just 2 wire. The Tonghui TH2830 was using the TH26011CS test fixture and not compatible with the fixtures for the DMMs, so it couldn't perform a proper 4 wire measurement.

Edit: Added a couple leaded resistor measurements where the TH2830 could use the proper 4 wire measurement.

Results were TH2830   KS34465A   SDM3065X   HP34401A    AG34401A
                   9.94513   9.9544        9.9472         9.9361         9.9480
                   1.99780   2.0000        1.9970         1.9906         1.9958 


The SDM3065X did quite well on the R and C measurements compared to the KS34465A, HP34401A, AG34401A and the new TH2830.

Best,

[Neper]

Received my SDM3045X today and I like what I see. Just one thing I've noticed: the beeper is rather low on High, barely audible on Medium and totally inaudible on Low. Is this normal?

Ralf

[klausES]

Here I go less from a software error or driving error, but rather therefrom this is due to the basic volume of your buzzer.
The series variance in these piezo buzzers (similar also whether such Buzzer "good" or distorted sound) to the base volume is relatively large.
Whether you have caught too quiet in misfortune, you can only try it quickly to replace it through another to the test.

PS.
Whereby it is somewhat unhappy with too quiet buzzer, Siglent has the smallest stage apparently actually made quieter than before.
Have just tried it at my 3055 (always i use only the middle level) and the lowest level is a whole piece of quieter than with the older versions of the firmware.

[Ringmodulator]

Received my SDM3045X today and I like what I see. Just one thing I've noticed: the beeper is rather low on High, barely audible on Medium and totally inaudible on Low. Is this normal?

Ralf

Hi Ralf,

my SDM3045x is about 3/4 year old.
The beeper acts normally and is clearly audible on Low.
I did set mine to Midlle. High is to much for me, but may be beneficial in a loud enviroment.

Buzzers have often a protection sticker to cover the opening, which should be removed befor final assembly.
Maybe this has not be removed.

If the unit is under waranty, I would have it exchanged.

Chris

[Neper]

Buzzers have often a protection sticker to cover the opening, which should be removed befor final assembly.
Maybe this has not be removed.

Might well be. Unfortunately there is a warranty sticker, so I can't simply open it and have a look.

BTW, I've done the opposite in a number of cases: Too loud buzzers which cannot be turned down via the unit's settings can be tamed by sticking one of those little self-adhesive rubber feet on their opening. Quite efficient and easily reversible.

I've had a DMU whose buzzer was so loud and shrill that our cats would flee from the room each time I was using it.   

If the unit is under waranty, I would have it exchanged.

I'll call the dealer, tomorrow morning. So far, they've been very fast and friendly.

Ralf

[Ringmodulator]

New Firmware available for the SDM3045X:
V5.01.01.07R1

With calibration option in the menu, but undocumented.

https://siglentna.com/download/22253/

Chris

[Ringmodulator]

Hi,

any info an the calibration option in the new FW yet?

Chris

[TheDefpom]

I’m waiting for the ok for me to publish my video showing how to do it… hopefully Siglent officially release the procedure soon!

[flash2b]

Anyone tried the 1.01.01.25 for the SDM3055 already ? I am still running the 1.01.01.22R1 and was happy that the USB logging worked again, so I am a little scarred to update....

[klausES]

Try the version.
It is not a problem because you can change by the SDM (so on my 3055 in any case many times)
but at any time cross and across any firmware version to each other.

PS.
I have not tested everything yet (in some things I do not remember) but it's "finally" a lot has become better with the currently new firmware.

[Bad_Driver]

Unfortunately it took some weeks longer than expected to come back with my promised measurements of warming up SDM3065 against HP34401A.
After running the SDM3065X for more than 3 weeks (about 500 h as  suggested) I run the "warming up" of the HP34401A against the SDM - results above.
Than I shut down the SDM for 24+ hour, kept the HP running and repeated the measurements with an "soft start" of the SDM.

Surprisingly fast the SDM came into specs, it took less half an hour (compared to 3 hours of th HP). So I assumed that the internal reference is always on power
as discussed before, what would explain the 4 watts which the SDM in "Power off" takes. So I repeated the same measurement next day after another 24 hours
"hard power off" - not plugged into power line. But the measured curve is nearly the same! There is no visible difference! May be during the last 25 years the
specs of the LM399 have changed?

Attached you find again the warming up of the HP against the SDM and than the 2 measurements of the SDM warming up (blue cold start, red warm start).

For your convenience now both diagrams calculated and visualized with the help of HKJs fine piece of software: TestController.
All measurements done with 100 PLC, 10 GOhms, SDm in 20V range, HP in 10V range. Temp about 23 degrees celsius in my lab.

[2N3055]

LM399 is not the reason for this warm-up drift.
LM399 will be fully up to specs in less than 10 seconds  and pretty much perfect in 30 seconds
It is rest of meter reaching thermal equilibrium.
New meters have fan that speeds up the process by mixing air.
34401A is fan-less design.. It has to distribute temperature passively...

[Bad_Driver]

That makes sense!

[iMo]

LM399 is not the reason for this warm-up drift.
LM399 will be fully up to specs in less than 10 seconds  and pretty much perfect in 30 seconds
It is rest of meter reaching thermal equilibrium.
New meters have fan that speeds up the process by mixing air.
34401A is fan-less design.. It has to distribute temperature passively...

MY HP34401a (with additional thermal isolation and thermal sensor built in) warms-up 3 hours too.
PS: actually it did 5.5hours in my experiments, but normally I wait 3h
https://www.eevblog.com/forum/projects/thermostating-the-hp-34401a-meter/msg2614842/#msg2614842