Siglent Bench DMMs SDM3065X, SDM3055 and SDM3045X

[tomud]

And I will say that some people are looking for problems where there are none.

I use the SDM3065X as a DMM for everyday work, bought because it is powered from the mains, which eliminates the problem of replacing the battery. Sometimes you need to check something quickly, whether it's capacitance or something else, and that's where this multimeter comes in handy. If I need more accurate measurements, I use better quality equipment.

As for the frequency measurement, it's such a quick test - photo below... MF2414B (Option 03 - oscillator 5×10^-10) turned on now for this test, so it's not warmed up and stabilized... The SDM3065X, according to the specification, measures the frequency in the range of 3 Hz ~ 1 MHz

I will add that measuring AC voltage and frequency at the same time is sometimes useful 

[KungFuJosh]

Aaaaaaaaaand then I remembered I own an SDG2122X which has a "High precision Frequency Counter" - and I'm guessing that's at least as good as a "cheap frequency counter." 🤔

I think I can trust my DMM's freq counter for my needs. Though now that I brought out the SDG, whatever. 🤣

[tomud]

and I'm guessing that's at least as good as a "cheap frequency counter." 🤔

This may sadden Siglent SDM owners - pictured Aneng A3008 for $20 

Sorry for the poor picture quality but this is a screenshot from a video I was taking...

[KungFuJosh]

and I'm guessing that's at least as good as a "cheap frequency counter." 🤔

This may sadden Siglent SDM owners - pictured Aneng A3008 for $20 

Why? It's less accurate than your own Siglent SDM photo above...

[Martin72]

Aaaaaaaaaand then I remembered I own an SDG2122X which has a "High precision Frequency Counter"

See...
And the 200Mhz could last for most of the purposes at home.

[alm]

Is there something stopping them from putting in a frequency counter that's at least as good as a "cheap frequency counter"?

Probably that few people would be willing to pay extra for a better time base in what would be a pretty limited frequency counter in terms of frequency anyway. Also, the input circuit of DMMs is generally aimed towards low noise rather than frequency response.

No question on the capacitance being better in an LCR meter, but it's still a nice feature to have, and the accuracy is fine for general use.

My problem with DMMs isn't so much the accuracy of their capacity measurement, but the fact that it's a DC measurement that is very easily fooled by non-ideal caps, rather than the AC measurement that can measure both resistive and reactive components (e.g. ESR and capacitance). Fixing this would be expensive because it would require an entirely new source and sense circuit. That for me limits the use to checking if I have the right chip capacitor.

Both boil down to that it's hard to put a good screwdriver in a hammer because the requirements for both don't overlap much. For example a hammer needs to be heavy far away from the grip, while the weight of a screwdriver should be close to the grip.n So then you end up with a screwdriver that you awkwardly need to hold by the head of the hammer.

[KungFuJosh]

That makes sense, but as far as your analogy goes...

Hammer drills and impact drivers are awesome. 😉😉

[tomud]

and I'm guessing that's at least as good as a "cheap frequency counter." 🤔

This may sadden Siglent SDM owners - pictured Aneng A3008 for $20 

Why? It's less accurate than your own Siglent SDM photo above...

Because it works up to 26MHz and not just 1MHz like SDM    It was just a joke of course 

My problem with DMMs isn't so much the accuracy of their capacity measurement, but the fact that it's a DC measurement that is very easily fooled by non-ideal caps, rather than the AC measurement that can measure both resistive and reactive components (e.g. ESR and capacitance). Fixing this would be expensive because it would require an entirely new source and sense circuit. That for me limits the use to checking if I have the right chip capacitor.

Yes, that's right, only from my experience in amateur use to determine during repairs whether an SMD capacitor is functional or not, such a simple measurement in a multimeter, whether it's Brymen or Siglent, is 99% sufficient. And because I have this multimeter on all the time, it's convenient, and if needed, I can put a dedicated LCR meter on the shelf and use it for accurate measurement. However, as I mention, I don't have to waste time on it because in most cases such a simple test is enough. The same applies to frequency measurement, in many cases even such a modest measurement is enough and I don't have to run a dedicated frequency meter to check something.

[delvo]

I tried comparing the frequency measurement of my SDM3065X against my SDG2042X (which thinks its a SDG2122X). Im testing at 10KHz which is is where it should be most accurate. It has an initial accuracy of +-1ppm (and it is still in Factory calibration for that). Using its own frequency counter obviously shows bang on 10KHz.
My SDM3065X shows 9.999975KHz. Thats 2.5ppm off. With the 1ppm of the function generator that is ~3.5ppm which is within the 24hour spec (5ppm). Of course that is just 1 sample but not bad.
I also compared readings using my DE-5000 as my "function generator" and im at most 4ppm off. And thats at lower frequencies where the spec is 20ppm @ 24 hours.
My 3065x calibration is just over 1 year old.

[Bad_Driver]

I have never wanted to use the SDM as a frequency counter, there are much better alternatives. To give the new owners confidence or interested people a little help for the decision, attached is a photo of my 25 year old HP 34401A and my two year old SDM3065. Both devices were allowed to warm up for about 90 minutes, the HP device needs a bit longer since it doesn't have a fan.

[tomud]

I have never wanted to use the SDM as a frequency counter, there are much better alternatives.

It all depends on the application we need. For measuring the frequency of the voltage when building an AC inverter, this function is better than using a typical frequency meter for RF applications. To tell the truth, I wouldn't want to mess around and risk connecting the output of several hundred volts from the inverter to my Anritsu MF2414B that works up to 40GHz. Especially since using this frequency meter for such applications doesn't make much sense...

[Martin72]

Today the first back to work after the holiday, I also tested the SDM3045X as well as a Keysight 34450A with the DMMCheckPlus.

FYI..

[Performa01]

Today the first back to work after the holiday, I also tested the SDM3045X as well as a Keysight 34450A with the DMMCheckPlus.

Sorry, but these numbers don't look very plausible to me.

Looking just at the 5.0000 reference, we can see it is only specified to 200 ppm accuracy: 0.0001/5 = 0.00002 = 200 ppm. For highly accurate DMMs like the Keysight 34450A and also Siglent SDM3065X, this is totally inadequate. Even when taking the worst case error into account (which we wouldn't actually see on a freshly calibrated meter), the reading of an SDM3065X must be in the window of 4,999745 V to 5,000255 V. Your table says 4.9970 V, which would be far outside its specification. The Keysight appears to be better, but still outside its tolerances.

Look at the attached picture what it looks like if an old, abused meter, that has never been calibrated for at least one decade (bottom left) is connected to a freshly calibrated calibrator and compared to a newly acquired instrument from eBay (top right), when both meters are serious lab instruments. The left one is less accurate than the SDM3065X according to its specs, whereas the right one beats most average 6.5 digit meters, despite having only 5.5 digits.

From the discrepancy of the more serious meters in your test (SDM3065X and KS34450A) we can tell that it's neither the reference nor the meters which are at fault. It quite obviously is a systematic error in the measurement, like random cables and plugs, without paying attention to temperature gradients and thermal voltages. It takes a bit of care and effort to make precise measurements down to ppm accuracy.

[Martin72]

It takes a bit of care and effort to make precise measurements down to ppm accuracy.

In other words, I was sloppy.

The test leads were different (Probe Master at home, Fluke TL at work), but they had the same length - but this should not play a major role in a voltage measurement, nor in a (constant) current measurement.
If it does, I would have to know from the manufacturer of the reference, which testleads he had used for calibration.
The ambient temperature may have been different, 1...2 Kelvin difference.
I'll repeat it again, with all meters in the same place.

[Performa01]

The most important thing is: you cannot make precision measurements with handheld probes. Even if the contact materials were superb quality, the temperature gradient produced by holding the probes in your hands will introduce massive errors.

Look at my photo: for precision measurements, I've used fixed wiring with quality banana plugs, shielded cables connected to a guard (most important for low current and high resistance measurements) and not touching anything immediately before and during the process.

[Bad_Driver]

Martin, as mentioned much earlier in this thread I kept my new SDM3065 running for about 500 hours to settle down before I started some comparisons with my HP 34401A.

My shown measurement was done with a Kelvin clamp on a batterie and two leads went to the HP and the other two to the SDM.

[slavoy]

Is there any way to check the runtime of SDM3065X in hours not in on/off cycles?

[Martin72]

It would be a nice thing to know the initial values.
From the "more serious" Keysight I know them, because they were in the form of a protocol with the delivery, since then it has been calibrated regularly(so you can see where the values drift over time).
From my serious Siglent, I only have a sheet of paper....
I will write to them, the serial number is individual, maybe there is a database with values.

[delvo]

I found another firmware bug. If you enable the dBm display it correctly switches the displayed unit to dBm. But if you then press DCV/ACV it still shows the dBm value, but it says the unit is V. Which leads to weird (and very wrong) displays of negative volts AC.
@tautech could you maybe pass this along too?

[tautech]

I found another firmware bug. If you enable the dBm display it correctly switches the displayed unit to dBm. But if you then press DCV/ACV it still shows the dBm value, but it says the unit is V. Which leads to weird (and very wrong) displays of negative volts AC.
@tautech could you maybe pass this along too?

Please confirm model and FW version.

[bateau020]

Looks like the post from devo is from a 3065, but I can reproduce on a 3055 that is on 1.01.01.27R2

[Performa01]

It would be a nice thing to know the initial values.
From the "more serious" Keysight I know them, because they were in the form of a protocol with the delivery, since then it has been calibrated regularly(so you can see where the values drift over time).
From my serious Siglent, I only have a sheet of paper....
I will write to them, the serial number is individual, maybe there is a database with values.

If you have your gear calibrated by some independent cal lab, you'll have to pay more for a calibration certificate with data than one without. Considering the price of a comparable Keysight, it appears quite appropriate if they provide a calibration report with data.

Even with data, you don't gain as much as you might expect. The data are just a snapshot of the instrument under certain test conditions in a certain environment at a certain temperature, taken with some references that have a certain tolerance which should be guaranteed by a traceable calibration.

By contrast, you use the instrument in a different setup, under different conditions, at a different temperature and some time has passed since the last calibration. I can tell from experience, that you cannot ignore this and especially a brand new instrument may drift significantly (yet staying within its one year specification) during the first months of duty, so the cal data would not be terribly useful for introducing "correction factors" anyway.

This still describes the ideal case, when the calibration is done carefully by well trained and experienced personell. While this should be the norm, unfortunately this isn't always the case.

In short, if an instrument has a certain specified tolerance, then you can have some confidence that the measurement results will be within this tolerance window - if you use an appropriate test setup, that is. Your own table above demonstrates how even fairly accurate instruments can deliver just house numbers under certain conditions. A calibration certificate with data might tell you that the meter was off by some 5 ppm in the 20 V range when tested in the cal lab. What would this info have helped in this very situation, where the reading is off by several hundred ppm?

[delvo]

Its a 3065X with firmware 3.01.01.12R1

[Martin72]

@Performa01:
I know this metrology stuff.
If I argued in front of our Q auditors like you did, they would rip the certificate off the wall and we would go broke.
The fact that I was able to establish Siglent in our company at all is solely due to the fact that I am responsible for the measuring equipment and not others.
Otherwise the stuff wouldn't have been ordered in the first place.
Now private is private and not commercial, you could say.
In the case of cheap products, I see it the same way, no question.
But 800€ is no longer cheap.
We both have different views on this and that's not a bad thing, you can't always agree.
Martin

[Kleinstein]

It would be a nice thing to know the initial values.
From the "more serious" Keysight I know them, because they were in the form of a protocol with the delivery, since then it has been calibrated regularly(so you can see where the values drift over time).
From my serious Siglent, I only have a sheet of paper....
I will write to them, the serial number is individual, maybe there is a database with values.

Chances are the initial values are relatively close to ideal. With instruments that don't sit a long time on the shelf there may be just the initial adjustment and than a check to essentially the same reference.
At least the LM399 reference is not expected to show much drift in the time it is powered off. So actually getting readings for the initial calibration may not be very useful: impressive looking numbers, but valid only for the first few hours. The initial drift can include some extra faster contributions, so even for a cal history the first few 100 hours and thus the inital data are of limited value.