Fluke 5700a repair

[veedub565]

A18 is finally completed, woohoo. All electrolytic capacitors replaced with new, higher spec ones. All carbon comp resistors replaced with properly sized metal film replacements (thick film for the 220k 3W ones). Only slight irritation is some of the smaller caps I ordered with 5mm pitch, actually arrived with a 2.5mm pitch and the legs bent out. So they won't sit flat to the PCB 

[/url]20210909_152224 by Ben Tocknell, on Flickr[/img]

On the solder side you can see some small burning of the PCB, I got to this one in time before more damage was done.

[/url]20210909_152247 by Ben Tocknell, on Flickr[/img]


Still can't get the images to embed, hopefully you can see them via the link.  I'm using the bbcode and "insert image" button from Flickr, it should work.

[picburner]

You have forgotten four carbon composite resistors, the ones highlighted in blue.

[veedub565]

No, I haven't forgotten them. Those are Carbon Film resistors not Carbon Composite, according to the parts list in the service manual anyway.

[veedub565]

Well it's one step forwards and two steps back with this at the moment. The replacement GPIB connector I ordered arrived, I couldn't source the OEM part so I used an alternative. This came with the wire latch which had to be removed. So that is all done and fitted now.

The I/O board is a bit of a PITA to remove, so before I refitted it I took the opportunity to test the relays. This unit has a relay fault somewhere!. Now the service manual says...

"Check individual relays by connecting an external power supply (a typical 9V battery
works well) across the set coil to place the relay in the set position or the reset coil to
place the relay in the reset position."

I used this method to test the relays on the motherboard, and it worked ok. So I applied the same method to testing the relays on the I/O board (can you see where this is going yet?)

K8 checked out ok, so I moved on to test K7...bang!!. A visual inspection shows it has blown the sh1t out of the relay driver U16, literally blown a hole in the middle of it. Looking at the schematic I'm not sure how this is possible?

So I need to source and replace U16, maybe U14 too as a precaution. Yet more money on parts! I'm reluctant to even attempt this again on other relays, it seems bad advice really. Repairing the I/O board is not too bad compared to some other boards in this unit.



By the way, attaching a 9v battery to test the relay on the A18 board did not work, it didn't activate the relay. Some investigations shows diode CR244 to  be s/c both ways, ho hum. So the question, is this a result of connecting a 9v battery across the relay coil with the wrong polarity... or was it like this to begin with. One of the original faults was error 3107 which relates to the DC HV amplifier. And this relay is related to that part of the circuit. I've either stumbled across one of the original faults here.... or blown something else up by attaching a battery like FLuke said to do.

[TiN]

Well in battle between common sense and manual, common sense wins  . Perhaps back in 80ies 9V batteries were not as advanced and powerful? I used SMU with current limit 100mA and 5V voltage (no idea why manual recommends 9V for 5V relay test). But most of the times I wouldn't bother testing relays and just swap the suspect ones (based on failing function/mode and circuit location). Obviously it takes longer time to figure out topology and how everything is related.

This thread reminds me of Hulk-3 project, which still have dead A18 before anything else can happen with it. 

[veedub565]

Yes I placed too much faith in the service manual I think here. I agree it is strange why they recommend 9v battery for 5v relay test. Your method sounds better, using SMU 5v with 100mA current limit. Less chance to do any damage that way. Using 9v battery seems quite a nice way to blow things up. Unfortunately I'm only learning this now 

I think I will forget about testing relays for now. It may not even be a relay that is faulty, could be a faulty relay driver or some other related problem. Like you said learning topology may be best, see what function doesn't work and track it back.

Lot's of bedtime reading to learn topology, I might have to make some large scale printouts.

[picburner]

In my 5700a I had an error 3409 (Phase lock loop fault) and I also checked all the relays of the A13 board without taking them apart.
I used a 5V smu with current limitation of course and I also paid attention to its polarity to avoid any damage to the connected relays driver and other circuit on the board.
The relays all worked well but their driver, a ucn5801a, showed no sign of life: none of its 8 sections worked.
For further confirmation I disassembled it and did a vector test with my eprom programmer: device completely dead.
Once the ucn5801a was replaced with a new one, the error disappeared.
We must be very careful when making tests of this type: it is absolutely necessary to avoid creating more failures than those already present, otherwise the repair bill goes up....
P.S. common sense is often better than what is written in the service manual (also for the resistors I showed you before: they are all carbon compounds, trust me!)

[veedub565]

I neglected to pay attention to polarity here which is probably what has blown the driver on A21 board. At least this mistake is repairable, but yes it puts up the repair bill.

The service manual lists those resistors as carbon film,  I looked up the part number listed (EB4741) though and yes it appears to be composite. Maybe Fluke made a typo, wouldn't be the first time that's happened.


This is certainly a learning experience !

[TiN]

Yep, I swapped those vertical resistors too. Round bobbin resistor = shoot on sight 

[picburner]

In my A18 board I have not changed any capacitors or resistors.
All are perfectly within their respective tolerances, I measured all one by one (even if on board) with my ESI Videobrige 2160 rcl bridge.
Yet the construction date of the instrument dates back to 1990 based on what is written on the integrated circuits.
Probably the instrument had a few working hours but I will never be able to know for sure because the CR2477 battery was completely empty then the ETIME? command returned an unreal value: 134214657.
Probably, in few years, I will have to replace all this components me too if I don't sell the instrument meantime....

[TiN]

picburner is brave man . I have my MFCs 24/7 unattended, so peace of sleep is paramount here 

ETIME? may be reset once PROM is formatted/replaced/corrupted. So it's not a protected counter to rely on.

[veedub565]

Another update

A18 board is now completed (again) carbon comp resistors I missed last time have now been replaced with suitable metal film. A7, A8, A10, A15, A16, A17, A19, A20, A21 board are also all completed with new caps and resistors. WOW there is some dust and dirt on these boards! So all have been cleaned. 

Broken GPIB connector on A21 board was replaced. I couldn't find the exact Amphenol part, so an alternative was used. This one had the bailey lock, but it was possible to drill out the rivets holding it and then left with normal connector. Blown relay driver was also replaced.

Backup battery on CPU board measures 3v, date/time are correct, but it may be worthwhile replacing anyway as I don't know when it was last replaced.



Unit was reassembled with following boards - A21, A20, A19, A18, A17 (and of course motherboard)unit powers up and works as expected with only limited boards attached. I made some voltage measurements for A19, A18, A17 boards (results attached) As you can see most of these are within tolerance, but a couple are out. In particular +5LHR and -5LHR warrant further investigation. Ripple on both rails is is 2.66v p-p (50hz)

I can see why it is a good idea to test these supply rails first before connecting any other boards. A bad supply rail could damage the other boards. This is why swapping cards willy nilly from a working unit into a faulty unit is a nice way to blow up working boards.

[TiN]

I didn't bother replacing coin battery on A20, it's not powering anything critical.

I like to use scope to poke onto voltage rails, to see for ripple/noise/etc. Also I remember that some labels are misleading a bit, with test point on non-rectified secondary voltages (where you will see "ripple") versus regulated DC rails.

[veedub565]

Yes the labels are quite misleading, for example +5LHR and -5LHR are actually meant to be 12v not 5v like you would think. In this case they are measuring more like 18v which is quite long way from what it should be. 2.6v ripple is within spec for the positive rail (3v) and slightly out for the negative rail (2v). Seems odd the negative rail would have a tighter spec for this than the positive rail.

Looks like +5LHR and -5LHR are derived from "5 LH AC1" and "5 LH AC2". So effectively +5LHR/-5LHR come straight from the transformer, via a little rectification and smoothing. So there doesn't appear to be any possibility to adjust the voltage, it is what it is. Strange then it's out of spec.

I use a scope to check for ripple too, haven't checked all of them yet.


One other thing I noticed, is that on my A16 board C32, C34, C31, C33 electrolytic capacitors are all Orange drop capacitors instead. It looks factory fit, so I've left them.

[TiN]

Be careful not to blow up scope, as many rails have isolated grounds so you can easily catch some hundreds volts pk-pk between test points.

[veedub565]

Yes I'm wary of things like this with floating earths. Sometimes best not to connect the scope ground at all, or even disconnect earth from the scope mains plug. Either that or I'll have to invest in a differential probe.

[alm]

Please don't disconnect the scope from ground and then go poking in an instrument that contains lethal voltages. Does the 5700A service manual contain resuscitation instructions like manuals of their previous calibrators?

If you don't have a real differential probe, the safe thing to do is use two probes on two channels in subtract mode and leave off their ground leads. Common mode rejection is not great, though.

[veedub565]

Please don't disconnect the scope from ground and then go poking in an instrument that contains lethal voltages. Does the 5700A service manual contain resuscitation instructions like manuals of their previous calibrators?

If you don't have a real differential probe, the safe thing to do is use two probes on two channels in subtract mode and leave off their ground leads. Common mode rejection is not great, though.

No it's not the best plan, although we have done it on occasion at work for certain instruments (can't remember the reasons why). Biggest issue is that the BNC's on the scope, or metal parts of the probe itself. could float at hundreds of volts potential relative to an actual earth. Fluke do still put the resus instructions in the manual.

Like you said, CMRR isn't great with two probes. And a differential probe may end up costing more than the scope (TDS3054)

[TiN]

HV Probe like EEVBlog HVP70 is perfect match for jobs like troubleshooting +/-44V or higher 5700A supply rails

[veedub565]

I got some information out of this unit finally (attached). Formatting is not the best, but readable. Unit was left to warm up for 24hrs, ambient conditions not tightly controlled but pretty constant 21deg and 58%.

Many results are ok, some have drifted a long way.

[veedub565]

Anyway, update on progress

All boards are now completed with new electrolytic capacitors, carbon comps replaced with suitable metal film.

DC voltages check out ok except:-
                           +5LHR and -5LHR unregulated supply is slightly high. Since this is derived straight from the xformer I'm not sure anything can be done to correct this.
                            +44S / -44S and +17S / -17S regulated rails slightly high

I would be interested to see what others measure on these voltage rails, and be able to compare to working unit. (I did have access to a working unit, but not anymore unfortunately. )

Awaiting a new differential probe so as to safely check AC ripple p-p

Self Diagnostic routine throws up the same errors as before, I tested this with limited cards fitted as follows:-
A20+A19+A18+A17+A16+A7+A8+A11+A12+A13 = A8 Relay Fault (four times)
A20+A19+A18+A17+A16+A7+A8+A11+A12+A13+A14+A9+A10 = no further errors
A20+A19+A18+A17+A16+A7+A8+A11+A12+A13+A14+A9+A10+A15 = HV Offset Fault (3107)
A20+A19+A18+A17+A16+A7+A8+A11+A12+A13+A14+A9+A10+A15+A5+A6 = 40db Attenuator fault (3918)

I did a few tests using a Datron 1281 (last calibrated by Fluke July 2020) *edit* I forgot to zero the DMM, updated measurement values below

5700A                 1281
10.000000VDC = 9.9986185VDC
9.999150ohm =   10.380749ohm (2w comp off) 10.023993 (2w comp on)



So plenty of work still to do to get this calibrator back working properly.

[ManateeMafia]

Here are some measurements that I made on a 5700A. The repair is incomplete, hence the missing values.
The ripple measurements are comparative and made with my multimeter. I just wanted to make sure the replacement parts didn't make things worse.

I have found that making these spreadsheets are easier than writing them down.

Good luck on your progress. You will find these are high maintenance devices. Relays will be your biggest problem, especially the ones behind the front panel. They show up when you run artifact calibration and the 1 ohm fails.

[veedub565]

Thanks, I appreciate you sharing that for me. Always helpful to have something to compare to in these cases.

I just re-measured the voltage rails on my A18 board and they are now well within spec. +/-5LHR now measures 14.6VDC and 0.2VAC (RMS) both using handheld Fluke DMM. 

A few possibilities
Maybe it needed some time to burn in after having all the new parts fitted.
When I made the original measurements only A20,A19,A18,A17 cards were fitted. This time it is fully equipped. Maybe the measurements are meant to be taken under load with the other cards fitted.
Or maybe I just got the wrong pins originally 

Well at least it seems to be in spec now


And yes, I can see this will be high maintenance. Nice to have though if I can fix it up. Tracking down the dead relay (or relay driver) is going to be tough. Like TiN said, learning topology is key here to narrow down the search. Replacing the relays behind the front panel might be worthwhile anyway if I can source replacements, I'm sure I saw a video where they were replaced.  A lot of these old relays are obsolete and hard to find now.

[TiN]

Maybe it needed some time to burn in after having all the new parts fitted.

No, 15 minute "burn in" is enough, SD and calcheck should not give out any errors without burningins..

When I made the original measurements only A20,A19,A18,A17 cards were fitted. This time it is fully equipped. Maybe the measurements are meant to be taken under load with the other cards fitted.

You went waaay too fast again.  Stop, remove all unnecessary cards (HV and AC) and test DC boards first and calibrator DC output.
Configuration should be
A20+A19+A18+A17+A7+A8+A9+A10+A11, nothing else.

Because :

5700A                 1281
10.000000VDC = 9.9986185VDC

Means its unhappy and broken. With original EEPROM (given you didn't format/erase it on A20) 5700 should never be that much off 10V.
Also forget about 2wire in resistance measurement for now, use only 4w mode. Resistors are fixed values here (unlike 55XX) from fixed resistors, so should be close to calibrator value as well, unless some relays in path are bad.

Well at least it seems to be in spec now

Wait, what? What is in spec? You still have a path ahead in repairs...

I got some information out of this unit finally (attached)

Cal check report is useless without actually running calcheck. You can see all shifts reported as 0.000 ppm. Meaning actual Calcheck procedure was never ran or aborted with failure. It takes bit under an hour to complete whole calcheck. No external connections needed.

[veedub565]

No, 15 minute "burn in" is enough, SD and calcheck should not give out any errors without burningins..

I'm talking about DC voltages on the A18 board test points here not SD or calcheck. +/-5LHR rails were measured around 18v after capacitor and resistor change, and now they are 14.6v. I was speculating that maybe the new components had burned in after I originally made the measurements. Either that or I got the wrong test points.

Wait, what? What is in spec? You still have a path ahead in repairs...

The spec for the DC voltage rails on A18 board. +/-5LHR rails are now 14.6v which is inside the spec or 12v +/-4v

Cal check report is useless without actually running calcheck. You can see all shifts reported as 0.000 ppm. Meaning actual Calcheck procedure was never ran or aborted with failure. It takes bit under an hour to complete whole calcheck. No external connections needed.

I noticed that after I posted, I did run calcheck, maybe I didn't leave it long enough to complete fully, or it aborted with failure.

You went waaay too fast again. 

I still have much to learn, I will go back to A20+A19+A18+A17+A7+A8+A9+A10+A11 configuration and take it from there.