Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[jan28]

Hello,

I'm trying to control my SDM4045X via SCPI. I've got most of it working but now want to know if the device is measuring, waiting for a trigger or 'idle' but can't figure out a way to do this via SCPI (on the display it is show in the upper left corner). I'm using a socket connection via ethernet and FW 5.01.01.09R2.

The SCPI manual briefly mentions the standard *STB?, *ESR?, *OPC commands but doesn't explain exactly what is supported or not.

I've tried:
- *OPC/*OPC? that doesn't seem to work for what I want to know.
- I'v also looked into some of the undocumented SCPI commands:  status:oper:cond:.... The standard operating register doesn't seem to get filled although the ENABLE/ENABLE? part to define the route the standard operating register to the status byte register seems to work.

Basically I would like to do the following:

Code: [Select]

STAT:OPER:COND?

This always seems to return 0. 

Any hints?

[Martin72]

Has been forwarded... 

Fast response:
https://siglentna.com/operating-tip/adding-customer-defined-temperature-sensors-dmm-sdm3000/

A pre-release version, so that you can get ahead for now.

[Mortymore]

Has been forwarded... 

Fast response:
https://siglentna.com/operating-tip/adding-customer-defined-temperature-sensors-dmm-sdm3000/

A pre-release version, so that you can get ahead for now.

To where can I send you a crate of beer? 

[Martin72]

Half of it should go to Mr. Rottach from Siglent Europe, for his very quick response.
I'm going to give this a try this weekend.

[tautech]

Has been forwarded... 

Fast response:
https://siglentna.com/operating-tip/adding-customer-defined-temperature-sensors-dmm-sdm3000/

A pre-release version, so that you can get ahead for now.

Thanks Martin, added to POI list.

[Mortymore]

In a quick attempt, manage to create a custom sensor, not for temperature but for a current clamp ICA32N that in the 100A range as an output of 10mV/A.

Readings are jumping more that desirable, neither are linear in the short range of current measured but that was not a concern. I just created a straight forward table with a dozen values to test if the thing works. I believe that it did, though I had to create the sensor in the internal memory, and when tried to store the sensor in the USB stick it allowed me to, but when tried to read the file I got an error and the meter freezed. I have to check that later, but even if it works only in the internal memory I'm OK with that.

PS: When choosing the UNIT, the option selected was "Custom Unit" and chose the letter "A" (since the intention is to read Amps)
PS2: The table was created with half a dozen values, later edited and more values were added, with no problem.

[RikV]

Martin72, thanks for this info. I wil certainly try it. Any idea how many values can be put into that conversion table? Since the behaviour of NTC (and others) is strictly non-linear, the more entrires the better the precision.
Suggestion to Siglent: why not allow for entry of a math equation to define a known sensor? (More difficult to implement in soft of course)??

[Johnny B Good]

Martin72, thanks for this info. I wil certainly try it. Any idea how many values can be put into that conversion table? Since the behaviour of NTC (and others) is strictly non-linear, the more entrires the better the precision.
Suggestion to Siglent: why not allow for entry of a math equation to define a known sensor? (More difficult to implement in soft of course)??

 I seem to recall a table size of 20 entries when I was trying to create a custom temperature sensor for one of my 'oddball' (ie unknown) thermistors recovered from various scrapped circuit boards. I'd initially assumed it would neatly interpolate the correct curve for any NTC thermistor from as few as just three entries only to be disappointed by the absence of any mathematical processing beyond a simple linear interpolation

 The thermistor manufacturers provide resistance versus temperature tables, commonly in 10 and 5 degree steps (and possibly even down to 2 or 1 degree steps if you ask nicely).  A twenty entries table using 5 degree steps covers a range of 100 degrees with pretty good accuracy for most purposes if the errors at the extreme ends of your chosen range aren't going to be a problem (typically true in most usage cases).

 I'm surprised Siglent didn't think to enhance the custom sensors with an NTC setup option using the Steinhart equation where you'd only need to load the reference temperature and corresponding reference resistance (eg, 10K at 25 degrees) and the thermistor's B coefficient (typically 3995).

 That option would allow you to experiment with the coefficients to more accurately characterise a randomly selected NTC thermistor such as the one I'm using to sense the heatspreader temperature attached to the base plate of an LPRO 101 rubidium oscillator which had measured 30K at 36 degrees and I'd simply plugged in a B value of 3995 as a starting point in the steinhart equation I was using in an Arduino sketch to stabilize the rubidium's base plate to 36.05 deg +/- 10mK or better.

 In this case I was using a series resistor of 30K to gain the best accuracy at my desired setpoint temperature. The benchmeter would be using a constant current source to measure the resistance more directly (no need to add a resistance calculation step as my Arduino sketch requires).

 From my own oversampling experiments (using floating point math, btw) to wring a few more bits of accuracy out of a nano's 10 bit adc (ranging from 64 right up to 1024 oversampling values), I rather doubt the floating point calculations would burn up more than a microsecond's worth of cpu cycles per measurement cycle.

 In view of the above, I'm rather surprised Siglent failed to add such a useful feature for so little cost in cpu cycles. Mind you, in view of the abysmal UI design of every model of SDG they've produced, perhaps I shouldn't be so surprised after all. 

[Mortymore]

After a successful input of a table directly in the internal memory of the SDM3055, I copied the table to an USB stick to get the structure needed of the .csv file

It goes something like this:

Code: [Select]

Name,./usr/usr/SAVE/ICA32N.sensor.csv,
Unit,A,
Model,DCV,
Sensor_name,ICA32N,
Data,0.000000,0.000000,
Data,0.001000,0.100000,
Data,0.002000,0.200000,
Data,0.003000,0.300000,
...
Built the table in Excel, with the data I wanted, dragging the cells, and later in the notepad (or whatever), replaced the ; for , and such.
In Excel you can apply any formula you need to convert the input data to the desired output.
With the new table in the USB stick, dumped it to internal memory of meter.

When the table has to many rows, it would be useful to be able to jump directly to the top or bottom of the table.
There's an empty space in the function menu that could be fitted with that functionality 

[Tjuurko]

Strange measurements in the "0.05PLC" mode

SDM3065X-SC

Mode: DCV
Range: 2V
Aperture: 0.05PLC
Auto Zero: Off
Input Z: 10G
Rel: Off

The signal -2.1978V -> +2.1898V (blue line) is supplied to the input and the measured voltage (red points) is recorded.
Only after ~40 measurements managed to get "+2.1898V"
It turns out that the real measuring speed is not "1000 Reading/s", but "25 Reading/s".

[Martin72]

Today a K-Type adapter arrived.
Looks good, fits perfect into the sockets of my SDM3065X 
Connect a suitable sensor and let the SDM measure the temp ( choosing "KIS90" in the menu), works fine.
BTW, the sensors are custom handmade in germany (sensorshop24)...

[Johnny B Good]

@Martin72

 I'd say the fit of the adapter is the least of your problems in this case. Unless your experience was hugely different to mine, I'm guessing you'll be using that dedicated handheld digital thermometer to measure temperature with K type thermocouple sensors rather than use the SDM3065X.

[tautech]

@Martin72

 I'd say the fit of the adapter is the least of your problems in this case. Unless your experience was hugely different to mine, I'm guessing you'll be using that dedicated handheld digital thermometer to measure temperature with K type thermocouple sensors rather than use the SDM3065X.

Maybe you missed or didn't fully understand this post: 
https://www.eevblog.com/forum/testgear/siglent-new-bench-dmm-sdm3055/msg5064094/#msg5064094

[Tjuurko]

Continuation of experiments from "0.05PLC":
Input signal ±2V (blue line)
Range: Auto (red dots)

[Kleinstein]

It looks like the SDM3065 uses some digital fitlering when in the fast modes. This is odd, as to a large part it makes the response slow, like 1 or 2 PLC.  Not sure if there is a way to turn of the filtering part - if not this is not a good feature.

It looks like the autoranging is however fast and after switching ranges there is a instant result - looks like not even missing 1 reading from going from 0.2 to 2 V range.

[Johnny B Good]

@Martin72

 I'd say the fit of the adapter is the least of your problems in this case. Unless your experience was hugely different to mine, I'm guessing you'll be using that dedicated handheld digital thermometer to measure temperature with K type thermocouple sensors rather than use the SDM3065X.

Maybe you missed or didn't fully understand this post: 
https://www.eevblog.com/forum/testgear/siglent-new-bench-dmm-sdm3055/msg5064094/#msg5064094

 I read that post so I guess it has to be the latter. However, my point had been the result of a needlessly fast update rate which appears to be 10PLC (5Hz refresh rate) with a very noisy sensor signal.

 All the digits to the right of the decimal point do a digital dance making it difficult to interpret the value to even just one decimal place. The remaining 3 digits are just a useless distraction. This is compounded by the less than stellar 'cold junction' temperature compensation (understandable considering the CAT I requirement). However, this may be a case of "RTFM" (at least as far as my dialing out the reference offset error is concerned) so when I can spare the time, I'll delve into the latest manual to better acquaint myself with the function(s) of the RefTemp and Rel buttons.

[EDIT] The manual was no help at all in this case.

 I gave up trying to obtain useful temperature readings with the KITS90 sensor option over a year ago and, right now, this is just a major distraction from my current MK III GPSDO project (basically a respin of the MK II but with a ZED9-T in place of the M8T module).

[Martin72]

Hi,

First and foremost, I posted it to express my delight that there is an adapter that also mechanically fits my pretty good K-type sensors.
But yes, the temperature display itself is very "nervous" - I had no time to look for it yesterday:
Is there no "Average" mode ?

[Johnny B Good]

Hi,

First and foremost, I posted it to express my delight that there is an adapter that also mechanically fits my pretty good K-type sensors.
But yes, the temperature display itself is very "nervous" - I had no time to look for it yesterday:
Is there no "Average" mode ?

 Not as far as I could discover (and believe me, I did search for such an option). As for that adapter, I have to agree with your assessment - I have one myself (among a collection other similar adapters). I'd been hoping to track one down that continued the chromel / alumel wire alloys into the pins to extend the "cold junction" all the way into the banana jacks.

 Needless to say, none of these adapters used chromel / alumel for the pins and, in retrospect, I guess there simply isn't any demand in this case since, unlike our Siglent bench meters with their 13 watt heat source to set up a 3 to 4 degree gradient with respect to ambient temperature, the hand held battery powered meters they're intended to be used with don't suffer such a thermal gradient issue due there being at most, only a few milli-Watts of heat energy to generate little more than a few mK temperature gradient.

 The cheap TM-902C K type thermocouple thermometers I bought only draw 0.6mA 0.35mA from a 3v battery. That's little more than 1mW to generate any thermal gradient between the chip and the socket that's worth correcting for. Obviously,"cold junction" compensation for ambient is applied, probably using an on chip sensor diode to keep instrument manufacturing costs to a minimum.

 I expect that your handheld digital thermometer is likely to have a similarly low energy consumption (on a par with a modern hand held multimeter) so there's little point in making such adapters using chromel and alumel pins. If I'd realised at the time that my quest for such a high spec adapter would be a futile one, I'd be about ten quid better off by buying just a single cheap Chinese adapter plug.

 I did manage to acquire a connector with alumel and chromel pins, a yellow K type plug used to terminate the K type thermocouple sensor wires. The pins were actually stamped with "AL" on the negative magnetic pin  and "CH" on the positive non-magnetic pin. Both pins on your adapter should prove to be magnetic assuming mine is completely identical to yours (appearance-wise they are as far as I can see).

 I seem to have purchased only the one which remains unused. I guess I must have realised that it would only ever be of any benefit if  I acquired a lab grade bench thermometer where a similar quality panel mounted socket was connected to a proper isothermal cold junction deep in the guts of the instrument via alumel/chromel connecting wires. The cheap TM-902C and similar handhelds wouldn't see any benefit from such a K type plug upgrade to any of my existing collection of thermocouples hence it remaining an unused spare.

===============================================================================================================
[EDIT 2023-09-22]

 It turns out that I'd actually bought a set of four of these K type plugs. I was looking through my collection of spare thermocouples today and found one with an upgraded plug already attached, swiftly followed by then discovering the other two plugs in a resealable polyethylene bag.

===============================================================================================================

 I landed up purchasing three of these cheap TM-902C meters. The first was a DoA for which I got a full refund and a spare thermocouple. Of the other two, one had the momentary on/off push button which defaulted to a ten minute power down time out, the other a slide switch which didn't have such an annoying time out feature. This was the one I chose to use, setting aside the push button one as a spare.

 A few weeks (or months) later, I experimented with the push button unit to see if there was any (undocumented) way to disable the rather annoying time out feature and, as luck would have it, it did indeed have such an option. Basically, after it has booted up and is reporting temperature, it's simply a matter of holding the button down until the clock icon disappears from the top left corner of the display (about ten seconds). Thereafter until you toggle it off or leave it on for the next three months to flatten the battery, it will continue showing temperature readings. However, it remains relegated as "my spare" since it under-reads the other by almost 2 deg C. The slide switch one tracks my other thermometers to within half a degree so remains my TC thermometer of choice.

[KungFuJosh]

I wanted a 2nd bench meter, so I found a decent deal on a used SDM3055. Naturally I had to do a quick comparison:


Results are less stable on the 3055. The X-E version has had a lot more use and break-in time. Does the 3055 also need a lot of break-in time for stability?

Thanks,
Josh

[KungFuJosh]

Still curious about accuracy comparing the 3055 to the 3055X-E, I ran some more tests. This time using the math function to show averages. A minimum of 2500 samples, some closer to 8k...

Tied to the same voltage source:
0.000V: within 2µV of each other
2.500V: within 9µV of each other
5.000V: within 13µV of each other
7.500V: within 16µV of each other
10.00V: within 20µV of each other

[tautech]

I wanted a 2nd bench meter, so I found a decent deal on a used SDM3055.

Results are less stable on the 3055. The X-E version has had a lot more use and break-in time. Does the 3055 also need a lot of break-in time for stability?

The more precise 3055 reference could do, not sure.
Q. Boot # for your new/old SDM3055 ?

[KungFuJosh]

The more precise 3055 reference could do, not sure.
Q. Boot # for your new/old SDM3055 ?

The used 3055 had 26 start ups when I got it. It has 30 now, compared to the 139 of the X-E. I also have left the X-E running for hours at a time.

I had both units on DCV with no leads plugged in. The X-E showed 00.0000 and the 5500 showed 00.0006. Both have the filter on. With significant samples, the X-E stayed the same, and the average dropped to 00.0005 for the 5500.

Then I switched to 200mV range, and the 3055 was closer to zero, but still less stable.

Do you think I need to have the 5500 calibrated? Will Siglent calibrate it for free since the calibration expired before the receipt date? Purchase date was March 21st 2023, and cal expired 2022-07-14. I think the cal expired for both meters before the purchase dates.

I'm assuming either way I should give it a lot of burn-in time first.

[Kleinstein]

Looking at the voltage with open input and presumably 10 M input resistance mode can give a hint on the input bias current (to gether with the actual zero reading with a low thermal EMF short).
The open terminal case is otherwise not that relevant.  A zero input both meter versions should give about the same stability and noise - no significant difference expected that far. Maybe the fan can help with a more stable thermal state.

A difference in the noise may be with a significant voltage, like 2 V at the input. Here the lower noise ref. in the better meter could be visible. Still some 6 µV_pp (at 2 V from the REF 5025 for the 0.1 to 10 Hz range) is not much and my be swamped with other noise sources.  The normal readings are usually lower BW, but also lower frequencies (more like 0.01 to 1 Hz) - so a direct comparison is not that easy.
The statistic function (e.g. observe over some 100 readings at the slow setting) may give more quantitative values for the noise.

[tautech]

The more precise 3055 reference could do, not sure.
Q. Boot # for your new/old SDM3055 ?

The used 3055 had 26 start ups when I got it.

Ah, almost like new then. 

It has 30 now, compared to the 139 of the X-E. I also have left the X-E running for hours at a time.
I had both units on DCV with no leads plugged in. The X-E showed 00.0000 and the 55 showed 00.0006. Both have the filter on. With significant samples, the X-E stayed the same, and the average dropped to 00.0005 for the 5500.
Then I switched to 200mV range, and the 3055 was closer to zero, but still less stable.
Do you think I need to have the 55 calibrated?

With that low boot count, no not yet, give it more run time.

I'm assuming either way I should give it a lot of burn-in time first.

[KungFuJosh]

Question:

What do you consider a standard calibration period? 1 Year? Something else?

If you purchase a DMM, and the calibration is over a year old, what would you expect of the manufacturer? Or let's say, a random number like...11 months old compared to the purchase date? What then?