Fluke 8845A (like 8846A) 6.5 digit Multimeter showing "overload" at all times

[t-17]

Hi all,

I have here a Fluke 8845A (6.5 digit bench multimeter) in good cosmetic shape, with a blown 440mA input fuse and probably some defects in the ADC. I do not know what happened to it earlier. However, I can see from the flux residue on the PCB that someone was already in the unit and apparently, changed rectifier diodes for the 3.3V rail of the analog part.

The unit is built around two Altera FPGAs, a LM399 reference, some Fluke ceramic resistor networks (I guess from their looks), as well as some fancy Opamps/ comparators by LT and AD, and a RMS voltage converter AD637.

What I have checked so far:

• All voltages are fine on the digital part (+1.5V for the FPGA core and +3.3V for I/O I guess, as well as +60V for the VFD).
• All voltages in the analog part are also fine (+1.5V again for the FPGA core, +3.3V I/O, symmetric +/- 5V, +/- 15V, +/- 20V)
• All clocks are running.
• Optical link between digital/analog part is working (checked by inserting a piece of paper inbetween-- Fluke is not using an integrated opto-coupler but went with a IR LED/LDR combination of some sort). When a piece of paper is inserted, the "measurement update" indicator in the display stops blinking.
• At least the LAN interface is working fine.
• Display and front panel in general are working fine.
• No errors are reported. However, the unit does not appear to have any reasonable self-test (checked the manual, did not find any).
• Unit boots up normally.

Observations:

• The 1.5V regulator for the analog part is getting very hot, but that is probably due to the fact that it's being designed in without a heatsink, and it generates 1.5V from the 4.8V winding of the main transformer.
• The Analog Devices RMS converter is also getting somewhat hot.
• There are slight burn marks on the PCB near the 3.3V rectifier diodes on the PCB, which are presumably used for generating the FPGA I/O voltage. The diodes have been replaced from what I can tell, and the rail is stable.
• The unit reports "overload" no matter what mode is selected (ACV, DCV, ACI, DCI, ...)
• When in continuity/ diode mode, it reports always "OPEN".

Any hints are very much appreciated. I'll report new findings into this thread. If it helps, I can also post images.

Side note
I have decided to start a new thread for this nice multimeter. I have already posted in another thread about the same piece of equipment, but this was to offer help for an entirely different problem (bad display): https://www.eevblog.com/forum/repair/fluke-8845a-repair/msg979347/#msg979347

[ManateeMafia]

Have you tried manual range selection? Also, single trigger mode vs continuous trigger.

I would start with 10v dc range and try to follow a test input signal or short the inputs. It could be a defective op-amp and that can be verified by checking inputs and outputs. Not easy without schematics.

Good luck.

[t-17]

Yes, I have already tried manual range selection. It works, but the result is the same. Single trigger works as well, but this will be mostly useful for checking the opamps. There are also a lot of analog switches/ multiplexers in there. I'll have to check them, too.

I'll try to check all of the ICs in the 10V range. This will be my next step.

If anybody could provide me with voltages/ signals at certain test points of the main pcb of a working unit, please contact me. This would help me a lot.

[Kleinstein]

There are two relatively easy to check points:
The reference voltages (should have test-points for the 7.x and maybe 10-12 V).
The current flowing in the ohms range.

Without a good selftest and schematics, the repair might get difficult. At least it take quite some time to draw a rough schematics from the board. If there is something wrong with the FPGA program, you might be lost getting replacements.

[t-17]

The reference voltages are all okay. (+/-7V and there is one +10V test point, also okay).
I have not measured the current flowing in the ohms range yet. I wanted to concentrate on one problem at a time, but maybe I'm missing something. Can you elaborate?

I'm making (slow) progress in tracing the signal. There is one opamp I'll have another look at. It's showing strange voltages when I apply 4V DC at the inputs in the 10V range (all opamps before are fine). I'm also trying to figure out how a multi-slope integrating ADC works. There are 4 transistors and 4 voltages regulators all 3-pin SOT-343 or something similar with various voltages at the pins. The strange thing is, although there are various positive voltages which appear to make sense (like +20V, +10V) on the negative side it's all off and I measure strange values. I suspect something is wrong there, too.

What I am pretty sure of is that the FPGA is ok. There are active digital signals between it and its SRAM and at the optocoupler. Also, the integrator is showing activity when I manually trigger it or leave it free-running. I am 99% sure it's a problem in the analog part.

For now, I have a suspicious looking opamp to investigate, so I'm quite satisfied for the moment

For reference, if somebody ever has this thing on the table (note: based on observations/ measurements, 99% certain):
DCV range selection/ input divider (by 1, 10, 100, 1000): Z1/U37 (ceramic laser-trimmed resistor network/ multiplexer DG444)
1st opamp: U36 (LF353)
2nd opamp: U35 (AD8510) (either dead in my unit, or has a special purpose-- strange voltages, need to double-check)
integrator: C131/U49 (AD8510)
comparator: U50 (LT1394)
FPGA: ALTERA Cyclone I EP1C12 (FPGA used in the analog part of the PCB)

I'll be updating this with new findings. Hopefully, this is useful to somebody one day.

[Kleinstein]

Measuring the Ohms current indirectly does a kind of check in the reference voltage. No real need to measure if the ref. voltage are there.

For the typical mulit-slope ADCs, there also should be a -10 V or similar reference. Just using +-7 V would be unusual, though not impossible. Strange negative voltages might be a problem. Usually many of the OPs should get something like +-15 V.

The usual configuration is an switchable 1:100 divider for the input- switching via relay of high voltage MOSFETs. Than the input amplifier with amplification of 1, 10  and often 100 - switching via DG444 would make sense. One should be able to test this input part with low frequency AC.

[t-17]

I made an interesting discovery. Two opamps (U35: AD8510 and U36: LF353) have supply voltages which change and shift with the input voltage applied to the input terminals. Before I continue with the search for the root cause, is there any application where it makes even remotely sense to design opamps in with this kind of behavior (variable supply voltages)? I don't think so. I suspect there is a problem with the supply. If I leave the input terminals unconnected, these opamps have about -5.5V on the negative terminal and around 5.4V on the positive terminal. With 4V DC on the input terminals, I have -1.5V and ~9.3V. When I disconnect the inputs, the voltages seems to undershoot a bit (V- for example then has -5.8V slowly coming closer to -5.5V, what is reached after half a minute or so). All other opamps in this section are all supplied with +/-15V.

If I measure the resistance to the voltage regulators (from the opamps supply pin to the voltage regulator), I measure about 8Mohms what does not make any sense to me. The connection between the opamps (the two nets V+ and V-) are low-impedance and seem to be intact. Is it possible that some trace or component has gone bad for some reason and the two opamps now do not have a proper supply any more?

I traced these nets to two transistors (Q31 and Q32) in the vicinity which have one pin directly connected to -20V (Q32) and +20V (Q31) and another pin to the V+ net or V- net of the opamps, respectively. I don't know yet what the third pin is for, but maybe this topology makes ring a bell for somebody. Why would these opamps have their rails switched by transistors? Does this make any sense?

I'll keep investigating this thing for the time being.

Thanks so far for your replies! They already helped me a lot!

[Kleinstein]

A supply voltage the follows the input voltage makes some sense, especially if the input voltage range is large compared to the supply of the OP. So it's less a problem with a LF353, but could still be used to reduce input bias / increase the input impedance.

The Keithley 2000 DMM uses this way in the input stage with an LTC1050 AZ OP. There two transistors as constant current sources from a high supply e.g. +-20 V provide current. 2 Zener diodes limit / stabilize the supply of the OP, an the center is dirven by a second OP.

[t-17]

I believe I have finally found something.

TP25 is at about 270mV in my unit. It is driven by an opamp and I believe it should be zero volts. I measured this with manual triggering on while the unit was idle, range 10VDC (manual).

Can somebody verify this with another 8845A/8846A unit?

[voltsandjolts]

Can somebody verify this with another 8845A/8846A unit?

Let me know if you still need this info, I can help this week.
Edit: I assume my 8846 and your 8845 will be similar enough that TP25 is the same point!

[t-17]

Hi,

thanks for the reply! I'd appreciate if you could provide me with this info. Also, when in 10VDC with auto-trigger on (default), I could make use of the waveforms visible at TP28 and TP20.

Thank you in advance!

[voltsandjolts]

Input shorted for all tests.

Manual trigger (idle), 10VDC range (manual):
TP25 +80uV (effectively 0V as you suspected it should be)
TP20 3.636V
TP28 -70uV (effectively 0V)
All clean and stable DC

When you press MAN trig button there is a burst of activity on these test points, similar to AUTO mode scope captures below.

Auto trigger, 10VDC range (manual):
See attached pics
CH1=TP20
CH2=TP25
CH3=TP28

TP28 seems to provide negative pulses which start/stop the acquisition cycle.
TP20 square wave is around 15kHz with varying duty cycle


I will keep the meter open for a couple of days if you need some more info.

[t-17]

Thanks a lot, voltsandjolts, this is very helpful! My waveforms look totally different and I suspect that either the CMOS switch (U57) or the opamp (U49) is bad. I'll report back once I know more. Thanks again for your efforts!

[t-17]

Could you provide the waveforms also with 4VDC at the inputs? (This is the voltage I have used so far for testing; I chose this value in particular because it is not present anywhere inside the unit.)

[TiN]

Those are multislope integrator waveform (triangle ramps) and it's comparator output, which is counted by digital logic into code. It will look similar at any input voltage, just "duty cycle" would vary differently.

If you have no ramps, you need to check charge control circuitry which connects various ratios of VREF to integrator. If no digital counter output, but charge ramps there - check comparators.   

[voltsandjolts]

4.00VDC input for all tests.

Manual trigger (idle), 10VDC range (manual):
TP25 +10uV (effectively 0V)
TP20 3.614V (effectively same as before)
TP28 4.00V
All clean and stable DC

Auto trigger, 10VDC range (manual):
See attached pics
CH1=TP20
CH2=TP25
CH3=TP28

TP28 seems to provide negative pulses which start/stop the acquisition cycle. Not sure about that now! Edit: Looks more like the input voltage to be measured which is periodically grounded to cancel internal offsets - differential measurement (?)

[t-17]

Thanks again! This will be all I need for quite a while. I'll now turn back to my TODO list and report back as soon as I find out something new. I will not need any more measurements/ waveforms for the time being, so feel free to put the cover on again. Your help is very much appreciated! 

[voltsandjolts]

Glad to help, good luck with the fault finding.
With all the 'clone' manufacturers around it's no wonder companies don't release schematics anymore but it is a PITA.

[TiN]

Often in bench meters there is autozero function used to remove offset errors on measurements. What it does is simply make measurement with A/D input grounded, then make actual input signal measurement. User's manual should have information how this  works in 884X exactly.

[t-17]

Hi TiN,

sorry for replying so late, but I have indeed overread one of your posts with valuable information. The comparator works-- I have checked the input and output waveforms and they look fine. However, I will replace the opamp used in the integrator, since it has a very large offset (>200mV) when it should be zero and have a look at the waveforms I see at TP25 (Ch1-- attached to this post). Apart from these zero drops I also have irregular (about 3-6 times a second or less) anomalies (the measurement in the attachment shows it horizontally in the center, as I've managed to let my scope trigger on it).

Feel free to comment if this makes ring a bell!

Thank you for your contributions!

[t-17]

I want to share something with you guys. After months of intensive troubleshooting and learning and most importantly, thanks to your help, the instrument is finally back to life! The culprit was indeed the AD8510 at the heart of the instrument. After replacing it, I had immediately reasonable readings. Of course, the unit is out of calibration now, but for me, this is still a huge success.

This is what I've done (please comment if I missed something important):
I've covered everything around the opamp carefully with Kapton tape in order to protect the parts from the heat of my hot air tool. I desoldered the old opamp, cleaned the pads carefully and used "no-clean" solder paste and again the hot air tool to solder the new replacement opamp. I did it this way in order to minimize stress to the PCB and to avoid flux residue. From my understanding, this is probably the most sensitive part of the PCB.

I came back to leave some images and once again say THANK YOU!

The first image shows the unit measuring the 4V DC of my power supply and the second image shows the "problem area". So far, all other modes of operation are working fine as well, so the blown input fuse and the dead opamp seem to be the only problems.

The thread can be marked as "solved" now, I guess

[ManateeMafia]

Congratulations on the repair and thanks for sharing your results.

[voltsandjolts]

  Nice result!

In UK my 8846A is around GBP180 for Fluke calibration or GBP70 for RS cal.
However, the Fluke report is more comprehensive.

[t-17]

My last post was a bit premature, I have to admit. The multimeter had more problems, which I discovered one by one after the central part of the multimeter (the A/D converter) was working again.

The entire ACV/ACI measurement was gone, but came back after replacing U18 (AD673) and U20 (AD8510). This was almost "easy" to debug and solve (RMS output pin was constantly at +15V, and after replacing U18 it did not react to input changes so I traced the input of U18 to the output of U20), but now I'm kind of stuck again with the so far only remaining problem.

Which is: frequency and period measurements. I have found that U16 (LM393) is a dual comparator, which compares *some* AC signal against +0.5V and -0.5V. Its outputs are connected together (logical OR, as they are open collector according to the datasheet). The output measures constantly at 3.3V which is the FPGA I/O voltage. Since this is the only comparator in the vicinity of the entire AC circuitry and I know that it is related since it does the range switches when the signal amplitude changes (and also the manual says somewhere the AC circuitry is used), I suspect that either another opamp or a multiplexer is bad (assuming no damage in passive components as the board looks absolutely clean without any traces of heat and all resistors and capacitors showed reasonable values as far as I could probe them in circuit). AC voltage and current measurements work fine now in addition to the previously repaired DCV/DCI part. I have tested all ranges. Continuity, diode and ohms work fine, too.

Could anybody provide me with waveforms of the comparator IC (U16) pins (output: 1 or 7 and input: 2,3 and 5,6) when measuring an AC signal (1Vpp and >5Vpp)? This would help me a lot.

[TiN]

Good job on repair, but..

The thread can be marked as "solved" now,

It's far from it, my young padavan  . I'd consider solved when meter fully working and calibrated .

Also using hot air gun for SO8 package is sorta overkill in my book. Usually better and safer technique is to put solder blob on one side (4 legs) and melt whole side with chisel tip. Then melt second side same way. And flip side quickly to other side back and forth. After 3-5 seconds opamp is flowing and can be removed easily by tweezer It sounds much complicated than actually is. Try on some dummy boards. This way you don't risk overheating nearest parts..