M-3650D - Old Multimeter Repair (drifting voltage offset)

[A5tro]

Hello everyone!
My grandpa gave me his old multimeter two weeks ago.
It is a Voltcraft M-3650D, which seems to be a re-branded Metex M-3650D. I think in the US it was sold as the RadioShack 22-168.

There is a problem when trying to measure voltage and I thought I could try to repair it, as I think it is a pretty nice meter otherwise.
I am pretty new to electronics (and new in the forum ^^) and after many hours of staring at the schematics I thought that I should probably document what I have and ask for some help. (Because if I can't solve it, I am not learning much either.) So, here I am!
Already I am very grateful for everyone who takes the time to read this.



===The Problem (initially)===
Probes connected to COM and V/OHM.
Range: 20VDC
Result:
The display shows a reading of -14V shortly after turning the meter on. It quickly rises to -12V and further, gradually slowing down.
I connected a 9V rechargeable battery and left the meter switched on over night. After tens of hours the offset was at approx. -0.1V.
I disconnected the battery to recharge it and tried it again a few days later with the same result of a slowly decreasing offset from 0V.


===My Initial Thoughts===
A temperature sensitive component?
A bad/missing reference?
A bad connection or defective component leaving some pin floating and collecting parasitic charges?


===Some Documentation===
The sigrok website has a little bit of documentation on the voltcraft-version
https://sigrok.org/wiki/Voltcraft_M-3650D
and a little bit more on the Radioshack-version (including very nice pictures!)
https://sigrok.org/wiki/RadioShack_22-168

There is a link to a service manual for the M-3610D and M-3650D, containing schematics and parts-lists.
https://www.eserviceinfo.com/preview.php?fileid=33728&file=Metex M3610D-3650D.gif

The schematics are cut in half which I found hard to read, so I exported the images (using XpdfReader) and stitched them back together.
It took me waaaay to long to understand how to read the range-selector matrix, so here is a small explanation:
Every ring on the range selector pcb is one column in the schematics. (Two PCBs with each 6 rings for a total of 12 rings and corresponding 12 columns in the schematics.) The contacts on the range selector wheel always connect two rings (and columns) that are next to each other. (1&2, 3&4, ...)
On the left of the schematic there is a numbered legend of the range options. I thought the numbers correspond to the numbers on the right side of the selector-matrix. This is not true. (I have no idea what those are actually for.) Instead the legend is aligned with the matrix in a way to directly lable each row. I added some colored bars to make it easier to read.

schematics

(How do I make the images appear actually inline at this position?    ... images are at the bottom)

I then tried to identify the main parts and functional blocks with ... mixed results. I can not find any documentation about IC6 and IC11. Could those be an amplifier and a frequency counter?
I added some lables and tried to highlight the signal paths for the 20V and 200µA ranges in green and red:

highlighted schematics



===The Problem (further investigation)===
Measuring voltages still works when accounting for the voltage offset. For example when the offset is at -8V, connecting a 9V battery returns a measurement of approximately 1V.

In the voltage ranges, the offset is the same, regardless of the selected range. For example, it measures -3V for the ranges 1000V, 200V and 20V, but out of limit for the ranges 2V and 200mV. As the offset drifts closer to 0V over time the lower voltage ranges can be used as well.
I thought this must mean that the offset-voltage exists across the entire resistor-chain (R1-R6) (used as voltage divider for the different voltage ranges) and that I should be able to force it to zero by shorting the probes (COM & V/OHM). But that did not work.
I was confused and thought that I should try looking somewhere else to get a wider picture of the problem.

I switched to the amp-range of the meter. It uses a different resistor-chain (R9-R12) but other than that the signal seems to take a very similar path throughout the meter.
For the different amp-ranges it displays the following readings (if V-range would show an offset of -10V):
200µA --> -1.0µA
2mA --> -0.010mA
20mA --> -0.10mA
200mA --> -1.0mA
That is always 0.5% of the selected range (contrasted to the v-range where the offset has a consistent absolute value across all ranges).
When in the 200µA-range (and only there) the offset can be reduced to zero by shorting the test probes (connected to COM and mA).


Something I noticed while looking at the schematic was that the "ground" connected to the COM terminal is not the same as the negative pole of the battery "V-" (foolish thought of a beginner, I guess ^^). I believe this is because some chips need symmetric supply voltages? As far as I can see "V-" is only connected to IC6, IC4 and Q9. As the offset is negative I thought I should check the related components and started by de-soldering C3 (close to IC6) as it looked to me like it would be connected to some negative potential and it is fairly close to the signal path.
I have one of those very cheap component testers. I put C3 into the socket and measured 337nF, 20Ohm ESR and 0.7% Vloss. I think that is normal for these small electrolytic caps?
(Afterwards I noticed that if C3 would leak current into COM that I would probably measure a positive voltage offset, not a negative one.)

===What I will do next===
As everything is so tightly packed (testing C3 wasn't easy/quick) I would like to avoid de-soldering and testing all the components in the signal-path. Instead I will go back to trying to wrap my head around what I would need to change in the schematic in order to cause the observed effects. I am still confused as to how everything connects together. Maybe I should make a simplified schematic with less zig-zag paths to help my monkey brain.
I will also try measuring the voltage across R69 and R70 (once I have put everything back together again) in order to see how much current goes through them.
In the meantime I thought I should post what I have so far. Not because I want someone to solve this for me entirely ... but I am very slow and sometimes feel a bit lost. So, if someone notices something really obvious or could point me to some useful troubleshooting techniques I would be very very grateful!


Greetings from Germany (and sorry if my english is a little bit funny  )

[retiredcaps]

Suggestions.

1. Check PTC1. It should be around 1.5k ohm.  You can measure it in circuit.

2. Measure the input impedance of the meter when the rotary switch is set to DCV and turned on.  You can do this by using another working meter and putting it's red probe into the V jack of the 3650 and it's black probe into the COM of the 3650.  See my picture at

https://www.eevblog.com/forum/repair/fluke-87-i-repair/msg390462/#msg390462

as example.

[A5tro]

Sorry for the long pause, I didn't get to it earlier.

I used a cheap second multimeter (DT9208L) to do the following measurements directly at the unpowered circuit:

PTC1 = 1.49kOhm
R69 = 0.98MOhm
R70 = 99.2kOhm
R1+R2+R3+R4+R5+R6 = 9.87MOhm

Considering 1-2% error, they all look fine.


I then reassembled it (but no casing) and measured the following at the input terminals:

old DMM in 20VDC-range but turned OFF, measuring from COM to V/OHM-terminal with second DMM set to 20MOhm-range
--> R_input=9.95MOhm

old DMM in 20VDC-range but turned ON, measuring from COM to V/OHM-terminal with second DMM set to 20MOhm-range
--> R_input=8.99MOhm

That almost looks like I only measured R1?


I then measured the voltage drop across R69 and R70 in order to see how much current flows through them, possibly influencing the measurement.

old DMM turned ON in 20VDC-range, showing an offset of approx. -3.5V
-->V_r69 = 100µV (COM of second DMM at IC8)
----> I_r69 = 100pA (go into IC8)

--> V_r70 = 35mV (COM of second DMM at IC6)
----> I_r70 = 0.35µA (go into IC6)


Then I switched the old DMM to the 200mVDC-Range and shorted its COM and V/OHM-terminals. This reduced the shown offset to zero! (I forgot to test that in my first post.)
I think this should mean that the shorted current is responsible for the offset.

Then, instead of shorting them directly with a wire, I used the second DMM (set to 20µA-current-range) to "short" COM and V/OHM of the old DMM.
--> old DMM showed offset of approx. -3.5mV
--> second DMM measured a current of approx. -0.35µA

This looks like I directly measured the measurement shunt resistors in the second DMM to be 10kOhm, which seems reasonable?

It is also very similar to the current flowing through R70. (A little bit of guessing was required for all measurements as the values were constantly drifting once the old DMM was switched ON.)

I guess that means that IC6 and its surrounding components are very suspicious.
Unfortunately I still don't know what IC6 is.

[Messtechniker]

Selective warming (not heating yet) or cooling (initially with air from a fan passing though a piece of hose, for example) specific components might be helpful in determining the troublemaker. Leaving fewer rabbit holes to go down thereafter.

[Kleinstein]

The large current through R70 seems to be the main problem - so something like a bad IC6 - whatever this chip is. It looks a little like RMS/DC converter, but not sure.

[floobydust]

I start with measuring the voltages on power, measure at the A/D or IC4 whatever it is "MD-08" it might be a charge-pump for -ve bias for the ICL7106 A/D, which the MCU then reads. Notice there are four grounds in the DMM, each floating to their own voltage. It's not easy to understand them or figure them out:
AGND (pin 30), ACOM (pin 32), DGND (pin 37) and battery V- (pin 26), and switched power V+. Try using V- as the common point for your measurements. Let us know what you get.
Sometimes I've seen a damaged A/D but still sort of alive, so it loads down say ACOM but readings have an offset and clip past a certain point.

[pjwum]

The large current through R70 seems to be the main problem - so something like a bad IC6 - whatever this chip is.

I had a 3640D with comparable erratic behaviour and it turned out that one gate of the 4066 was dead. A short in the 4066 could possibly drain current through R70.
The 4066 has four electronic switches. Check the voltage at the control inputs and then the according switch contacts (meter powered on of course). You also could look for temperature changes of the chip package.

[Kleinstein]

A bad 4066 could possibly explain to much current toward IC6. This would be an easy to get cheap part.

Another possibility could be a bad electrolytic cap around IC6.  IC6 seems to be some daughter board with a SO8 chip under a black coating, so there could still be a chance to fix it.

[A5tro]

Thanks for all the answers!

Testing caps in circuit with the second DMM (DT9208L)
C3=310nF
C4=3.7µF
C5=3.5µF
C6=8.5µF
C9=9µF
As this meter seems to always measure capacities a little bit low, it doesn't look like there are obvious faults.

Warming the whole upper PCB and IC6 with a hair dryer (not full throttle) does not influence the offset in a significant way. Selectively warming IC6 through a conical paper funnel didn't do it either.
I was not able to test whether IC2 is sensitive to temperature, as it is hidden below the top-PCB.

But touching IC6 (somewhere close to pin 3 I think) does change the offset. When trying it this time it went from -0.3V to -1V, but it can go much further when given more time or contact area. In fact, touching the PCB close to IC6 has the same effect and might work even better.
I thought that maybe I am acting as an antenna or something, so I also touched a ground connection with the same finger that touched IC6, but the result was the same (see pictures below).
So, I don't think that IC6, IC7, IC8 or any other component close to them on the top-PCB are temperature sensitive.
It seems more like the capacitance or resistance of my body was the influence.

Measuring power and ground potentials at IC2, all in reference to the negative battery terminal "V-":
Vcc (at battery) = 9.24V
V+ (pin1) = 9.15V
V- (pin 26) = 0V
AGND (pin 30) = 6.15V
ACOM (pin 32) = 6.15V
DGND (pin 37) = 5.47V
(While measuring, the offset drifted from -1.1V to -0.9V.)
I appended the datasheet of IC2. I think on page 5 it shows how ACOM is maintained stable by the internal bandgap-reference? (Haven't read it yet.)
Is DGND supposed to be like this? With a difference of 0.68V to AGND it almost looks like there is a PN-junction between them somewhere?


Checking IC8 (4066) (DMM ON; 20VDC-range)
Voltages relative to "V-" (COM of second DMM)
VDD (pin 14) = 9.16V
VSS (pin 7) = 6.15V

Voltages relative to "VSS" (COM of second DMM)
VDD (pin 14) = 3.01V
VSS (pin 7) = 0V

control 1 (pin 13) = -0.65V    (pulled to DGND by CPU with a little series resistance?)
input 1 (pin 1) = 0V
output 1 (pin 2) = [-10mV , -50mV] for offset= [-1V , -5V] (directly correlated)
R_ON_1 (pin 1-2) = >2k Ohm

control 2 (pin 5) = 3.01V
input 2 (pin 4) = [-10mV , -50mV] for offset= [-1V , -5V] (directly correlated)    
output 2 (pin 3) = [-10mV , -50mV] for offset= [-1V , -5V] (directly correlated)
R_ON_2 (pin 4-3) = 400 Ohm    (measured with second DMM in 2kOhm-range) (value looks reasonable from what I found in datasheets)
U_4_3 (pin 4-3) = 0V       (no voltage drop, independent of offset --> no current through the this switch)

Switches 3 and 4 are not connected.

While furiously measuring I also noticed that the offset is extremely sensitive to humid air, as I was breathing close to the multimeter. Intentionally blowing on it made it jump to -20V and further.

There are a couple of things that I don't understand yet  ... but to me there is nothing that stands out as obviously broken.
The black stuff on IC6 is unfortunately very hard (similar to those IC-packages), so I don't think it will be easy to get it off.

[pjwum]

the offset is extremely sensitive to humid air, as I was breathing close to the multimeter.

Did you already examine the boards carefully for residues of battery leakage? Especially the connectors between the two pcbs? You could also apply your funnel technique when breathing to your multimeter.

[jkostee]

Did you ever discover the problem?  My Radio Shack version is doing the same thing.  I've had this thing so long it almost has sentimental value.

John

[mqsaharan]

As Kleinstein suggested, IC6 is TRMS converter and gets switched in when AC/DC switch is pressed. Looks like it is still available for purchase at
https://www.tekkna.it/product/22686/CIRCUITO-IBRIDO-ADC-HD-201CR-A.asp

My suggestion is to check R68 first with another meter. And if you can remove it you can check whether there is something wrong with the IC2 circuit, i.e., the components surrounding IC2 or the chip itself.
Input to the ADC is through R68 which is directly connected to pin31 of IC2. Removing it or at least one leg of it will separate ADC input from the rest of the meter input circuit and you can check if meter still shows any residual reading on the display, it shouldn't. You can also input small signal like 100mV through a 1M resistor regardless of the rotary switch position* and read the display to confirm IC2's operation. I'm sorry, I don't remember, should the display read 100 or 1000. With no input at pin31 or shorting pin31 via 1M resistor to pin30 or meter's COM terminal the display should read zero.
Anyway, it is a quick way to isolate the fault between IC2 or the rest of the meter input circuit.
Please refer to the MAX130 datasheet for checking on proper operation of IC2.

*After thinking about it a bit, I think the rotary switch should be positioned at some V range for that test to be true since the display is controlled by a microprocessor.

I hope it helps.

[A5tro]

Did you ever discover the problem?  My Radio Shack version is doing the same thing.  I've had this thing so long it almost has sentimental value.

John

Hi John, unfortunately the project is on ice since my last post.

As Kleinstein suggested, IC6 is TRMS converter and gets switched in when AC/DC switch is pressed. Looks like it is still available for purchase at
https://www.tekkna.it/product/22686/CIRCUITO-IBRIDO-ADC-HD-201CR-A.asp

Hi mqsaharan, thank you so much for your effort! I am sorry that I won't test it asap, but I promise to look into it again in the future and post updates here. Your suggestions look good. I think I am always way to hesitant to just remove components to make troubleshooting easier. Definetly something I need to learn.
I am also supprised that you found the HD-201CR ... it even is listed with the same name as on the schematic. How did I not find it when I looked for it?!?! Seems like I also need to learn how to use a search engine. ^^

Anyhow, thanks again to everyone that tried to help me with this. Actually, I have no idea why you are doing this. I don't have enough brainpower to help random people on the internet with complex problems like this. So, hats off and big thanks to everyone here!