Goldstar DM 441B zero adjustment/calibration

[Leuven]

I just bought a second hand Goldstar DM 441B. Nice old school unit, 20 000 counts, manual range only. I've always wanted a bench multimeter and in the end it was between this one and the UT-803.

One thing got me a bit baffled though on the DC voltage measurement. While the unit seems to be measuring in the same ball park as my fluke 115 (I don't have a voltage reference, but would expect the Goldstar to be the more accurate unit), there is something strange going on with the zero adjustment. When just turned on cold at room temperature, the unit starts with +0.026V and as it warms up, it eventually creeps down to zero (takes about 15 min), and then beyond into negative voltage.

Obviously, correct DC voltage measurement can only be obtained when on 0.000V. Would I be expected to check and adjust zero voltage every time I want a precise measurement? Or calibrate once when it's fully stable, but have to wait an hour for a full warm up? That seems a bit off to me. Plus, whatever drift I get when I'm not on zero, that drift doubles up for a reverse polarity measurement.

Is this a faulty unit or just the way it works?

[MosherIV]

Nice score, 4.5 digit meter 

The manual is hilarious
"4.1 Zero Adjust
Everybody can adjust to "0" at initial screen to get a exact reading value."
 

From the nice picture you posted, do you see the label "ZeroAdj" and the hole next to it?
My guess is that this is for adjusting the zero, with a pot trim tool (small flat blade screw driver if you do not have one).

Yes, wait till the meter warms up and stablizes. This can take upto an hour or two.
Then adjust the zero.

Would I be expected to check and adjust zero voltage every time I want a precise measurement?

Yes but in practice you probably only need to adjust it once a month or once a quarter.
The important part is to let the meter warm up and stablize.
You only need to worry about it if you want the Precise measurement. Most of the time you will be OK with the ballpark figure ie you do not need to worry about the warm up time ie do you really need to know it is 3.289V or is 3.3V good enough?

Or calibrate once when it's fully stable, but have to wait an hour for a full warm up?

Yes, just what I said above.

Is this a faulty unit or just the way it works?

My guess is that is the way it works.
I have a Thurlby 1905a (5.5 digit) and a Solartron 7060 (6.5digit) - yes I have to wait for 1 hour for warm up before I can take accurate readings.

FYI, it is worth getting yourself a voltage reference such as the AD584 - it is a single chip precision voltage reference that provide 2.5V, 5V and 10V +/- 30mV worst case. Plenty good enough for a 4.5 digit meter. Just buy the chip, put it on a circuit board (strip board will work) and provide it with 12V or more and it will just output those voltages.
If you get a chance to check the AD584 against a calibrated meter - you then have some form of traceability.

[Gyro]

I'd say that level of warm up drift looks excessive. My 'old school' 20 000 count digit bench meters (1970s Datrons) have warm up drifts of the single digits microvolt level, and that's on the 10mV range. 26mV drift seems way over the top!. The zero adjust pot should be just to zero the LSD on the most sensitive range.

I'd have a look at things like the power rails, caps etc. Does it still show the drift with the inputs shorted? If not then it could be input bias current shifting.

[The Soulman]

My "vintage" 4,5 digit meters (keithley179 and metrawatt ma5d) read only zero's directly after power on.
So I'd guess your meter requires some tlc..

Is it in dc volts only or also dc milivolts and dc amps?

Service manual:
http://www2.produktinfo.conrad.de/datenblaetter/100000-124999/120901-sp-01-en-DM_441_B_Tisch_Multimeter_Kalibrier.pdf

ADC datasheet:
http://ww1.microchip.com/downloads/en/DeviceDoc/21478c.pdf

[Kleinstein]

From the service manual, it looks like this is a relatively poor design, with a quite drifty JFET OP at the input for all ranges. So Ac performance should be OK, but DC performance has this high amount of drift.
As the same amplifier is used for AC and DC, there is no really easy fix for this: with an AZ OP (e.g. ICL7650 an a small riser board) to fix the DC performance one would end up with limited AC bandwidth.

One might be able to fit a better OP - with these old instrument you might be lucky and it is in a socket.

[Leuven]

Thank you very much for the replies.

You're perfectly right MosherIV, it takes a full 2 hours to stabilise completely. I can now measure the total drift, it's about 35mV from cold (at ambient temp 18 degrees). But as you say, one decimal place does me for most of the time.

@Gyro Drift is the same is with the probes shorted or not, makes no difference. Maybe it's indeed excessive, or simply a design shortcoming as Kleinstein points out?

The drift is all across the DC range, BUT - if at 20/200/100V ranges it drifts 35mV, for 200mM/2V range it drifts 0.35mV (no typo!). In other words, under 2 volts it is 100 more precise than over 2 volts, in a cold state.

[Gyro]

The drift is all across the DC range, BUT - if at 20/200/100V ranges it drifts 35mV, for 200mM/2V range it drifts 0.35mV (no typo!). In other words, under 2 volts it is 100 more precise than over 2 volts, in a cold state.

That would tie up with it being amplifier drift as this is situated after the input attenuator. Even so it is at a pretty high level. This also means that it should be of sufficient level for you to trace down the chain with a handheld dmm. It's worth taking a look before accepting the situation. Also, although you can't use an autozero opamp, there are many lower drift fet opamps than the LF411 these days, so it would be worth looking for compatible subtitutes (after verifying that it is the problem). It may even be a degraded part.

[Leuven]

Just had a look inside, the op amp is an Intersil CA3140AE through-hole mounted. Is that what I'm looking for?

[Gyro]

Well if it's a CA3140 then I'm not surprised it's that drifty, they're first generation Mosfet. We are talking U1 here, right? The LF411 shown in the manual is a low drift, low offset jfet one. Does it show signs of having been replaced at some time? Maybe it was replaced with what someone had in their scrap box. You can probably find something even better with the same pinout these days. Check around and make sure that's where the offset is coming from.

EDIT: The AD711 from a quick search.

EDIT1: Ah no, you're looking at U5 in the ohms circuit.

[Leuven]

Well if it's a CA3140 then I'm not surprised it's that drifty, they're first generation Mosfet. We are talking U1 here, right? The LF411 shown in the manual is a low drift, low offset jfet one. Does it show signs of having been replaced at some time? Maybe it was replaced with what someone had in their scrap box. You can probably find something even better with the same pinout these days. Check around and make sure that's where the offset is coming from.

EDIT: The AD711 from a quick search.

EDIT1: Ah no, you're looking at U5 in the ohms circuit.

I found it, it was hidden on the front board. U1 is a HD74LS02P. To be perfectly honest, I wouldn't know how to troubleshoot this thing, but I'd be willing to pluck it out and replace it with a more performant/low drift part if suggested.

[The Soulman]

Well if it's a CA3140 then I'm not surprised it's that drifty, they're first generation Mosfet. We are talking U1 here, right? The LF411 shown in the manual is a low drift, low offset jfet one. Does it show signs of having been replaced at some time? Maybe it was replaced with what someone had in their scrap box. You can probably find something even better with the same pinout these days. Check around and make sure that's where the offset is coming from.

EDIT: The AD711 from a quick search.

EDIT1: Ah no, you're looking at U5 in the ohms circuit.

I found it, it was hidden on the front board. U1 is a HD74LS02P. To be perfectly honest, I wouldn't know how to troubleshoot this thing, but I'd be willing to pluck it out and replace it with a more performant/low drift part if suggested.

Nope wrong U1, in the bom of the service manual the U1 you're looking for is labelled U01, could be it is also labelled that way on the pcb.
In either case it should be close to the zero adjustment pot.

[ModemHead]

Look here.

[Leuven]

Sorry, yes that's the one, it's an LF411ACH as per manual.

[Kleinstein]

So they spend some money on using the TO99 version of the LF411, but it looks like this is not really good enough. It is specified at typical 7 µV/K, so not really low drift. The observed drift seems to be even higher.

There are better more modern JFET based OPs, but many of them have a different offset trimming or none at all.

An AZ OP ICL7650 would be slightly slower (e.g. 2 MHz compared to 3 MHz GBW), but this may not be that bad. This one would still need a small adapter board for the external capacitors. There should be no need for an offset adjustment in this case, as the ADC should also have a very low offset.