Yokogawa GS610 source measure unit service manual needed => have it => useless!

[_Wim_]

Hi,
I know it will be difficult (impossible?) to find, but  recently acquired an SMU from Yokogawa (model GS610) that has a problem and I am looking to repair it.
I bought this unit on Ebay as “correctly working”, but it outputs approx. 3.5mA to much on all ranges below 0.5A (even on the 200µA range). On the 20µA range it output approx. 1.6mA to much! Ranges from 0.5A and above are within spec.
I have let it warm up for several hours, performed the zero offset calibration, and executed the cal function on all ranges, but the problem remains. I have probed around in the unit to try and reverse engineer it, but the unit is quite complex, and without any schematic or guide how to interprete the test points, I am afraid I will not be able to repair it.
The seller claims it was correctly working before  he shipped it, but I have my doubts. I hope I can return the unit (claim ongoing) or (preferred) I see a possibility to repair it (as I would really want to keep this unit if correctly working).
Any help would be greatly appreciated.

[_Wim_]

Just an update so others do not make the same mistake as me:

I contacted Yokogawa (german distributor) explaining my problem. The proposed to send it in for diagnose, but the diagnose only would cost 345€+VAT+return shipping. After diagnose they would quote me for the repair cost. I figured the actual repair would be way to expensive for me.

So I asked for a service manual. They quoted me 75€+VAT for the service manual, which I off course immediatly bought. In the quote they mentioned no full shematics would be available, but I block diagram was.

When I received the manual however (doc ID: sm765501-01e), I was very very dissapointed. The block diagram is the same as in the user manual (very very high level), and the only other diagram is how the pcb are connected together. The trouble shoothing guide is essentially "if no dispaly => replace display board", "if no power, replace power board"... but no usefull information at all for actualy debugging it. All PCB are full with test points, but these are not descirbed.

I contacted Yokogawa again stating my dissapointment, but they only replied the following:

"sorry, but the items you expected are confidential.
The service policy from the Headquarter is to repair it at on board level in the service department in Japan.
Outside of Japan the rule is to change boards.

Sorry, but I can give you no better answer."

So, if you expect to fix a Yokogawa unit yourselft, do not count on any support, and also do not buy the service manual, there is no usefull info inside it.

[nctnico]

The same goes for any equipment made since approx. 2000 from any manufacturer.

Still, with some time and effort you should be able to figure out how it works and hopefully fix the problem. Usually it is something really simple but it can be hard to find.

[_Wim_]

What worries me is that I can see somybody has been in there before and already replaced multiple parts. But I have not given up yet, but the reverse engineering is hard with a multilayer pcb and small surface mounts components...

[Performa01]

Thanks for sharing your experience - what a bummer!

So the general rule for buying used gear is: make sure a comprehensive service manual including all the schematics is available before pulling the trigger!

An SMU is a pretty useful tool for sure, and usually quite expensive. Sorry it didn't work out for you.

[nctnico]

What worries me is that I can see somybody has been in there before and already replaced multiple parts. But I have not given up yet, but the reverse engineering is hard with a multilayer pcb and small surface mounts components...

Better return it or get ready to get your hands and probes dirty. I had a similar project a couple of years ago and only a block diagram to work with.
https://www.eevblog.com/forum/repair/e4421b-restauration-project/

[_Wim_]

I do not have the option any longer to return it. The claim was going nowhere, and sending the unit back before I got my money back was not looking interesting. And when Yokogawa told me they would sell me a service manual, I stopped the claim because I really wanted to keep this unit (in hindsight not a smart move). So hopefully I can fix it, becasue I paid 800€ for this...

[cncjerry]

Is here an offset you can apply to the output in the UI?  The fact that it is consistently high sounds like calibration maybe, or current sense resistor out of spec, or somebody applied an offset through the UI.  You should be able to find the sense resistors pretty easily. Also, since it is only on one range, maybe a relay is stuck?

I really like their equipment but have yet to find one I can justify.

[_Wim_]

Thanks for your feedback. There is indeed a possibility to provide an offset, but depending on the range, it become problematic (on a 20µA range a 3.5mA is quite big). The sense resistors are indeed easy to find, but the problem exists on all ranges below 0.5A, and this is covered by multiple sense resistors. This is for sure not only a calibration issue, as a deviation of more then 10% is considered a broken unit, and on the low ranges the deviation is >1000%...

As this is mainly an analog unit, I am looking at all opamps that could have an offset problem. More specifically, I am trying to find out what is different between the >0.5A ranges, and the one below.

I am also trying to figure out how the analog board interfaces with the digital section. So far I have found out (I think) the following: there are 2 DA converters (LTC1590 and LTC1654) that generate the analog setpoint (voltage or current) Both are used together somehow (one for the smaller increments, and one for bigger increments). So far I have only found 1 part that I think could be the AD (a yokogawa custom part). I think this AD does not read the actual value, but only the difference compared to the generated setpoint (seems logical to me?)
 
This unit is completely filled with analog switches and quite a few yokogawa custom parts, which makes is quite hard to find out exactly what is going on. One fortunate thing is that the analog board can be made good accesable with everything still connected.

I have attached a picture of the analog board just for reference. I made some annotation on there with the parts known.

[nctnico]

I'd also check the power supply voltages and see if known chips (the ones you can find a datasheet for) have a sensible power supply voltage (like equal positive/negative)

[_Wim_]

Thanks, I will do that also. From my notes I have checked the power supply of the analog board, and the 2 voltages references:

+15V = 15.210 V
-15V=  -15.217 V
5V = 5.064V

LT1009: 2.4977 V
LT1021BCN8-10   : 10.006 V

So these are for sure OK. I did not check every individual chip (yet).

[_Wim_]

Also, from my notes, these are the currect deviation for each range:

Range/Deviation
20µA range:  1.595 mA ? SPEC = +-(0.03% of setting + 50nA) =>NOK
200µA range: 2.863mA ? SPEC = +-(0.03% of setting + 300nA) =>NOK
2mA range: 3.422mA ? SPEC = +-(0.03% of setting + 3µA) =>NOK
20mA range: 3.463mA? SPEC = +-(0.03% of setting + 30µA) =>NOK
200mA range: 3.484mA ? SPEC = +-(0.03% of setting + 300µA) =>NOK
0.5A range: -1.4 mA (last 2 numbers  fluctuate)   ? SPEC = +-(0.03% of setting + 5mA) => OK
1A range: -1.4 mA (last 2 numbers  fluctuate) ? SPEC = +-(0.03% of setting + 5mA) => OK
2A range: -1.4 mA (last 2 numbers  fluctuate) ? SPEC = +-(0.03% of setting + 5mA) => OK
3A range: -1.4 mA (last 2 numbers  fluctuate) ? SPEC = +-(0.03% of setting + 5mA) => OK

[Performa01]

Well, these specifications aren't spectacular and shouldn't be difficult to meet.

Have you actually checked the current ranges 0.5A and above to be within spec at full scale also? What DMM are you using for that? The constant -1.4mA offset error together with an instability >10µA makes me suspicious - that should not happen.

If I were in your shoes, I'd probably do the following:

Try to figure out where the sense lines of all the shunt resistors finally go, Supposedly some switches for range selection, but finally the signal has to go into some amplifier that does the current regulation.

I strongly suspect that the problem will be in this feedback path or with the amplifier itself. Try to find out what OpAmp is used here. Normally, I would expect an autozero CMOS amp, which could be defective or it is one of the kind with external charge storage capacitors, which in turn might be bad and should be replaced by proper polypropylene types. But with these specs it might just be some industry standard JFET amp - and that is rather unlikely to fail.

My first suspects would be the range switches anyway. If they're mechanical, check them thoroughly. If it's relays, some of them are very prone to increasing and unstable contact resistances with time. But then, from your photo, I only see modern relays, some of them from NAIS as it seems, these should be fine and reliable. But maybe there are also some analog switches/multiplexers involved. During operation, you should check the voltage drop across them to make sure they're ok.

Finally, there's always a chance for broken wires, corroded connectors or even cold solder joints.

The easiest way to know if there's a problem in the sense signal routing, is to measure the voltage drop between shunt resistor and amplifier input. It will not be zero, even if the sense current is negligible, because of thermal voltages, but whenever you see voltage drops in excess of some 50 microvolts, you should take a closer look.

Best of luck!

[_Wim_]

Well, these specifications aren't spectacular and shouldn't be difficult to meet.

Have you actually checked the current ranges 0.5A and above to be within spec at full scale also? What DMM are you using for that? The constant -1.4mA offset error together with an instability >10µA makes me suspicious - that should not happen.

If I were in your shoes, I'd probably do the following:

Try to figure out where the sense lines of all the shunt resistors finally go, Supposedly some switches for range selection, but finally the signal has to go into some amplifier that does the current regulation.

I strongly suspect that the problem will be in this feedback path or with the amplifier itself. Try to find out what OpAmp is used here. Normally, I would expect an autozero CMOS amp, which could be defective or it is one of the kind with external charge storage capacitors, which in turn might be bad and should be replaced by proper polypropylene types. But with these specs it might just be some industry standard JFET amp - and that is rather unlikely to fail.

My first suspects would be the range switches anyway. If they're mechanical, check them thoroughly. If it's relays, some of them are very prone to increasing and unstable contact resistances with time. But then, from your photo, I only see modern relays, some of them from NAIS as it seems, these should be fine and reliable. But maybe there are also some analog switches/multiplexers involved. During operation, you should check the voltage drop across them to make sure they're ok.

Finally, there's always a chance for broken wires, corroded connectors or even cold solder joints.

The easiest way to know if there's a problem in the sense signal routing, is to measure the voltage drop between shunt resistor and amplifier input. It will not be zero, even if the sense current is negligible, because of thermal voltages, but whenever you see voltage drops in excess of some 50 microvolts, you should take a closer look.

Best of luck!

Thanks for the many great tips and sorry for my late reply. Due to holidays and other priorities, this has been delayed a little. Following your advice, I started looking again from the sense resistors.

The switches devices seems ok, 1 sense resistor is selected for each range:
0.5A to 3A range = 0.05000 ohm sense (4-wire)
200 mA range = 1.0000 ohm (2-wire)
20mA range = 10.000 ohm (2-wire)
2mA range = 100 ohm (2-wire)
200µA range = 1 K (2-wire)

The relay for switching between 2-wire connection and 4 wire connection also switches correctly. I also checked electrical connections (wires) to the front panel, no issues found.

But I did found something very strange: the voltage over the sense resistors matches the requested current setpoint, but the actual current is incorrect (checked with Brymen BM869S and Philips PM2525, both give identical results)

200mA range (Shunt  1 ohm) => with setpoint of 100mA => 100.087mV over shunt but actual current going out is only 96.3mA 
20mA range (Shunt  10 ohm) => with setpoint of 10mA => 99.94mV over shunt but actual current going out is only 6.51mA  ?   
2mA range (Shunt  100 ohm) => with setpoint of 1mA => 99.70 mV over shunt but actual current going out is  -2.4 mA 
200 µA range (Shunt  1Kohm) => with setpoint of 100µA => 99.93 mV over shunt but actual current going out is  -2.7 mA 
   
The opamps in the feedback path are mostly BB OPA627AU type, but so far I am still trying to figure out the exact routing (4-layer PCB with lots of ground plane, making it very hard to follow internal traces).

As to your question of the full scale range above 0.5A, they indeed to also not meet spec:
3000mA setpoint => 2990,7 mA actual (max deviation allowed: 5.9mA) => NOK
2000mA setpoint => 1992.8 mA actual (max deviation allowed: 5.6mA) => NOK
1000mA setpoint => 995.2 mA actual (max deviation allowed: 5.3mA) => just ok
500mA setpoint => 496.4mA actual (max deviation allowed: 5.15mA) => just ok

Edit: shunt high current is 0.05 ohm instead of 0.5 ohm

[Performa01]

When looking at your measurements, it is clear that in the ranges up to 200mA the output current always seems to be some 2.7 ~ 3.7mA too low, for the 1xx setpoint. How is it at different setpoints? Is the error constant or proportional?

Anyway, we have some strong hints, as the voltage over the shunt resistors appears correct, but the output current is too low by either a pretty constant current or a constant percentage of the setpoint (please check what is actually applying here) regardless of the range. Okay, it looks a bit different for the ranges 500mA and above, but not by much, as it essentially just seems to be twice the current  compared to the lower ranges. The difference is, you have fully exploited these higher ranges, whereas you only used 50% of the lower ranges, so this is a hint that the error is actually proportional to the setpoint, i.e. some percentage of it, but still totally independent of the range.

Now we should try hard to find a hypothesis how something like this could happen…

The first conclusion is that it has something to do with the voltage, as this is the only constant across the ranges. Since the voltage across the shunt appears correct, it most likely equals the setpoint voltage 0 ~ 200mV. No wonder you couldn’t find anything wrong in this area.

The errors can only be explained by some additional voltage across the shunt, NOT caused by the output current. It has to be an extra current from somewhere. Look at the picture below:



This is a basic current source. V4 is the reference voltage for the setpoint and goes from 0 to 200mV. R1 is the external load, R2 is the shunt resistor (for 2mA range).
The interesting part is R3, which sends some extra current through the shunt, which is then missing at the output of course. The graph to the right illustrates this.

It is actually two graphs, which happen to appear as just one, but there are two different vertical scales.
1.   The voltage across the shunt resistor vs. setpoint voltage. Both go from 0 to 200mV – exactly what you see in your device.
2.   The current through the external load. It goes from -3.4mA to -1.4mA and is -2.4mA at the 100mV setpoint – again exactly what you see.

To verify this, just don’t connect any load and measure the voltage drop across the shunt again – if it equals some 3mA multiplied with the shunt resistance, then my theory is confirmed.

Since the error is about the same for all ranges, there is no need to look for a problem near the shunt resistors, rather behind the switches at the input of the OpAmp.

It could be a defect of the OpAmp input.
If there’s an analog multiplexer used for range selection, this could be faulty.
Of course it could also be just dirt and moisture forming a resistor from the positive supply rail to the inverting OpAmp input, but I take it you would have noticed that by visual inspection already…

EDIT: Sorry, the above circuit and graphs illustrate the case of a constant current error.

For the more likely proportional error, the parasitic resistor goes from the output of the current source (the OpAmp in my example circuit) to the shunt and obviously has to be quite a bit lower in value.

In this case, the error will depend on the load resistor and you'll see no voltage drop across the shunt with no load connected.

EDIT2: From your measurements, the proportional case would be even more tricky than that, since you even get negative output currents. But before I go on speculating, I'll wait what you find out and if it's actually a proportional error.

[_Wim_]

When looking at your measurements, it is clear that in the ranges up to 200mA the output current always seems to be some 2.7 ~ 3.7mA too low, for the 1xx setpoint. How is it at different setpoints? Is the error constant or proportional?

Anyway, we have some strong hints, as the voltage over the shunt resistors appears correct, but the output current is too low by either a pretty constant current or a constant percentage of the setpoint (please check what is actually applying here) regardless of the range. Okay, it looks a bit different for the ranges 500mA and above, but not by much, as it essentially just seems to be twice the current  compared to the lower ranges. The difference is, you have fully exploited these higher ranges, whereas you only used 50% of the lower ranges, so this is a hint that the error is actually proportional to the setpoint, i.e. some percentage of it, but still totally independent of the range.

Now we should try hard to find a hypothesis how something like this could happen…

The first conclusion is that it has something to do with the voltage, as this is the only constant across the ranges. Since the voltage across the shunt appears correct, it most likely equals the setpoint voltage 0 ~ 200mV. No wonder you couldn’t find anything wrong in this area.

The errors can only be explained by some additional voltage across the shunt, NOT caused by the output current. It has to be an extra current from somewhere. Look at the picture below:



This is a basic current source. V4 is the reference voltage for the setpoint and goes from 0 to 200mV. R1 is the external load, R2 is the shunt resistor (for 2mA range).
The interesting part is R3, which sends some extra current through the shunt, which is then missing at the output of course. The graph to the right illustrates this.

It is actually two graphs, which happen to appear as just one, but there are two different vertical scales.
1.   The voltage across the shunt resistor vs. setpoint voltage. Both go from 0 to 200mV – exactly what you see in your device.
2.   The current through the external load. It goes from -3.4mA to -1.4mA and is -2.4mA at the 100mV setpoint – again exactly what you see.

To verify this, just don’t connect any load and measure the voltage drop across the shunt again – if it equals some 3mA multiplied with the shunt resistance, then my theory is confirmed.

Since the error is about the same for all ranges, there is no need to look for a problem near the shunt resistors, rather behind the switches at the input of the OpAmp.

It could be a defect of the OpAmp input.
If there’s an analog multiplexer used for range selection, this could be faulty.
Of course it could also be just dirt and moisture forming a resistor from the positive supply rail to the inverting OpAmp input, but I take it you would have noticed that by visual inspection already…

EDIT: Sorry, the above circuit and graphs illustrate the case of a constant current error.

For the more likely proportional error, the parasitic resistor goes from the output of the current source (the OpAmp in my example circuit) to the shunt and obviously has to be quite a bit lower in value.

In this case, the error will depend on the load resistor and you'll see no voltage drop across the shunt with no load connected.

EDIT2: From your measurements, the proportional case would be even more tricky than that, since you even get negative output currents. But before I go on speculating, I'll wait what you find out and if it's actually a proportional error.

Many thanks again for you detailed reply. Much appreciated!

I just tested the 200mA range in steps of 10mA, and the error is indeed somewhat proportional to the setpoint (varies between 3.62mA and 4.01mA). But, while testing I also wrote down the measured value by the GS610 display. This error (deviation from setpoint) is also proportional to the setpoint (from 0.096mA to 0.587mA). If I deduct both errors from each other, I get a nearly constant error, although slightly decreasing with higher setpoints.

As to your question of dirt and moisture, I did indeed already check for this before, but did have to clean of some gunk when I got the unit. As there was worked on this unit before, I could be left over from previous repair (dried flux?), and the gunk was only in placed were some additional soldering was performed.

[_Wim_]

This is the 1A range (in steps of 100mA)

[_Wim_]

The errors above are independant of the external load. The above graphs were created with an 4 ohm load, I just tested a couple of points with a 470 ohm load, same error...

[Performa01]

As a first approximation we can assume a constant error that we should concentrate on. Once this is fixed, we can see what's left and take care of that, if necessary.

So we are indeed looking for a current source, which will most likely be some undesired resistive path to the positive supply rail.
Have you checked the voltage drop across the shunt when in 200µA range with no load connected already? It should be significant and would finally confirm my theory.

Tracking down the error will be a bit difficult on a multilayer SMT board. I'd probably do the following:

• You have already identified the OpAmp for the current regulation.
• Find the range switches - it will most likely be some analog multiplexer like (A)DG4xx or something.
• Desolder the MUX and check with an ohmmeter that there is no parasitic connection between shunts and negative OpAmp input left.
• Make a temporary wire connection between shunt and negative OpAmp input.
• Check if the error has gone. If so, you need to replace the analog MUX.
• If the error only decreases, but is not actually gone, then you need to replace the OpAmp (also).

There might be some passives in the feedback path, but I don't expect any of them causing the symptoms you're observing, since none of them will go to the positive rail.

[PA4TIM]

Is there a performance test chapter in the SM ?  I do not know the instrument but 99% of the repairs I get here (I repair test and calibration gear not supported anymore by the factory ) are without service manual and sometimes I can not even test them because they are part of an installation.

First thing I do, if possible, is checking if things work and what is the fault, often is my contact at the customer not the user. And if there is a trouble shoot section or performance check section I follow that.
Then I check things like solder joints, burned parts etc.
Then I check the powersupply with a scope. If there are separate boards I check if the rails there are OK too.

If there are adjustable parts mark them so you can return to the settings if you adjust them. Make sharp detailed pictures. If you fry a component or forgot the mounting direction you can use the pictures.

If someone already was in there before you there is a change they messed up the alignment. Often there is an certain order in making adjustments because things can depend on each other. In those cases it is very difficult to get it good again.

Then I use a scope to poke around. You can check many parts without knowing there function in the whole. Look up the datasheet and often they give test circuits to test the component. That way you can test analog switches, opamps, optos, mosfets, diodes, transistors, capacitors, resistors and many other parts. Sometimes in situ sometimes you first have to desolder them. With a breadboard, PSU an function generator you can test many components.

Besides that you can poke around them with your scope using the "standard rules", things like the voltdrop over diodes or Vbe from a transistor, check if both inputs from an opamp are equal. And things like analog switches, is the control line is high, the switch must be closed so you measure around the same voltage on out and input. Look if a an inverter indeed inverts. Very easy and basic tests.

Measure voltage over components to get an idea of the current (like knowing if the led from an opto is "on")

Also it is usefull just to probe around things like logic IC's. Look for distortion, runts etc.

Use an infrared thermometer or thermocouple to look for hotspots.

Use protection when and where it is possible. Like mounting a temporary fuse in a powerrail, I made an electronic adjustable fuse that I can use for AC and DC. Sometimes you can replace a powerrail with a lab supply that has adjustable current limiting.

Hope this helps you a bit

Fred

[_Wim_]

Have you checked the voltage drop across the shunt when in 200µA range with no load connected already? It should be significant and would finally confirm my theory.

This is not possible. When the unit does not detect any load, it turns the output immediatly off. Initially I though it detected that no banana plugs were inserted, but that is not the case, as I got the same result with banana lead with no load.

I will answer on your other questions later this evening (just started looking into this again)

[_Wim_]

Is there a performance test chapter in the SM ?  I do not know the instrument but 99% of the repairs I get here (I repair test and calibration gear not supported anymore by the factory ) are without service manual and sometimes I can not even test them because they are part of an installation.

First thing I do, if possible, is checking if things work and what is the fault, often is my contact at the customer not the user. And if there is a trouble shoot section or performance check section I follow that.

Your absolutely right, I was jumping in. Reading the performance specs again, it states that the source function should be accurate, independant of the measure function. When I checked voltages before, I compared what was indicated on the display of my unit, with my DMM. This was within spec (see below). However, according the the performance specs, the source function should go to the requested setpoint within a certain tollerance. This is also not ok (see below).

I will repeat the same test also for current generation as wel as current measurement.

[_Wim_]

The current source and measure performance...

Remark: when in voltage source mode and measuring current on the GS610, I can select auto range or manual range. By doing this I noticed that current measurement is also OK in the high current range (>0.5A) and not ok in the low current range (<0.5A). So it seems the current measurement in the low range is off only. If I look at the zero offset, it is possible a zero offset calibration was performed on this unit with the faulty current measurement (if you deduct the zero offset, the unit is almost correct, and especially on the low ranges, I am unsure if my Brymen BM869s is accurate enough to verify this...)

[_Wim_]

Then I check things like solder joints, burned parts etc.

Yes, done that. So far no luck. Spotted some sticky stuff which I cleaned off, and some previous soldering work

Then I check the powersupply with a scope. If there are separate boards I check if the rails there are OK too.

Yes, done that too. Currently I am checking every individual chip power supplies. Checked the main rails also with my scope, nothing abnormal to see.

If there are adjustable parts mark them so you can return to the settings if you adjust them. Make sharp detailed pictures. If you fry a component or forgot the mounting direction you can use the pictures.
 

No adjustable components found so far. Did take already detailed pictures. Good tips btw!

Then I use a scope to poke around. You can check many parts without knowing there function in the whole. Look up the datasheet and often they give test circuits to test the component. That way you can test analog switches, opamps, optos, mosfets, diodes, transistors, capacitors, resistors and many other parts. Sometimes in situ sometimes you first have to desolder them. With a breadboard, PSU an function generator you can test many components.
Besides that you can poke around them with your scope using the "standard rules", things like the voltdrop over diodes or Vbe from a transistor, check if both inputs from an opamp are equal. And things like analog switches, is the control line is high, the switch must be closed so you measure around the same voltage on out and input. Look if a an inverter indeed inverts. Very easy and basic tests.
Measure voltage over components to get an idea of the current (like knowing if the led from an opto is "on")
Also it is usefull just to probe around things like logic IC's. Look for distortion, runts etc.
Use an infrared thermometer or thermocouple to look for hotspots.
Use protection when and where it is possible. Like mounting a temporary fuse in a powerrail, I made an electronic adjustable fuse that I can use for AC and DC. Sometimes you can replace a powerrail with a lab supply that has adjustable current limiting.
Hope this helps you a bit

Fred

Busy with that also, but lots of stuff inside this beast… It consists of a digital board (all OK I guess), a display board (also all OK), an analog board (this is the board I am currently working on) and 2 power identical but mirror image amplifier boards (1 for current sinking, the other for current sourcing??).
These boards are stacked in a square around the heatsink, which makes it quite hard to measure them in situ… Luckily the analog board is on the top and can be accessed for probing when the unit is on (well, at least the top side of the pcb…)

[_Wim_]

Following the advice of Performa01 and PA4TIM (much appreciated), I am making progress: the 3.5mA current deviation is fixed (see updated performance results).
It was a broken opamp (U71 on analog board, an OPA2132U) that was connected to the muxes selecting the measurement resistors. I am unsure what the exact function of this opamp is, but its output was railed against the positive rail, while it was wired as a simple buffer with an input voltage of only 0.2V.

The new performance results for the current are now a lot better, but still not within spec. The results for voltage were almost not affected.
When looking at the measurements results for the current, I get the exact same measured values as before, even though the actual current is quite different. Somehow it seems there is no feedback between the current measurement and the current generation (both work independently?).
According to the manual I can perform an offset calibration (for deviations between source and measure), but this does not correct the error (you do see the unit trying). The offset calibration does work for a setpoint around zero (smaller deviations), but for bigger corrections it seems the unit has insufficient range to adjust.

Anyway, I still have not checked everything (far from!), but just wanted to post an update.