Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[KungFuJosh]

Following up from the other thread talking about SDM3065X stability:

After 1 hour warm up time with both the meter and the reference on.

10.00460 (claimed by reference seller - not certified)
10.0039? on SDM3065X



I will post again after some hours.

ETA: TestController log rate is set at 0.1s.

[TheDefpom]

It is FAIRLY settled after an hour, it is still warming up for about 4 hours, after that although there is still some movement but it is small.

[KungFuJosh]

Okay then. I've passed the 4 hour mark now. So I'll post the info now, then reset the data and let it run for 4 more hours and see if there's a difference.



Note that there were some minor changes in temp and RH.

[bdunham7]

ETA: If we pretend to believe the stated value of 10.0046, then the bottom number to stay within would be 10.000197976. Seems safe. 🤷

10.0046 would be displayed as 10.00460 on a 6.5-digit DMM and is 4.6mV or 460ppm over the nominal 10V.  If your combined tolerance is specified as 48ppm of 10V or 480µV (40ppm  * 10V + 4ppm * 20V) then your lower reading limit would be 10.00460 - .000480 or 10.00412. 

[KungFuJosh]

ETA: If we pretend to believe the stated value of 10.0046, then the bottom number to stay within would be 10.000197976. Seems safe. 🤷

10.0046 would be displayed as 10.00460 on a 6.5-digit DMM and is 4.6mV or 460ppm over the nominal 10V.  If your combined tolerance is specified as 48ppm of 10V or 480µV (40ppm  * 10V + 4ppm * 20V) then your lower reading limit would be 10.00460 - .000480 or 10.00412.

Apparently I'm still confused. If the resulting accuracy for this range is:
0.0040 + 0.0004 = 10 x (0.0040 / 100) + 10 x (0.0004 / 100) = +/-0.00044

So you're saying that the specified .00044 is an offset number instead of a percentage number? Damn it. Okay, I guess that makes more sense than another percentage.

Thank you. Learning is fun? lol

I guess there's no point in worrying about the 10.00460 since it's a bullshit number. Still pretty close anyway.

[Martin72]

You can imagine why I want to have my reference (yes, it's not forgotten) calibrated externally...
And even then it wouldn't be “perfect”.
In the calibration protocol of my SDM3065X, the measurements were done like this:
Fluke voltage reference set to the range, measured the actual current value with a Keysight 8.5 digit meter, then measured with the SDM3065X and its displayed value is compared to that from the Keysight.
That would be the way.
Then the reference value would be almost irrelevant, as you measure with a much more accurate meter.

[KungFuJosh]

Weeeeeeellll, I think this test is pointless. My ref isn't stable enough, nor is the environment. PP was solid at 3ppm for the first two hours. Then after I sat in the room for a bit, it jumped to 3.6ppm. Then when the air conditioner shut off, it jumped again up to 4.2ppm.

None of that is bad of course, but for a tighter tolerance I'd need an enclosed ref and better environment control.

You can see the jump in the chart after the AC shut off.

[mawyatt]

Okay then. I've passed the 4 hour mark now. So I'll post the info now, then reset the data and let it run for 4 more hours and see if there's a difference.



Note that there were some minor changes in temp and RH.

Think you are showing the same long term plot twice, they both look identical!!

Best,

[KungFuJosh]

Think you are showing the same long term plot twice, they both look identical!!

Nuh uh! 😉

The air conditioner came back on, and the voltage went back down (or at least started to before I shut it off).

[mawyatt]

Just noted that in all the SDM3065X plots this is showing SD (assume Standard Deviation) as 0.000000V. Don't think the SDM3065X, nor the Reference are quite that good noise wise!

Best,

[KungFuJosh]

Just noted that in all the SDM3065X plots this is showing SD (assume Standard Deviation) as 0.000000V. Don't think the SDM3065X, nor the Reference are quite that good noise wise!

I assumed that was a bug when I noticed it, but also wasn't going to stop a 4 hour test to upgrade TC. 😉

Nope. Another PEBKAC error. Apparently I needed to go to the Current Values tab, right-click and increase the number of digits and/or choose SI for SD to work.

[KungFuJosh]

Tomorrow I'll test both from cold start. Or should I warm up the ref first, to separate that? It makes sense to me to warm up the ref first remove the ref's drift from the test.

Ideally you'd do it both ways--with the ref warmed up and meter cold then vice versa.  But try it the first way first.

Looks like 30 minutes for mine if uV resolution is needed. Immediately useable if only needing mV.



I assume if I need super accuracy I should stick with what Scott said and warmup for 4 hours. Or get solar panels and never turn off the meter. 😉

[KungFuJosh]

Looks like the ref is pretty good after 5 minutes or so, depending on the environmental stability.




I also decided to point a hot air wand at it. 100C at 50% power. Alternated a couple times between 3 to 6 inch distance.

[bdunham7]

Looks like 30 minutes for mine if uV resolution is needed. Immediately useable if only needing mV.

I assume if I need super accuracy I should stick with what Scott said and warmup for 4 hours. Or get solar panels and never turn off the meter. 😉

Actually those results are pretty good, in fact better than the typical 34401A. I wouldn't worry about warmup times now that you've characterized it.  Anything after the 5-minute mark is a very small proportion of the specified tolerance so TC and long-term drift will dominate any measurement uncertainty.  IMO for 99% of uses you can use it a soon as it boots and for the rare occasion where you want maximal stability 5-10 minutes is plenty.  If you are doing single-ppm relative measurements or calibrating the meter, then you might go for a 2 hour warmup.

Now you need to independently verify its absolute accuracy!   

[KungFuJosh]

Now you need to independently verify its absolute accuracy!   

It's still in cal, and I don't have a better reference than the one I used. It also agrees with my other meters. Good enough, or you got some fun suggestions? 😉

[KungFuJosh]

I'm a schmuck and decided to spend the money on a better ref that I don't need. I bought the VREF10-001 from Doug. He clearly does good work.

I captured the 1st, 2nd, and 3rd hours after 4 hours of warmup. The first hour was probably the most accurate, as I was back and forth working on my bench in the following two hours. Even with my interference (including soldering 1 foot away), the stability was excellent.

It was also much less influenced by environmental change as you can see by the minor shift related to the temperature change. With the previous (cheapo) ref, that would have been much more significant. The bigger temperature shift after about 1.5 hours was when I closed my office door and the AC became more effective.

[Martin72]

Here a link to the reference:

https://voltagestandard.com/shop/ols/products/vref10-001-r9

[KungFuJosh]

I was impressed with the results so I emailed Doug to thank him. In his reply he said "It is interesting that the Avg reading was within 1uV during all 3 tests;  the uV noise on the Reference output is low frequency noise inherent to the LT1021 Reference chip- it looks like your DVM does an excellent job of averaging."

Does the SDM3065X do any averaging outside of the statistics box?

[tautech]

I was impressed with the results so I emailed Doug to thank him. In his reply he said "It is interesting that the Avg reading was within 1uV during all 3 tests;  the uV noise on the Reference output is low frequency noise inherent to the LT1021 Reference chip- it looks like your DVM does an excellent job of averaging."

Does the SDM3065X do any averaging outside of the statistics box?

What NPLC settings are engaged ?
Any Math engaged ?

[KungFuJosh]

Does the SDM3065X do any averaging outside of the statistics box?

What NPLC settings are engaged ?
Any Math engaged ?

10PLC and the statistics box was open.

[tautech]

Does the SDM3065X do any averaging outside of the statistics box?

What NPLC settings are engaged ?
Any Math engaged ?

10PLC and the statistics box was open.

Yeah you said.
No Min / Max / Average as well ?

[KungFuJosh]

No Min / Max / Average as well ?

That's in the statistics box. I haven't setup anything else. It's always like this old screenshot:

[vulubalulu]

I was scouting the internet for a benchtop dmm for instantaneous autoranging for voltage and resistance. I saw a video of a sdm3045X being really quick and it beeps shortly when it detects voltage, i like that feature too. But I feel like it's overkill for my application. I just repair non-audio electronics as a hobby. At most I measure AC that comes out of the socket and usually it's DC voltage ranging from 1v to 50v. I don't really need so many counts either, i just want it fast. the 45x looks really neat but at almost 450€ the price point is high. Are there alternatives you can recommend. Please advise.

[tautech]

I was scouting the internet for a benchtop dmm for instantaneous autoranging for voltage and resistance. I saw a video of a sdm3045X being really quick and it beeps shortly when it detects voltage, i like that feature too. But I feel like it's overkill for my application. I just repair non-audio electronics as a hobby. At most I measure AC that comes out of the socket and usually it's DC voltage ranging from 1v to 50v. I don't really need so many counts either, i just want it fast. the 45x looks really neat but at almost 450€ the price point is high. Are there alternatives you can recommend. Please advise.

369 Euro plus whatever taxes is the list price.
https://www.siglenteu.com/digital-multimeters/sdm3045x-digital-multimeter/

[vulubalulu]

the 369 Euros are without tax. it comes down to 450