Fluke 5700a repair

[Ugur]

Hello,
The problem may not be just the semiconductor parts.
You have to be patient, yes it is not easy but you can do it if you want.

Now can you tell me what errors did you get as a result of the diag test?
Can you do the reset test?

First of all, can you remove the relays behind the front panel that I mentioned in my previous message and check the contact resistors on and off?

[veedub565]

So I have some good news finally

I found that CR5 is short circuit. I should have spotted this far sooner, it's connected back to back with another diode so when I measured it in circuit I assumed the low reading was because of this. I lifted a leg today to test it and found it short circuit.

This would mean that SCOM and ADC COM were shorted together, which would really upset the calibrator !

I put a replacement diode in (1N4148) and it has fixed the rather misleading A8 Relay error.


The voltage output is still way off though, even after warmup and zero cal it still outputs 9.997V instead of 10.00000v. A good DAC card in this calibrator produces a good 10.000v so we know there is still a problem with this DAC card !

Bit strange in the past couple of days I've found CR5 and Q23 are both faulty. They are both in the same area, and both arranged as voltage clamps/limiter. I wonder if some spike or surge got on the line. Maybe someone put a voltage up where they shouldn't.

[SometimestheDragonWins]

I think i can dig up my parts list i ordered from digikey when i replaced all the caps on the 5700a i used to have if you think it would be handy. Something like 400 bucks in caps but all from the same place.

[Ugur]

Maybe you just need a adjustment calibration.

[veedub565]

Maybe you just need a adjustment calibration.

Yes this is what I've been thinking. I think it was Kleinstein or TiN said that a 5700 should never be that far out of spec. Thing is though I've refitted all the cards back into the unit and it now passes it's self diagnostics with no errors at all. So there can't be that much wrong with it now.

If someone had done a bad cal in the past, like tried to cal without fixing the fault. Or tried to cal it using the wrong reference standards... maybe that could put it a long way out of spec. The UKAS lab at work might do a cal on it for me if I ask them nicely. I don't really want to go to all that trouble though only for it to fail.

[veedub565]

I think i can dig up my parts list i ordered from digikey when i replaced all the caps on the 5700a i used to have if you think it would be handy. Something like 400 bucks in caps but all from the same place.

I did replace all the electrolytics some time ago, but yes it might be handy to have a list. I've not done all the tants yet (I'm living dangerously) I wanted to make sure I could get it fixed first before spending too many hundreds on new parts

[Kleinstein]

From the plan it looks like the reference is part of the DAC card  (not sure about this). In this case it could still be a question of the calibration (from numbers from a cal in defect state).

The self test can do a lot of checks, but there are also limitations. So there are some possible defects the self test can not find (HW limits) or that are just not included in the software, as there are way to many ways things can go wrong and they did not consider this as a probably failure mode.

After the repair I would wonder if the zero and offsets are still correct - I would expect the described defect diode to possibly effect the zero in some ranges. The other point with a broken DAC would be that the linearity could be off - which could cause the ACAL procedure for the higher lower ranges to be possibly be wrong, but it should hardly effect the 10 V point.

[veedub565]

Yes the reference is part of the DAC card. It uses a very stable 13v reference voltage, which it can then divide or multiply as required to generate the different outputs.

I did perform a zero cal following the repair, didn't make any difference. The faulty diode seemed to be functioning as +/- voltage protection. When it went short it would have shorted two of the commons together that shouldn't have been.

I would speculate that the ADC part is working ok, when CALCHK is performed it is correctly measuring the output and saying, hey this is 1,000% drift from the cal data I have stored. It can read it's own output and it can see that it's a long way out. I've been unable to find anything wrong with this part of the circuit, or the null amplifier. So chances are the zero cal works properly too.

 I know there's nothing wrong with the 13v reference as I've checked it. Maybe something wrong with the DAC itself, it's got the right data, it's got a good 13v reference, but for some reason the output is still incorrect.

[Kleinstein]

If the reference and DAC are changes one would also need different calibration constants. The references are stable at some 13 V, but individual units may be different quite a bit, maybe up to 1% (AFAIK no official specs for the LTFLU reference).

I would not expect much to go wrong with the DAC itself to cause such a large error in the scale factor.  The DAC gain is more or less following the PWM ratio with only a small part (some 2% each - which is surprisingly much)  from the fine part and neg offset The fine part should be checked during selftest ACAL.

In theory there could be some trouble from the linearity correction part:  e.g. U38 or Q4 in the DAC not working may reach about the order of the observed error of -3 mV at 10 V nominal out.

[veedub565]

Thanks I'll investigate these area's more closely

[veedub565]

I've been having a bit of a poke around the linearity control area

U2 and U38 both replaced with new parts
Q4,5,6,7 all tested working
VR7 tested working

One slight difference is Z2 which says 14k and 11k on the schematic, but I measure it as 14k and 14k (even with U38 removed). it seems rather too precise to be a fault, possibly an error in the schematic

[Ugur]

Z2
14 Kohms from 1 to 2 pin
11 Kohms from 2 to 3 pin

Legs 2 and 3 are connected to legs 2 and 6 of U38.

Maybe there is a revision as you said!

[veedub565]

So a small update on this

Having got my 5700a to pass all it's diagnostics tests. I was lucky enough to be able to take it into the UKAS lab at work and perform an artifact cal. I left it switched on for 24hrs to warm up and stabilise. I had access to all the correct (calibrated) 10v, 10k, 1ohm standards.

The 10v part went through with no issues, and a small change of around 7ppm (which doesn't seem to account for the large PPM this instrument is out by)
The 10k part also went through, but with a rather larger change of 137ppm

The 1ohm part failed though with the following error message - "Couldn't Ext Cal 1 Ohm (Ext Meas) (2582)".

[veedub565]

Had another couple of tries at this today and got the same error "Couldn't Ext Cal 1 Ohm (Ext Meas) (2582)" and also "Couldn't Ext Cal 1.9 Ohhm (Int Meas) (2587)"

I also ran the Cal_Chk report again, now that I've got it to pass it's self diagnostic routine, results attached.

Quite interesting to note a couple of things

1. It doesn't look like this unit has been calibrated since 2006, that's a lot of years for this to have drifted a looong way out of spec. Plus recent change of resistors and capacitors and trauma of being opened up and worked on.

2. The full scale shift on 11v range is -193.91ppm or  -0.002133v. Well a shift of -0.002133v corresponds nicely to the 9.997v I'm reading when the unit is set to 10v. When running the artifact cal with a lab standard, it showed a reference adjustment of 6.5v=7.2ppm and 13v=5.6ppm ?


 I'm beginning to think this just needs a good calibration now... except it fails artifact cal on the 1ohm part, and I've tried it 3 or 4 times now, in a lab, with the proper calibrated lab standards.

Watching one of the XDev video's and their calibrator was even further out than mine is, reading 9.6v when set to 10v. So also kind of makes me think mine just needs a good cal... if I could get it to pass the 1 ohm. 

[veedub565]

Eugh dirty relay contacts. The contacts inside the relay aren't any better, worse if anything. This isn't going to help matters !

I switched the relays around and did the artifact cal again. It still failed 1ohm, but for 10v and 10k the ppm changes were drastically different, like hundreds of ppm. So probably these relays need changing.

Potter and Brumfield (now TE) R10 E6297-3 relays, custom made for Fluke I think, gonna be hard to get hold of.

[TheDefpom]

I have had good success removing the relay covers and cleaning them with deoxit gold on a stip of paper placed between the contacts and worked backwards and forwards a dozen times, I have done this a few times now on different equipment and it has worked every time, and continues to work to this day (they haven't failed again yet).

[veedub565]

Another update on the long running repair of this Fluke 5700a, and I've resorted to sending it to Fluke UK for investigation.

After the latest issue with it failing the 1ohm and 1.9ohm part of artifact calibration, I checked out the A9 Ohms cal and A10 Ohms main boards. I couldn't find anything wrong though, and to be honest was a bit lost at this point for where to look.

I contacted Fluke and surprisingly they agreed to look at it for me. I'm surprised firstly that they would even deal with me as a private customer not a business. And secondly these units are 2yrs out of support now. They quoted me a standard investigation fee, I won't say what it was but it was hundreds of £££. And any technician time to repair a fault would be added on top of that. I was a bit worried this may spiral out of control !. However as a minimum for the cost of the investigation fee they would test all the boards and be able to tell me what was good, and what was faulty.

Anyway I dropped it off, and a couple of weeks later they contacted me to say all boards tested. 1ohm fail caused by A3 Motherboard, 1.9ohm fail caused by A8 Switch Matrix. Additionally the 3 big relays behind the front panel were swapped out with new ones, and no change was observed so the original ones were refitted. They did not have any replacement boards available, and seemed to think there wasn't much further they could progress with this.

I did try and buy some spare relays from them, but it seems even Fluke is having difficulty getting hold of them. So I accepted their findings, and surprisingly Fluke dropped the investigation fee and just charged me for the technician time which was substantially less.

I have to say the service from Fluke has been excellent, and something a certain other large test equipment manufacturer could learn from. The only thing I was a bit bummed about was I sent the unit in to them in a proper FLuke Calibration box, with handles,and the proper foam inserts. As standard practise they binned it, and returned the unit to me in a large generic box, with no handles (making it awkward to manage) and (badly) packed with expanding foam 


So armed with this new information that A3 and A8 are the source of the latest faults, I did a bit of fault finding on A8. lo and behold CR13 is short circuit. Clearly wasn't causing any self test errors, but perhaps contributing to this new fault in some way ?. Looking at the schematic it's related in some way to "Out Lo", "Out/Sense Lo", "PA Com", "OSC Sense Lo", "Int Sense Lo"  . Looks like CR12 and CR13 are protection diodes limiting the voltage to +/-0.6v so CR13 short, I'm not sure what effect that would have. I've ordered a new part anyway, so will try it and see.   

 

[veedub565]

Another small update. Replacing CR13 on the A8 Switch Matrix didn't fix anything, but I didn't really expect it too. It was faulty and needed replacing anyway.

I took a look at the A3 Motherboard, and I found some burnt out tracks. This isn't good news, means a lot more work now to cut out the burnt bits and repair tracks. From what I can work out the following tracks have burnt open

K13 Pin 11 to K6 pins 6/11
K13 Pin 9 to K7 pins 13/4 (this trace also goes to K6 & K8 pins 13/4)

Also burnt out areas at J201 (A8 Switch Matrix) B FB and J611(HV Control) also B FB

Bit tricky to work out what's happened, but seems to relate to SCOM, OSC SENS LO, OSC LO GND, and OUT LO (front panel binding post), and B FB line.

The diodes that I've found to be S/C on A11, and A8 seem to be for voltage limiting / protection diodes.

Trying to piece together what's happened to this unit, maybe someone put too much voltage or current up the OUT LO binding post.... or maybe something attached like external amplifier unit because B FB line seems to be related to that function.

[veedub565]

Well it's been a while, but I'm still working on this on and off. I've got so many other projects on a the go and work commitments.

After stripping the 5700 down to remove the A3 motherboard, I salvaged an old LED light panel to use as a kind of lightbox. It's great to be able to put the board on and see right through it, and see where the damage is.

Some of the damage was under the motherboard relays, so I removed these to be able to see more clearly, and also I was able to test them out of circuit to verify they work. Is there a datasheet or pinout for the custom COTO relays as I'd like to test these too.

I also tested RV1 Varistor and it measured 5ohms, which seems a bit low. I tried applying some voltage to it, and found it was already drawing 500mA at only 2v so I think this is probably faulty. Also the two tracks leading to this component are both burnt out! presumably any standard 22v varistor replacement would work here.

Finally got around to being brave and getting the dremel out and cutting into the A3 motherboard PCB to cut  out the bad bits. I've never done this before and I was rather nervous, but hey it's broken anyway so nothing to lose right. I was quite pleased with how it went and has turned out. Now to epoxy repair the PCB and then the really hard part of linking all the broken tracks and getting it right !

[veedub565]

Ok so I have another update for this long running Fluke 5700a repair. I've been a bit sidetracked with other projects lately, mostly I've been working on a HP 8341B and setting up my own 8510C VNA system.

Anyway, the 5700A. I think last update I'd sent it to Fluke UK who tested all the boards and reported back that A8 Switch matrix and A3 Motherboard had a problem. Problem with A8 turned out to be a couple of blown protection diodes, which I replaced. And the motherboard turned out to have some burnt out tracks. 

I finally got around to cutting out the burnt tracks on the motherboard and soldering in some wire links. It looks like the tracks from RV1 to K7 and K13 had burnt out, RV1 itself was blown, the "B FB" track had burnt out in 3 places between J611 High Voltage Control and J201 Switch Matrix. And "PA Com" had burnt out between JP1 E16 and J611 High Voltage Control. I tested all the A3 motherboard diodes and relays and all tested ok.

I've no idea what could have caused all this damage, I can only suspect something external was put up somewhere it shouldn't have gone. especially as Fluke confirmed all the other cards were ok.


So the latest update, After I did all the wire links I fitted minimum amount of cards for it to work (DCV only) and tested the output.... still 9.997v Hmm, I performed a "zero", now previously this would perform straight away as soon as the instrument was switched on. Now however it says unable to perform zero calibration. Hmm, that's different, so i left it switched on for 30mins to warm up and tried again. And it went through ok. I suspect with PA COM open circuit it wasn't referencing anything properly, probably just seeing some fixed value. now it's seeing true value and takes time to warm up and stabilise.

I tested the output again... still 9.997v Damn !

Then I remembered that the DAC card in this unit isn't the original one. The original DAC card was faulty and swapped with a working DAC from another (broken) unit. In hindsight this was a bad move, and I should have repaired the original. Anyway, it probably means that the original calibration data is a long way off and pretty worthless now.

I decided to do a rough and ready DCV calibration using a Time Electronics DCV source. I know this isn't really good enough source, heck I didn't even have the case fitted to the Fluke, it was still open. I just wanted to see if it did actually adjust the output to be more accurate, or at least did something.

DC source was verified at a stable 10.0003v using Datron 1281 DMM. Fluke 5700a DCV calibration completed succesfully, biggest reported change was some 5,900% on 220mV range (this concurs with the thousands of % error I saw in the self test report).

 I tested the output..... and it's now 10.0001048v

Thats much better,  error from 10v is now 0.0001v instead of 0.003v. It's not perfect, but remember this is a very quick and dirty cal.

So this is a positive development, Fluke say all the other cards are ok, but I think I should still double check the High Voltage card.

I'm lucky I have access to a UKAS calibration lab, so I am able to perform a proper cal on this now against primary standards. Fingers crossed I now have a working FLuke 5700A Calibrator !

[veedub565]

Bah ! well I thought it was all fixed 

24hrs after I did the ACAL I ran Cal Check Report and found that the 220v DC range was some 300% of spec, other ranges didn't look toooo bad, although 220mV range is 60% of 12 month spec, that's not good either.. I ran Cal Check again after 48hrs and 220v DC range was now some 800% of (12 month) spec. Additionally some other ranges had shifted quite a bit too now.

I measured the DC outputs with 8.5 digit DMM.
1,000v = 1000.00v
10v = 10.000001v
1v = 1.000013v
100v= 98.7v

Clearly a problem with 220v range


So, a bit further along but still not happy. Going to be hard now because there's no obvious fault, it's essentially working but drifting out of spec.  Suspects at the moment are the switch matrix and power amp assy. Following the troubleshooting doesn't reveal a problem with either 

Also wondering if I left it long enough to warm up prior to conducting ACAL. Could be 11V reference wasn't up to temp and stable enough, and alot of other DC ranges depend upon that. Any small drift of the 11V ref would be magnified by the power amp assy. Just a thought

[veedub565]

Time for another update

I'm not sure how this happened, or how I missed it before. Anyway, K17 on the A8 Switch Matrix was fitted backwards, i.e. it was reset when it should have been set and visa-versa.  Not sure what effect this would have had, but it can't be good.

I refitted it the correct way, and made some checks to confirm it was operating as it should be. I was hoping that might be the final piece of the puzzle... HAHA no chance ! I now get a new error when performing artifact cal or cal_chk

 "unable to cal fine tune (14bit) DAC (2520)"


There's very little information in the manual about the 14bit DAC, and no information at all about error (2520). The output of the 14bit DAC is checked during the self test and this passes ok, so one would assume this is working properly.

Speaking to the team at XDevs even they aren't really sure on this one, but  it seems this is some issue with the AC calibration constants. Given all the repair work on this unit, and the fact it has a replacement A11 DAC card fitted it's not surprising if the original factory cal constants are causing a problem. Next step is to backup the CALROM and format to factory default.

Interestingly I was speaking to an ex Fluke engineer about this, he remembers working on 5700A and has apparently seen this error before. He told me that replacing the A11 DAC card fixed the problem. Interesting because A11 was already replaced on this unit, shame I don't have a spare working A11 DAC to try, or the original one to put back in.

[veedub565]

Picture here from October 2021 shows A8 switch Matrix, and you can just about see K17 top right of picture... and it is backwards compared to every other picture I've seen of A8. Pretty sure at that point I had only just replaced all the CC resistors. I mean that is backwards right ? because this new error has only occurred after I switched it around.

https://www.flickr.com/photos/193832816@N04/51648147365/