TTI 1604 Bench Multimeter (MAX134 based Multimeter) Negative Voltages/Offset

[TheBay]

As my country is in lockdown I decided to go through some of my test equipment and check it all over, carry out any maintenance, calibration or preventative maintenance that is due.

I've worked through a load of power supplies and decided that my TTI 1604 could do with a calibration and checking over (Last cal was in 2010)
surprisingly it was still pretty close and only a few uV off. Contrary to other peoples views I'm quite fond of this bench multimeter, granted it's got one of the slowest continuity tests I've ever seen on a multimeter but I only ever use it for Volts/Current measurements so the other features don't matter to me, plus I've lost count how many hand held multimeters I have.

So I decided to check the calibration on the 1604 and get the service manual out and calibrate it.
I powered it up and let it settle for a while, then connected one of my voltage references to it, I wasn't paying much attention as there are test leads everywhere and I had plugged them in backwards, no problem it's just displaying a negative voltage so I just swapped the leads back over and I noticed something I'd never realised in all the time I've owned this. The reverse polarity negative voltage was reading a little higher than a positive voltage.

Initially I thought something hadn't warmed up properly or a difference in resistance on the test leads, so swapped the other end instead, then grabbed some more test leads and found the same thing.

It's not out by much, approximately 200-300uV below 4v on the 4.75 range and sometimes 1mV on other ranges, not much but it bugged me.
I thought no problem I'll just carry on and calibrate it, which I did.
I worked my way through the service manual and adjusted anything if it was out. However I did expect it to have a offset adjustment or some form of calibration for negative voltages - It doesn't.

I left it on all day and checked it later on and there was no drift, happy with that I put it all back together.  I had never used this meter for negative voltages before, its usually just set on one mode and monitoring something for voltage or current changes, I never use it for fault finding. Hence not noticing this before.

As I now know about it, it bugs me. So I went through the service manual again to have a look if I'd missed anything or if there is anything on the input chain that could be changed to adjust this.

This is basically is a dual slope type ADC based multimeter, using a MAX134CPL DMM ADC, I tried a few things earlier and removed all the varistors on the input to see if any of the front end protection was a bit leaky but no change, I then took the integrator capacitor out and reversed it, hoping the reading would go the other way as that would be a simple fix, but nothing I did would remove that small 200-300uV/1mV difference between polarities.

I had a look at the MAX134CPL datasheet and nothing is obvious there and the zero input offset is done internally, so I don't think there are any  external components influencing the zero offset. Also it does always display 0 when not measuring something and no probes plugged in.

I know this is a very small difference and probably well within it's rated specifications, but I would like to find a way of correcting it but I'm at a loss, hopefully someone is familiar with this multimeter or any others based on the MAX134.

One of my Fluke 77's has a very slight difference when reading the opposite polarity but that is when a voltage difference is outside of its displayed range and it can't quite decided when to go to up a count. It does the reverse when you reverse the leads. And my AVO Badged Metrawatt does it on lower voltages. I tried a load of other multimeters here and they don't do it so luck of the draw with those.

Any information would be gratefully appreciated, I'm probably wasting my time but I haven't got much else to do in the workshop over lockdown.

[TheBay]

Just noticed if I put it in mV mode it does the opposite, when reading the same voltage reversed it shows a slightly Lower reading when measuring negative voltages.
Still only by a very small fraction.

[Kleinstein]

The input "amplifier" (actually gain < 1, so more an attenuator)  of the max134 is inverting, so it makes sense to see the opposite behavior for the mV range (and likely current).

A possible problem with linearity could be from the integration capacitor. This is a general weakness of dual slope ADCs - dielectric absorption in the capacitor can cause INL errors. In this converter it may effect both polarities in a different way.
AFAIR is was already difficult to get full 4.5 digit accuracy with a normal PP capacitor with the ICL7135.
The observed error is in about that order of magnitude.
I did some tests on DA and found a possible better alternative to the PP capacitor ( TDK brand MLCC with C0G dielectric, other brand C0G were more like similar to PP). I have not tested the THT form factor, but it may be worth a try, as they are not very expensive.


An obvious point would be checking the supplies. 

[TheBay]

The MAX134CPL is technically a 3.75 digit ADC so they are cheating slightly on this as it is a 4.75 Digit Multimeter with some additional logic and multiplexing etc. Though the MAX134 is a 40,000 count ADC.

First thought was the integration capacitor, but I assumed (possibly wrongly) taking that out of the board and putting it back in reversed the difference would change, i.e. positive would read higher instead of negative, but it didn't. If I was wrong here then changing the capacitor to what you recommended or similar might be worth trying.
The integrator capacitor in this is a Wima Polypropylene 4.7nF (Green case) and I assume its 5% tolerance.

That capacitor is quite microphonic as the readings will change if I tap it, but that is to be expected at that size and value. I was tapping the board with the back of a wooden brush to see if there are any cold joints.

Thanks for pointing that out with the mV range, that makes sense!

The input "amplifier" (actually gain < 1, so more an attenuator)  of the max134 is inverting, so it makes sense to see the opposite behavior for the mV range (and likely current).

A possible problem with linearity could be from the integration capacitor. This is a general weakness of dual slope ADCs - dielectric absorption in the capacitor can cause INL errors. In this converter it may effect both polarities in a different way.
AFAIR is was already difficult to get full 4.5 digit accuracy with a normal PP capacitor with the ICL7135.
The observed error is in about that order of magnitude.
I did some tests on DA and found a possible better alternative to the PP capacitor ( TDK brand MLCC with C0G dielectric, other brand C0G were more like similar to PP). I have not tested the THT form factor, but it may be worth a try, as they are not very expensive.


An obvious point would be checking the supplies.

[Kleinstein]

Reversing to capacitor polarity would change the parasitic capacitance from the outer layer, but not the effect of leakage and DA.  4.7 nF is the standard value for the MAX134 and the tolerance does not matter much in a dual slope ADC.

With an alternative capacitor the expensive part would be shipping - so maybe something to keep in mind for the next order.

[TheBay]

Reversing to capacitor polarity would change the parasitic capacitance from the outer layer, but not the effect of leakage and DA.  4.7 nF is the standard value for the MAX134 and the tolerance does not matter much in a dual slope ADC.

With an alternative capacitor the expensive part would be shipping - so maybe something to keep in mind for the next order.

I get free shipping from component supplies as I have a trade account with a few, I will try RS first as they tend to be next day.
The only through hole capacitors were exactly the same as what is in there, but I would get a few to try, I'll have a look at the SMD's as I can easily fit one across the pcb.