Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[tautech]

but it is too deep to fit my lab table .. oh them problems..

Comments in this post and some following may be of help to you:
https://www.eevblog.com/forum/testgear/siglent-sdl1000xsdl1000x-e-electronic-load/msg2511993/#msg2511993

[Kleinstein]

The SDM3068 uses a LM399 ref., but for the ADC chip the voltage needs to be divided down. At least for the short time scale the divider drift may be similar to the reference dirft. So it is no for sure that the temperature drift of the 3065 is actually much better than the 3055.
For the long term dirft and possible humidity (e.g. seasonal) effects the metal can reference can be a plus.  I don't know if the SMD3068 does an internal direct check of the 7 V refrence.

A large step up for the 3068 is having high impedance also for the 20 V range and not just the 2 V range.
It has a few nice features, but may not be the best choice if ultimate scale factor stability is needed.

[Bad_Driver]

Do we know what ADC is used in the SDM3065X?
The same as in the lower range DMMs?

[oz2cpu]

Do we know what ADC is used in the SDM3065X?
dave only did teardown of SDM3055 ?
https://youtu.be/6nqSFYVKnP4

The DefPom about SDM3065
https://youtu.be/k6HE8qV3CA4
did Siglent address the issues he mention in this video ? firmware updates ?
this video is from 2017, it is really good, but it dont reveal any inside hardware details

[Kleinstein]

There are pictures around of the SDM3068 interals. They show and AD7175 ADC, so a really good one. That is low noise and low INL for an SD ADC chip, especially if they don't use the internal buffers.
The SDM3045 and 3055 use a lower grade  AD7190 , still quite good for 5.5 digits and way overkill for 4.5 digits.
A point that is somewhat limiting for the SDM3055 is that they have the 2 V range as the prime range and the 20 V already seem to go though the input divider and thus 10 M input impedance only and extra noise and drift from the divider.
The 3068 has the 20 V range without the 10 M divider and only an extra, lower resistance divider after an initial buffer.

[tautech]

Do we know what ADC is used in the SDM3065X?
dave only did teardown of SDM3055 ?
https://youtu.be/6nqSFYVKnP4

The DefPom about SDM3065
https://youtu.be/k6HE8qV3CA4
did Siglent address the issues he mention in this video ? firmware updates ?
this video is from 2017, it is really good, but it dont reveal any inside hardware details

See them here:
https://int.siglent.com/download/firmwares/?ProId=24

Link to SDM3065X PCB pics is in the OP POI list.

[Bad_Driver]

Hi Tautech,

I used my VPN to connect direct to Shenzen for the international Siglent web page but what and where is „OP POI“???

I could not find any pictures.

[tautech]

OP = opening post for this thread.
POI = Points of interest list that contains link to post with pics. 

China website:
https://www.siglent.com/?l=1

This is accessible from the link in EU and US websites but in some browsers it defaults to the English version:
https://int.siglent.com/

[Bad_Driver]

Thanks, I got it!

[Bad_Driver]

OZ2CPU found some strange looking noise with the Siglent SPD3033X power supply during his investigations.
We tried to analyze it a little bit more. For this we both made some recordings with the SDM3055X (oz2cpu) and I (SDM3065X) for further FFT with EXCEL.
You find the results in the SPD3033X thread.
https://www.eevblog.com/forum/testgear/how-to-calibrate-siglent-spd1000x-spd3303x-spd3303x-e-series-power-supplies/msg3923762/#msg3923762

During this tests I discovered some strange behavior of the SDM3065X (the SDM3055X seems to be better) which I want to point out.

Problem:
If you record continuous measurements to USB with SDM3065X, the time interval between measurements is not constant, there are big deviations.
I tried several series with different USB-drives and different settings. As you can see in the attached photos, we find a distribution pattern. (y-axis: time interval between single measurements in seconds)

But for low frequency FFT we need a nearly constant interval or with other words a constant acquisition frequency.

Surprisingly the SDM3065X behaves during very fast recording (0.05 PLC, acquisition frequency up to 1 kHz) best.
It's not a problem of the USB transfer speed, one data set is 32 byte, at 1 kHz acquisition frequency we need only 32 kByte/s.

The problem with the not constant triggering was pointed out here long ago.
https://www.eevblog.com/forum/testgear/siglent-new-bench-dmm-sdm3055/msg3466342/#msg3466342
I tried external triggering as well but found no improvement.

Attached you find the distribution pattern of different recording (4096 measurements - limit for Excel FFT) of the 3055X and the 3065X.

btw: in one of the first Siglent Operating Tips from March 2018 it was shown that there is no time stamp with recordings to the internal storage
https://siglentna.com/operating-tip/sample-and-store-sdm3000/

Now we have 2022! Siglent - why?

My wish/expectations to Siglent for the next firmware update:

#1 time stamp for internal recordings
#2 (nearly) constant acquisition/internal triggering frequencies, as with the 3055X, in best case controlled by user

As mentioned this would enable some interesting possibilities (low frequency noise measurements etc.)

Thanks to oz2cpu for his support.

Tautech, I will be very grateful if you can direct this to someone who is in charge of this product. Thanks.
 

[Kleinstein]

If the meter has no time stamping, how are the times measured ?
In a different thread they used an external triangle / ramp signal to see the times when the input is actually read.

[tautech]

If the meter has no time stamping, how are the times measured ?

PLC

[Bad_Driver]

USB recording has time stamp, internal recording not.
Sounds strange, but it works so..

I tested the time stamping with a low frequent square wave, it seems to be ok.

[oz2cpu]

while we are at it, suggestion great and usefull features,
for SDM3055:  add user setting of variable power line avarage, not just fast, med, slow as it is today,
but 1-1000 mains setable periodes. this will make it possible to get a lot better and more stable readings,
now we do this in external data recording software, but when comparing readings with different equipment, like keysight, raw data looks very different,
due to different filter settings..

[2N3055]

while we are at it, suggestion great and usefull features,
for SDM3055:  add user setting of variable power line avarage, not just fast, med, slow as it is today,
but 1-1000 mains setable periodes. this will make it possible to get a lot better and more stable readings,
now we do this in external data recording software, but when comparing readings with different equipment, like keysight, raw data looks very different,
due to different filter settings..

While being old, 34401A is a 6.5 digit meter running on completely different principle. It uses discrete converter that has full control of measurement cycles and full freedom to synchronize it to either mains frequency (for 50/60Hz rejection), to exact timing and level triggering. It performs true integration for a period specified. 

SDM3055 is a 5.5 digit meter that uses off the shelf ADC (AD7190  I believe). With that option, you need to work the way ADC wants. It has internal filtering and averaging, mains interference rejection etc. built in.
It will have sweet spots in data rates where it will have best performance. For emulating longer "integration" times, you basically take 10 or 20 shorter acquisition and average them etc. etc. ....

Keysight 34450A (their 5.5 digit meter) also have only fast, medium and slow mode...

[oz2cpu]

we dont know exactly how the cpu reads the ADC ?
how it is synced to mains ? is it the sync pin ? filters inside the ADC ?
or done in software ?
anyways, i perform a lot of side by side tests at the moment, using Test Controller and SDM3055 parallel voltage input to Keysight 34465A,
and the signals are very different
specially the 34465A can clearly do magic when the user configurable PPM filter is adjusted,
i just say : i would be super happy for a similar feature, this will give me 2 more useable digits on the data it sends out,
no visible difference on the display, since it only show much less

By the way ; there was a trick ? to make the SDM3055 show more digits on its display ?
it will only make sense to apply such hack, if a more slow filter can be enabled too

[RikV]

Is there another way to connect SDM3065X to TestController than TCPIP? USBTMC is not supported?

[HKJ]

Is there another way to connect SDM3065X to TestController than TCPIP? USBTMC is not supported?

Only TCPIP, but you can use either Socket or LXI (Click on the value in the Type column to select).
LXI do not need the IP address, it can poll the network for all LXI devices and automatically find it.

[2N3055]

we dont know exactly how the cpu reads the ADC ?
how it is synced to mains ? is it the sync pin ? filters inside the ADC ?
or done in software ?
anyways, i perform a lot of side by side tests at the moment, using Test Controller and SDM3055 parallel voltage input to Keysight 34465A,
and the signals are very different
specially the 34465A can clearly do magic when the user configurable PPM filter is adjusted,
i just say : i would be super happy for a similar feature, this will give me 2 more useable digits on the data it sends out,
no visible difference on the display, since it only show much less

By the way ; there was a trick ? to make the SDM3055 show more digits on its display ?
it will only make sense to apply such hack, if a more slow filter can be enabled too

It is not synced to mains. Rejection is part of ADC internal algorithm.

[Kleinstein]

we dont know exactly how the cpu reads the ADC ?
how it is synced to mains ? is it the sync pin ? filters inside the ADC ?
or done in software ?
anyways, i perform a lot of side by side tests at the moment, using Test Controller and SDM3055 parallel voltage input to Keysight 34465A,
and the signals are very different
specially the 34465A can clearly do magic when the user configurable PPM filter is adjusted,
i just say : i would be super happy for a similar feature, this will give me 2 more useable digits on the data it sends out,
no visible difference on the display, since it only show much less

By the way ; there was a trick ? to make the SDM3055 show more digits on its display ?
it will only make sense to apply such hack, if a more slow filter can be enabled too

The SD ADCs inherently have digital filtering build in. In some cases this can (not with the AD7190) be quite close to a fixed integration time, but often there are other filters, like the sinc³ filter, as 3 times the filtering like integration in series. This gives better suppression of the 50/60Hz and thus usually no sync to the mains phase needed and available. It comes at the price of longer settling, So the consecutive readings are no longer fully independent and there is some settling (like 3 conversions) after a step. So 5 readings per second on the SDM3055 is different from 5 readings per second on an 34401, more like comparable to 2.5  readings per second (10 PLC AZ).  The ADC is inherently different and the comparisone is not easy.

Chances are the slow, med, fast settings are just 3 filter / speed options available from the AD7190.  There are more internal options (including higher speed) and the µC could also do some averaging in software to get closer to something like 10 PLC classic integration from averaging 10 consecutive readings at 50 SPS. Currently this can be done on the PC.
More averaging could still be useful when the signal source is noisy. The meter itself and especially the ADC should be OK with stable 5.5 digits already for the 50 SPS mode.

More resolution on the screen may be available from the math function, as the average values with manual reset.

With the Sigilent firmware we could be happy with bug fixes in reasonable time. I would not have too much hope for much extra functionality. This would be more like a thing with the 3048 - that has HW that is capable of more. Not sure why they keep separate FW and not switch over to a SMD3055E with essentially the same FW and only lower specs, higher drift. It would be one FW version less to maintain, and I doubt there would be much complaint's from the users.

[purpose]

but it is too deep to fit my lab table .. oh them problems..

There is a French girl who's been working on her 3055 for a coupe of years (cough), so hang on for a while longer and you might have a 7.5 digit 3055 in a nice small package.

https://www.eevblog.com/forum/testgear/hacking-the-siglent-sdm3055/

[rfclown]

Does anyone know how to programmatically put the DMM back to LOCAL after making a remote reading? I'm using LAN right now. I searched the manuals and Googled to no avail. We just got some SDM3045Xs at work. When automating measurements I like my instrument displays to keeps updating when I'm not making readings, which for some instruments requires sending a command to go LOCAL after making a measurement. Only way I've seen to do this so far is by sending a *RST (which resets everything).

[oz2cpu]

my 3055 still reads anything the remote connection does, on its own display
dont understand your question

[rfclown]

my 3055 still reads anything the remote connection does, on its own display
dont understand your question

If there is 2V at the input, and I send the command "READ?", the display then freezes showing 2V. If the input voltage changes, the display won't change until I send "READ?" again. I don't want to have my program continually sending reads just to make the display update. I want to programmatically get the instrument back in LOCAL mode so that it changes automatically. Many instruments either don't freeze the display on a read, or allow you to send a command like LOCAL. It might be that I need to use a different command besides "READ?" if there is another one that doesn't freeze the display.

[Algoma]

Sound like the trigger is going to a stopped state after the single read command?