Keithley 2100,2000, etc or Chinese bench multimeters?

[KungFuJosh]

Yet EasyDMM is easier to use should you not have any programming skills.

Yes, EasyDMM is easier to use. It still sucks by comparison, and TestController doesn't need any programming skills if your device is already supported. You can program stuff/functions, but you don't need to.

[Caliaxy]

Yes, the drift slows down with time, until a point when it's not relevant anymore. SDM3065X only provides data for the first year (on par with 34461, indeed). Looking at the KS datasheets (for 34461 and 34465), the drift slows down a bit in the second year (by 25%). DMM6550 is more impressive, in two years drifts as much as the others in one. But they are all very small numbers.

I am afraid you misunderstood what some of these specifications mean in datasheets.
1 year and 2 year columns do not mean drift in 1st and 2nd year. These are maximum errors with 1 year and 2 year calibration intervals.

I understand that I just inferred that if those maximum errors guaranteed by the manufacturer at delivery embed the drift of the reference (which, whatever it is, gets smaller with time) the lower degradation of accuracy in the second year after calibration reflects a one year longer-aged reference, right-shifted on its exponential curve. In turn, this would show a longer-aged reference in DMM6550 than in KS34665, for instance.

Of course, this doesn't mean I am not wrong. There must be other factors/components that make DMM6550 better than KS34465 and KS34665 better than KS34461 and SDM3065. And no instrument gets better by just aging...

What specs in the data sheets made you infer that KS34465 might have an aged reference (as opposed to KS34461 or SDM3065)?

[Caliaxy]

i'm waiting for a Siglent sdm3065   well see  .... but the reference voltage used is no par with the oldies with lm399 or the ltz1000 (3458a)

Rigol DM3068 and Siglent SDM 3065 also use LM399...

But not aged, allegedly (at least not in Siglent SDM3065), which makes them prone to drift in time, requiring more frequent re-calibrations.

What do you mean "more frequent" calibrations?

As needed to keep it on par with the ones with aged references (if precision matters that much).

I don't think that will work. For as long as the drift hasn't made the readings go out of specification, there is no reason to adjust. A reference just needs to age to get rid of the initial drift.

Agree, I might make wrong assumptions. A 100 ppm drift would alter those specs, a 5 ppm would not.

[nctnico]

Yes, the drift slows down with time, until a point when it's not relevant anymore. SDM3065X only provides data for the first year (on par with 34461, indeed). Looking at the KS datasheets (for 34461 and 34465), the drift slows down a bit in the second year (by 25%). DMM6550 is more impressive, in two years drifts as much as the others in one. But they are all very small numbers.

I am afraid you misunderstood what some of these specifications mean in datasheets.
1 year and 2 year columns do not mean drift in 1st and 2nd year. These are maximum errors with 1 year and 2 year calibration intervals.

I understand that I just inferred that if those maximum errors guaranteed by the manufacturer at delivery embed the drift of the reference (which, whatever it is, gets smaller with time) the lower degradation of accuracy in the second year after calibration reflects a one year longer-aged reference, right-shifted on its exponential curve. In turn, this would show a longer-aged reference in DMM6550 than in KS34665, for instance.

Of course, this doesn't mean I am not wrong. There must be other factors/components that make DMM6550 better than KS34465 and KS34665 better than KS34461 and SDM3065. And no instrument gets better by just aging...

Probably by selection. Keep in mind that Keithley / Fluke and Keysight have decades of know-how where it comes to making high precission multimeters and how to age / select references to meet specifications.

[KungFuJosh]

Probably by selection. Keep in mind that Keithley / Fluke and Keysight have decades of know-how where it comes to making high precission multimeters and how to age / select references to meet specifications.

It's not that simple. All these major brands know how to do that. The same question also applies to different meters within the same brand(s).

In theory, Siglent keeps the price of their bench DMMs lower by not spending the investment pre-aging the references. I don't know to what degree that's true, but that's what I've read. My SDM3065X has excellent stability, especially for it being so young. 😉

[nctnico]

Well, you don't learn how to make samurai swords in one day. It takes decades of training and devotion.

BTW: do you have a good enough reference to back the claim that your SDM3065X has excellent stability? You'd need some really sophisticated gear to determine stability for a 6.5 digit DMM.  In general most people / labs don't. You'd need a metrology grade voltage standard.

[Martin72]

I bought a handful of SDM3045Xs for “my” test field. We don't need more than 3 decimal places of accuracy and as long as the annual external calibration “says” that the Siglent meets its own specifications, everything is fine.
The same applies to my private SDM3065X.

[KungFuJosh]

Well, you don't learn how to make samurai swords in one day. It takes decades of training and devotion.

Ha! That's a silly analogy. Making swords is a significantly higher level skillset. I have no experience selecting or aging ref chips, but I'm sure if I was working with somebody who was experienced, I'd be fine, and refer to them for confirmation. Electronics = measurable specs. Sword making is an art that takes hundreds or thousands of hours to be mediocre at. 😉

BTW: do you have a good enough reference to back the claim that your SDM3065X has excellent stability? You'd need some really sophisticated gear to determine stability for a 6.5 digit DMM.  In general most people / labs don't. You'd need a metrology grade voltage standard.

Absolutely! For my needs. Actually, for better than my needs. 🤣

I tested with a VREF10-001 r9 and posted the results from TestController in the Siglent DMM thread. I also shared the results with Doug (the maker of the VREF) and he was impressed with the Siglent's performance, and said the noise seen in the results is right on with the specs of the reference chip itself.

Thanks,
Josh

[coromonadalix]

for the based lm399  ltz 1000 and such,  aging will make them better and less drifty ...

i had proof of that  with oldies like 34401a  and their replacement L4411a and 34410a,  theses later where as good ...

even old Keithley 196 and Tek DM5120  based on lm399  where darn good too,  but 300 volts, not 1kv

surely  other in the k2xxx series would be as good

I'm not bashing Siglent or others   but ...  they will age in a sense ... all electronics do age,   but temp controlled zenner will stabilize over time

but now  all comments including mine are derailing the @op thread,  he'll decide whats best for him,  not us  loll



Now other brands tries to get the best bang for the buck,  but some  will never be as good as the past or actual main actors  ...  see the top end 8.5 digits

Nothing as really changed over the years,  just some black"ish"  phenomenon 

Fluke / Wavetek, Advantest, Keysight, Keithley,  Datron, Prema, Solartron   ...  some of them haven't got any successor or extended models,  just a black 3458a  ... 
Or the ones we find on fleabay, that can be repaired and saved ...

Now there is the new REF xxx series   who seem's good, pretty good ?, well see how they evolve ...

[KungFuJosh]

I just watched the rest of Tony's video. The Siglent was not warmed up in Tony's video, but it was more accurate according to the outdated cal sheet. 😉 Then I watched a couple reviews of the DMM6500.

Now I'm even more interested to see the SDM4065A when it comes out.

[Kleinstein]

....
What specs in the data sheets made you infer that KS34465 might have an aged reference (as opposed to KS34461 or SDM3065)?

For the 10 V DC range the 34465 specs are 10 ppm for 24 hours, 20 ppm 90 days and 30 ppm for 1 year. So only 10 or 20 ppm of room for drift over 90 days / 1 year.
The specs for the LM399 allow for a typical drift of 20 ppm for the frist 1000 hours. If not preaged this drift part alone would add at least 20 ppm for the first 40 days.  To reach the tighter specs it would need some preaging / burn in and possibly also some selection.

A good 6 digit DMM should have some burn in or at least pre-aging of critical parts to get the typical 35 ppm accuracy specs. Without it the LM399 ref. is barely stable enough for this. The meters based on a SD ADC chip have the additional drift part from dividing the reference down for the the ADC and scaling the input signal from 10 V to some 2.5 or 5 V range of the ADC.

[Zucca]

K2001 (and M) sometimes appear on ebay for a bit less than $800 so it's indeed a great value meter, but the voltage range noise as tested by myself suggests it's rather disappointing for reading the 7th digit (the noise can hardly be averaged down to 0.1ppm of range).

Carefull with cheap K2001 on ebay, they are often corroded by the leaking caps.
Yes it is a bit noisy for a 7.5, but the noise is not due to the HW but for the SW/DAC quantization algorithm.

That said you don't buy the K2001 because of the low/high noise, but for all the rest.

[djsb]

Thanks everyone for your replies and general interest.
It looks like I'll be buying a Keithley DMM6500. I'll post back here later when it arrives.





PS, It arrived yesterday morning. I bought a Certified Pre Owned DMM6500 from TestEquity in the UK. It has 3 years guarantee and the calibration runs out June 2025. It came with a cal certificate and still had the plastic screen protector. The instrument was shipped to me from their warehouse in Germany via the UK office in St Asaph in North Wales. The cost was £984 inclusive of VAT and postage. I'm going to be very happy with it.