Fluke 8502A - fuse blown

[Rax]

Success!!!!!

I pulled Q27 and Q9 for the A/D board which are dual n-channel jfets and Q27 tested badly on the curve tracer, basically no curves at all on both sides. It’s late at night but I thought I would just try to replace Q27 with 2 J113 jfets that I had and the meter has come to life. I guess there may be flow on effects but for now I will run the meter until something fails I guess. I will try to run a calibration tomorrow before ai get too excited. I may try to find a data sheet for the old jfet but the markings and manual don’t provide much of an indication of the model and maker.

Thankyou to all who assisted with suggestions etc

Richard

Richard - great work!!! So awesome when a repair succeeds. Well done!

BTW - I've replaced C13 in mine, which is the only electrolytic on the board, so suspect by default at this age, but also measured out of spec. May be a good idea to check it and replace if off. I think this improved things quite a bit on my AD module.

I actually wonder if my unit has issues in the same area (maybe Q27 failing intermittently?), as "Error 5" has to do with the REMAINDER (though I don't think you've seen this particular error, have you?). Did you install the two J113s in physical contact? I'm curious what will be your permanent solution for this part. I wonder if any of the dual JFETs from LS would work.

[RichardM]

Success!!!!!

I pulled Q27 and Q9 for the A/D board which are dual n-channel jfets and Q27 tested badly on the curve tracer, basically no curves at all on both sides. It’s late at night but I thought I would just try to replace Q27 with 2 J113 jfets that I had and the meter has come to life. I guess there may be flow on effects but for now I will run the meter until something fails I guess. I will try to run a calibration tomorrow before ai get too excited. I may try to find a data sheet for the old jfet but the markings and manual don’t provide much of an indication of the model and maker.

Thankyou to all who assisted with suggestions etc

Richard

Richard - great work!!! So awesome when a repair succeeds. Well done!

BTW - I've replaced C13 in mine, which is the only electrolytic on the board, so suspect by default at this age, but also measured out of spec. May be a good idea to check it and replace if off. I think this improved things quite a bit on my AD module.

I actually wonder if my unit has issues in the same area (maybe Q27 failing intermittently?), as "Error 5" has to do with the REMAINDER (though I don't think you've seen this particular error, have you?). Did you install the two J113s in physical contact? I'm curious what will be your permanent solution for this part. I wonder if any of the dual JFETs from LS would work.

Thanks Rax

I will test C13 again now you mention it. Thanks for the tip.

No, I don’t recall seeing Error 5 but you may want to check those jfets anyway. I found the one that does work a little noisy on the curve tracer as well. I didn’t tie the two j113 together but that’s prob a good suggestion. It’s a bit tricky as the legs/pin out were inconsistent and needed to be bent at strange angles. I am considering pulling out Q9 again and capturing the curve tracer curves and then searching for a suitable match long term. For now I will see how it performs.

Richard

[Kleinstein]

The J113 may be a reasonable choice. The input is not that high in impedance, so input leakage is not an issue. The J113 are reasonable low noise. For the pin bending it may help to know that the drain and source are interchangeble.
It may need some selection for a reasonable low drift for the pairs (probably in circuit, or if the drift is too high adjust one of the drain resistors)

The needs for Q9 should be similar.

If really desparate and performance really suffers, there would be the last resort of replacing the JFET pair and follwing OP with just a single modern high performance OP.

[RichardM]

The J113 may be a reasonable choice. The input is not that high in impedance, so input leakage is not an issue. The J113 are reasonable low noise. For the pin bending it may help to know that the drain and source are interchangeble.
It may need some selection for a reasonable low drift for the pairs (probably in circuit, or if the drift is too high adjust one of the drain resistors)

The needs for Q9 should be similar.

If really desperate and performance really suffers, there would be the last resort of replacing the JFET pair and following OP with just a single modern high performance OP.

Thankyou for your expert advice. I did notice that the curve tracer gave the same output regardless of how I connected the drain and source. As a novice I still thought they needed to be connected as per the datasheet. I may not need to bend them now, thanks.

Not sure whether to also replace the other dual opamp ? If it aint broke.... ?

[RichardM]

From tears of joy to tears of despair!!

The last module I put back into its plastic shell and installed back in the Fluke was the filter module. Unfortunately the TO99 opamp I replaced was sitting a little too high and touched the tin shielding inside the plastic case and out came the magic smoke!! I hope its just a case of replacing 2 burnt resistors and ordering another opamp for $20..... Sigh

I wish there was another high voltage opamp that was equivalent to the LM1436H.

Richard

[SilverSolder]

From tears of joy to tears of despair!!

The last module I put back into its plastic shell and installed back in the Fluke was the filter module. Unfortunately the TO99 opamp I replaced was sitting a little too high and touched the tin shielding inside the plastic case and out came the magic smoke!! I hope its just a case of replacing 2 burnt resistors and ordering another opamp for $20..... Sigh

I wish there was another high voltage opamp that was equivalent to the LM1436H.

Richard

A good trick is to take a "Before" picture with your phone, and then try to make the repair look the same.  You rarely go wrong by having the new parts installed exactly the same way as the old ones - and it is a great sanity check (did I put that ferrite bead back on the correct leg, etc.)!

[m k]

At this point, I have a pretty high degree of confidence "high res" triggers "Error 5." I left it run measuring about 13Vdc with all digits possible outside "high res" for just about 24hrs, the error never came up. I switch it to "high res," it locks up on "Error 5" right away (may take up to a minute). I did the same with my AP box - for a calibrated AC level - and it locked up when I requested "high res."

This happens on both ACV and DCV.

How you measure that Error 5?
Since it's high res can it be that summing part it the problem?

Simplified analog part of A/D in O&S page 52/354.
Explanation in previous page.
https://www.manualslib.com/manual/888772/Fluke-8502a.html#product-8502A

Digital of A/D in O&S schematics pages 353-354/354
ID0 is Reset
ID1 is AutoZero.
ID2 is Buffer In and Remainder In, topmost NAND with inverted inputs is OR.
ID3 is Channel Y.
ID3 inverted is Channel X.

So topmost Tr of pictured U17 is both AutoZero and Buffer In.
It's pretty likely fully operational.
It also possibly shouldn't happen if Error 5 happens before it.

Anyway, with those data lines you should be able to analyze how intermittent that High Res Error 5 is.
For now it seems that you have a temp and noise situation.

[Rax]

At this point, I have a pretty high degree of confidence "high res" triggers "Error 5." I left it run measuring about 13Vdc with all digits possible outside "high res" for just about 24hrs, the error never came up. I switch it to "high res," it locks up on "Error 5" right away (may take up to a minute). I did the same with my AP box - for a calibrated AC level - and it locked up when I requested "high res."

This happens on both ACV and DCV.

How you measure that Error 5?
Since it's high res can it be that summing part it the problem?

Simplified analog part of A/D in O&S page 52/354.
Explanation in previous page.
https://www.manualslib.com/manual/888772/Fluke-8502a.html#product-8502A

Digital of A/D in O&S schematics pages 353-354/354
ID0 is Reset
ID1 is AutoZero.
ID2 is Buffer In and Remainder In, topmost NAND with inverted inputs is OR.
ID3 is Channel Y.
ID3 inverted is Channel X.

So topmost Tr of pictured U17 is both AutoZero and Buffer In.
It's pretty likely fully operational.
It also possibly shouldn't happen if Error 5 happens before it.

Anyway, with those data lines you should be able to analyze how intermittent that High Res Error 5 is.
For now it seems that you have a temp and noise situation.

m k - thank you very much for this overview. Would I need a logic analyzer to look at these data lines? I am not versed in this type of work. Not sure if what I need to look at is visible with a scope (HP 54522A). Some other instrumentation available at hand but could use some guidance on what would be useful.

Would you also elaborate on the "temp and noise situation?"

Thanks a bunch!

[Rax]

m k - further on your points. I've had a hunch on the summing nodes (Q11-Q21) due to the triggering of the error by "high res," along the lines of your train of thought, I think.

Also, when "Error 5" occurs, TP5 hangs on either 0V or -7V (most often -7V), and stops switching:

When this happens, TP5 hangs either on a high state of -7V, or 0V (instead of the switching behavior I placed screenshots of above; during "Error 5," it flatlines).

That said, I am unsure of next steps.

[m k]

You can store what happens in ID7 but I'm not sure how to trigger it.
And if I understand it right it takes quite a long time to complete.

Temp part of temp and noise is because you said that some times the error is not coming right away after selecting High Res.
Noise part is because if the error is staying away without High Res it's not very big and maybe it's even an adjustent thing but there are not much to adjust.

Somebody should verify my understanding.
Every D/A converter part is used more than once in every value generation and one or some switching reference parts are used only once?

[Rax]

You can store what happens in ID7 but I'm not sure how to trigger it.
And if I understand it right it takes quite a long time to complete.

Temp part of temp and noise is because you said that some times the error is not coming right away after selecting High Res.
Noise part is because if the error is staying away without High Res it's not very big and maybe it's even an adjustent thing but there are not much to adjust.

Somebody should verify my understanding.
Every D/A converter part is used more than once in every value generation and one or some switching reference parts are used only once?

m k - thank you for elaborating your thoughts (and I appreciate their "seeking form," which I mean in the best way of "thinking on this with me"). The noise part could also be leakage, and I'm thinking of the FETs in the summing node (or maybe differential inputs elsewhere?). The temp part... gut tells me no, because the fault occurs just too quickly after creating what I've determined to be conducive conditions, at least lately.

It'd be great learning to acquire and interpret the data/control lines. I wonder if this could be supplied with affordable hardware.

On a different track, I grabbed a parts unit off of ePay and may try to continue isolating by swapping modules from the donor unit. Are the codes on the modules cases enough to match them (for instance, "fast A/D Converter" = 383984), or codes from the PCBs are required for safe swapping? I'm thinking maybe software accounts for some variations and don't want to upset the unit. Of course, interdependent adjustments may also throw things off, though I assume that depends on the exact module replaced.

[m k]

Maybe I understand it a bit better now.
All parts are used many times or all the time and switching time is what makes it closer to goal.
(I have that HP Journal but it's in the que)

With noise I didn't mean ripple but more generally something that flips the error flag up.

My understanding of the value generation is seeking it using iterations and flipping comparator U7 is telling it's too far now.
That bit value is then put out to controller using ID7.
You can try trigging it with U6 pin 13 of digital board.
(right of text of down of 4 D-flops of upper left corner)
If you get much less than 20 bits you're probably getting only partial value and next part is completely its own thing and next pin 13 down/up.
It's also possible that you're getting same bit values.
The pin is operating so that first D-flop's R is up and Q is down, then R goes down allowing next S put Q permanently up, clock is never happening so D is also never happening.
With storing the trigger channel also you can see how long you should get bits.

Why High Res.
Last reference is so close to input that comparator misbehaves.
If iteration starts from center and comparator indicates less in both directions the value is exactly there in the middle.
But if comparator starts indicating more in both directions it's an error.

I don't understand those completion part offsets.
(3-73 of O&S)
Finally Q4 and Q8 are opened, Q5, Q6 and Q7 closed and offsets stored.
Where, the sample is complete and circuits are autozeroed.

The operation of the meter is pretty slow compared to what is available now.
So pretty much any logic analyzer will do.

I have many never used China gadgets.
Two months shipping brings up anybody's inner fortune teller.
Anyway, one of those is USB logic analyzer and no idea how it works, close to
https://www.sparkfun.com/products/15033

A/D 383984 seems to be the only module version in all models.
Much more likely you are upsetting the software with almost anything else.

[Kleinstein]

The gain from step to step is x 16. So From the 5 bit DAC (plus maybe the sign) only 4 bits should be really used in the later cycles. Changes are the highest bit is mainly used in the first conversion and should not be in used in the later cycles. It should still not be a problem with high res mode, but not the normal mode.
 
The residual signal is amplified from cycle to cycle. So the comparator would not see a weaker signal later. A signal just at the edge may happen (just random at any stage), but this should not get the ADC to an error state. Worst case get a metastable state and a bit error. I don't see a much higher frequency for the high res mode here.

A point that could upset the ADC could be if the gain from cycle to cycle is too large and from this it may need the highest bit. In theory this could be fixed with some math, but my feeling is the old ADC is not using numeric error correction but would need the trimmers to be set right.

The sampling type ADC is a bit different from modern DMMs with there integrating ADC. This makes the old Fluke meters a bit noisy / sensitive to hum  or slow settling if they use the analog filter. The meter was not that bad for it's time.

Normally TP5 should get a 0 V only for the auto zero phase. So stuck at 0 would mean something like a stuck control sequence, not somerhing like noise or an offset in the analog part. Maybe have another check on the supply for the digital side.

[SilverSolder]

[...]I grabbed a parts unit off of ePay and may try to continue isolating by swapping modules from the donor unit. [...]

That is a great idea and a huge troubleshooting time saver with these units, given how easy it is to swap modules.  Even if you get both of them working, it never hurts to have two DMMs (and the ability to swap modules between them!).

I believe the A/D converter is one of the boards that works in pretty much any version of the instrument, so if it was mine, I would go ahead and try the "new" one.

[Kleinstein]

For the new unit, I would still first check the supplies: a broken regulator can do quite some damage.

The modulales should not interact very much. So the adjustments are local to the modules and not much need for a new adjustment with swapping modules.

[Rax]

For the new unit, I would still first check the supplies: a broken regulator can do quite some damage.

Good point, it's also typically my first point of evaluation, as it can give me a first indication of any possible big faults in the modules themselves (such as a short, etc.). It's neat that with these units, the PS can be pulled out of the unit very easily and be verified (and then adjusted) by itself.

[m k]

Instruction Manual's 4-52 has A/D Converter calibrations.

First there is to check that between TP7 and 6 is -7V.
Next is zero adjustment, pretty easy also and without any meters, just a good short needed.

[Rax]

Instruction Manual's 4-52 has A/D Converter calibrations.

First there is to check that between TP7 and 6 is -7V.
Next is zero adjustment, pretty easy also and without any meters, just a good short needed.

Thank you, m k. This far, all -7V references where I measured are what I'd see as dead on. On the AD it's -7.0000V on my Fluke 8840A (occasionally going -7.0001V, which may account for minute environmental or otherwise variations in either meter in my mind) - BTW, the calibration test equipment per SM requires a 0.01% voltage accuracy DMM, while the 8840A is I believe 0.005%. On the ohms module, for instance, it's -6.9999V, sometimes going -6.9998V. I think the manual says +-0.0001V is acceptable. I've looked a bit, but I've not yet figured out whether there's one main -7V reference for all modules (with local adjustments adjustments) or each has their own.

I will check in the new modules, but, overarchingly, I will ever be amazed at how well these instruments hold with age.

I think most posters said the calibrations and adjustments on different module are typically independent of each other (they don't interact), which is great to know and a very positive scenario.

[SilverSolder]

Instruction Manual's 4-52 has A/D Converter calibrations.

First there is to check that between TP7 and 6 is -7V.
Next is zero adjustment, pretty easy also and without any meters, just a good short needed.

Thank you, m k. This far, all -7V references where I measured are what I'd see as dead on. On the AD it's -7.0000V on my Fluke 8840A (occasionally going -7.0001V, which may account for minute environmental or otherwise variations in either meter in my mind) - BTW, the calibration test equipment per SM requires a 0.01% voltage accuracy DMM, while the 8840A is I believe 0.005%. On the ohms module, for instance, it's -6.9999V, sometimes going -6.9998V. I think the manual says +-0.0001V is acceptable. I've looked a bit, but I've not yet figured out whether there's one main -7V reference for all modules (with local adjustments adjustments) or each has their own.

I will check in the new modules, but, overarchingly, I will ever be amazed at how well these instruments hold with age.

I think most posters said the calibrations and adjustments on different module are typically independent of each other (they don't interact), which is great to know and a very positive scenario.

The reference is brought out on the backplane bus, so any module that needs it has access to it.

[Rax]

The reference is brought out on the backplane bus, so any module that needs it has access to it.

Awesome, thank you for confirming that. Then I'd say in my main unit the -7V Vref is confirmed working within spec 100%.

[m k]

If you need more spares I have
https://www.ebay.com/itm/174945154128
discounted in my cart.

USA reroute is possible, everything else is unknown.

Rerouted.

[Rax]

Anyway, one of those is USB logic analyzer and no idea how it works, close to
https://www.sparkfun.com/products/15033

I got this: https://www.amazon.com/gp/product/B07K6HXDH1/ref=ppx_yo_dt_b_asin_title_o01_s00?ie=UTF8&psc=1. ***Luckily,*** in the US we have the big elephant in the room and things can be had (delivered) sometimes the same day.

I've got it running on my Windows 10 laptop (PulseView/Zadig) - seems to be purring 100%. Now I wonder if anyone can point me to the applicable reading material. I'm entirely not sure this is good for the job, or if it is even compatible with the hardware on the Fluke. Any pointers greatly appreciated - thanks a lot in advance.

This is at least until I get the unit for parts and try swapping got get ahead with this (early next week).

[m k]

Somewhere was mentioned that some parts are groupies.

In the case like that you should change all of those that are selected to that group.
Not much more than 2 components I guess.

Calibration is of course the next thing in that situation.

E,
Sigrok seems to be suitable.

You can always try installing it again.
Go to device manager and uninstall its drivers.
Then with luck reinstalling picks better usb side base drivers.

E2,
I mean the device.

[Rax]

Well, drat. I got the parts unit with spare modules, and swapping the AD module still returns Error 5. I'd have thrown a couple of coins on the table this was gonna fix it. I may try other likely modules as see what's happening, but I'm a little taken aback and will be mulling over this for a bit.

[Rax]

You can always try installing it again.
Go to device manager and uninstall its drivers.
Then with luck reinstalling picks better usb side base drivers.

No need, I probably wasn't clear but the install went very smoothly. I think it's working OK.

I just have no idea how to connect it, where, etc., but not knowing which end to grab this "using a logic analyzer on a DMM" issue, I thought I'd ask to be sent to read somewhere the basics if that resource exists. I'm not sure how complex the topic is (this is pretty far afield for me, but eager to learn), so I wanted to be respectful of everyone's time. Maybe it's doable through simple instructions and then I'll just get right on to it.