Fluke 5200A repair (?)

[Rax]

I'm the happy owner of this 5200A from a member of this community I have loads of thanks for (you know who you are!).

It's been measuring brilliantly (see attachment). I think the seller was very conservative in their assessment of the condition of the unit. I have a "calibrated" (long form Z540-1 available) Fluke 8502A I can compare readings with, and I think it's likely within cal most everywhere. BTW, the 8502A has a pretty amazing RMS module, being able to measure dead on to 1MHz to a certain level without a hitch. I am using it to measure the 5200A output, and the next column after that lists the values displayed by the 8502A for standard lab excitations.

Now, what's not so good is that sometimes it seems to go on a state where "OVERLOAD" light flickers on and the output starts fluctuating. This is rare (this far), and I'm still examining the mechanism by which OVERLOAD activates, but if anyone has experience with this unit and this looks familiar, I'd appreciate input.

Thank you.

[Rax]

My first assumption with this "flickering OVERLOAD" annunciation issue is a component becoming stressed by overheating.

When the light flickers, there's an audible noise from within the case - I don't think it's a relay (though I'm still looking over the OL circuitry, which I don't have a handle on yet) - I think it's rather a component exhibiting mechanical stress that becomes audible.

The instrument has been on for about 24hrs - not all of which supervised - but I feel it's been fine (not doing the above) for most of its current cycle.

[Ugur]

It could be an old capacitor. or an unstable op-amp...

[Rax]

The other condition that is occurring is that the OVERLOAD light never comes on during startup. Per the startup procedure, I believe it's supposed to be on at the beginning, the turn off after 5-10 seconds.

Instead, in my unit OL never turns on, but the STDBY/OPER lights are initially both off, then - just about when OL should probably go off - the STDBY turns on. Then, I'm able to turn that knob and put it into OPER to get signal out.

[Rax]

It could be an old capacitor. or an unstable op-amp...

Completely possible. I have not yet really opened the unit to look for any evident trouble - such as stressed or burned resistors, leaked caps, etc. - other than to see what's moving loose inside (I assume a screw...). This is because upon receiving it, I noticed/heard this and opened the case and internal shields. The unit had quite a lot of screws missing everywhere. It's a miracle it arrived in the shape it did, as structurally, the unit was essentially pretty "flexible." I have not yet found what's moving inside, but I rested the unit on its rear handles carefully and turned it on.

Other than this OL condition, the unit is in very good shape, operationally. I'm assuming the issue must be pretty localized to some specific condition or fault.

[tridac]

Still worklng to try and get just one 5200 fully working, Bought two other for parts donors and now have the basic system working on all except one voltage range. The power amplifier in particular, is fragile and there is no current limiting, so if it goes down, there are low wattage sacrificial resistors on the psu pcb that burn out to protect other areas. Rebuilt the psu, low value resitrs on pcb pins, fitted heat sinks on a couple of transistors and fitted a fan, cross flow across the psu compartment. Some of the psu parts run too hot to touch and discolour the boards over time, tracks lift and as a design, a bit on the edge. Psu is fine now, but mine has the o/l led flashing at times, so the next step is go through the cal procedure as far as possible. If still issue, then dig deeper. According to the the manual, just about all the basic functionality can be checked out, range by range, with the power amp board removed and a jumper fitted, which might save some trouble. Afaics, the overload circuit shorts the input to the power amp on an anomaly, so if that's happening, the issue might be earlier in the chain. Fully direct coupled signal chain as well, ouch.

Run the zero offest calibration procedure on the power amp at least, as it is direct coupled to the output ratio transformers, which can burn out, making the whole thing scrap. Also, replace the two power amp 68K 2w comp resistors, which run quite hot, as they will blow the power amp if they fail with age and go oc. (mine did) Replaced with 2w metal oxide. Old, but a really good calibrator, with accuracies of a handful of microvolts when right, but Fluke were stretching the art at that time. As shown by things like the power amp bootstapping technique, to enable the use of high frequency, but low voltage and power (TO5 can) transistors in the output stage. +/- about 300 volts on the power amp, ~600v dc total, so be carefull where you stick probes and for your own safety. Good news is that the power amp uses sockets for most transistors, so you have to ask, why they did that. The schematics are amazing and not too difficult to work out what each board does. Very clever design for it's time.

Definately a keeper, if you can keep it alive :-)...

[Rax]

replace the two power amp 68K 2w comp resistors, which run quite hot, as they will blow the power amp if they fail with age and go oc. (mine did) Replaced with 2w metal oxide.

Thank you for all that input - very appreciated.

I assume these are R40 and R46 on the A7 (Power Amplifier) module, right? For something needing high reliability and some muscle, I usually resort to Vishay Dale WW (such as RS, probably 5W here). Not non-inductive by default, unless needed (I don't think here that's the case?....).

[Rax]

a really good calibrator, with accuracies of a handful of microvolts when right, but Fluke were stretching the art at that time.

This thing's quite amazing, given the relatively wide bandwidth, keeping the levels so precise is pretty breathtaking. In the RF world, fractions of a...dB!! is an amazing level of performance, this is obviously in a different world of precision where one doesn't need to quantify in dB but %. Not quite ppm, but % is pretty dope.

I am currently trying to decipher any patterns in the occurrence of the OVERLOAD aberration - may be temperature related, maybe just something relatively random (= not yet determined) about the failing part?... Should be said that for all such work I always start from the power supplies, and in this case I haven't even yet started that examination. So maybe I just need to pull the sleeves on this thing. That said, today's been working without skipping a beat. I had nothing to go by, it's been flawless.

[tridac]

It is quite amazing, considering that the first edition manual is dated 1975, which means they must have started the design in 1972/73. Might look a bit clunky in places, but that was the tech of the day, what was available. The reference board is worth  a look. The usual Fluke dc reference amplifier chip design, the output of which is then pulse width modulated and filtered, to get  a variable reference, to compare against the output from the precision rectifier. Both the filter and galvanic isolation are interesting. If you take the cover off the reference board, there are wound ferrite toroids either side of a screen plate. a few turns on each toroid, then  a single turn loop across the screen, to couple the two together, pwm pulse on and off.

There is  an earlier note on some 5200 mods here:

https://www.eevblog.com/forum/metrology/fluke-5200-ac-calibrator-fan-notes/msg5001235/#msg5001235

The resistor mentioned are the right ones to change, bias current from the + / - 300 volt rails. They  can be unsoldered carefully from the board top side, then use a solder sucker to clean the vias. Don't need to take the board / heatsink assy apart, but do form the replacement leads to standoff from the board, to allow air flow. The two power amp adjustments are output offset and quiescent current adjust. All in the manual, but suggest power off, connect probes, then power on again procedure to adjust, to avoid possible smoke :-). You will need an card extender board for serious debugging, but found an Ebay seller in Korea making copies of the Fluke oriiginal. Seems unobtainium otherwise...

[Rax]

It is quite amazing, considering that the first edition manual is dated 1975, which means they must have started the design in 1972/73. Might look a bit clunky in places, but that was the tech of the day, what was available. The reference board is worth  a look. The usual Fluke dc reference amplifier chip design, the output of which is then pulse width modulated and filtered, to get  a variable reference, to compare against the output from the precision rectifier. Both the filter and galvanic isolation are interesting. If you take the cover off the reference board, there are wound ferrite toroids either side of a screen plate. a few turns on each toroid, then  a single turn loop across the screen, to couple the two together, pwm pulse on and off.

I think this was the only AC calibration fixture in the Fluke lineup for quite a while, before they started packing features into the multifunction calibrators. So a great unit to have and the only one at that... (for hobbyist-level metrology geeks like yours truly!).

The resistor mentioned are the right ones to change, bias current from the + / - 300 volt rails. They  can be unsoldered carefully from the board top side, then use a solder sucker to clean the vias. Don't need to take the board / heatsink assy apart, but do form the replacement leads to standoff from the board, to allow air flow. The two power amp adjustments are output offset and quiescent current adjust. All in the manual, but suggest power off, connect probes, then power on again procedure to adjust, to avoid possible smoke :-).

This is very thoughtful, thank you. I've been cruising it a little apprehensively, though keeping the space this is in a little cooler (SoCal is currently seeing some pretty serious heatwaves). But I'm an absolute believer in overdesign (and, consequently, huge fan of eliminating under-design wherever identified.... so many instances in the late '70s and throughout the '80s at the onset of solid state). I'll likely have appropriate resistors coming next week and will do the replacement. But it's been almost constantly on since I started the thread and no more OVERLOAD flickering. Hope for the best!

You will need an card extender board for serious debugging, but found an Ebay seller in Korea making copies of the Fluke original. Seems unobtainium otherwise...

I just got a couple of versions of extender boards for the 5440A calibrator, maybe I shouldn't stop at that... I'll look into how the corresponding extender cards looked like, but if you have some info, I'd appreciate it.

[bdunham7]

I think this was the only AC calibration fixture in the Fluke lineup for quite a while, before they started packing features into the multifunction calibrators. So a great unit to have and the only one at that... (for hobbyist-level metrology geeks like yours truly!).

Oh no, there's another that matches your 5440A.

https://download.flukecal.com/pub/literature/1260112E_w.pdf

I just got a couple of versions of extender boards for the 5440A calibrator, maybe I shouldn't stop at that... I'll look into how the corresponding extender cards looked like, but if you have some info, I'd appreciate it.

Having worked on one once without extenders, I'll say that it sucks but it can be done. 

[alm]

I think this was the only AC calibration fixture in the Fluke lineup for quite a while, before they started packing features into the multifunction calibrators. So a great unit to have and the only one at that... (for hobbyist-level metrology geeks like yours truly!).

Indeed, the next calibrator to offer AC was the 5100 series, but it had higher uncertainties and didn't have the frequency range or flexibility (only discrete frequencies). It did go to 1 kV. The unit that replaced the 5200 was the 5700 in the late eighties. The 5200 remained in the catalog for a few years after the 5700 was introduced: it's in the 1994 catalog, but not 1996. I'd say the 5200's biggest limitation is that to go beyond 120V you need an amplifier that is even bigger and heavier. Later units by Fluke and others incorporated this amplifier, although the Fluke 5700 still needs an amplifier to go beyond 1 kHz in the 1000V range.

I just got a couple of versions of extender boards for the 5440A calibrator, maybe I shouldn't stop at that... I'll look into how the corresponding extender cards looked like, but if you have some info, I'd appreciate it.

I bought an extender on eBay years ago. It was quite crude, just a piece of routed copper-clad with an edge connector but no plating or chamfering on the edge connector, or solder mask. I think it's the same seller that's still offering the partial kit on eBay.

Oh no, there's another that matches your 5440A.

https://download.flukecal.com/pub/literature/1260112E_w.pdf

No, but that one could be used to calibrate a 5200A in a slightly overkill way. That's just a very fancy AC voltmeter. The 5200A was the last AC-only calibrator from Fluke. Datron continued longer with the 4200(A) and later offered versions of the 4708 configured for just AC volt/amps (no DC or resistance). It wouldn't surprise me if some of those were used with the Fluke 5440 since Fluke had no AC calibrator with the same specifications until the 5700. The Datron calibrators all had the 1 kV amplifier built in, sometimes optional. They have a better frequency range at high voltage than the Fluke 5700 without amplifier.

[Rax]

The unit that replaced the 5200 was the 5700 in the late eighties.

Isn't the 5700 a multifunction calibrator, though? Not ACV only. 

[Rax]

Oh no, there's another that matches your 5440A.
https://download.flukecal.com/pub/literature/1260112E_w.pdf

I think the 5790A(B) is a current model, no? Or maybe the started making the "A" version in the '80s?

Now that you're mentioning this model, I think I recall an "AC-only" calibrator, but it can't be the 5790A, nor the 5700 (the latter being a multifunction). Am I correct in this?

So which exact model for AC exactly matches the 5440A?

[alm]

Isn't the 5700 a multifunction calibrator, though? Not ACV only.

Yes, but it's the first unit that surpassed the 5200 in uncertainty. Fluke hasn't made an ACV-only calibrator after the 5200, or a DCV-only calibrator after the 5440. They've only made multi-function ACV/DCV/ACI/DCI/R calibrators since.

[Rax]

Fluke hasn't made an ACV-only calibrator after the 5200, or a DCV-only calibrator after the 5440. They've only made multi-function ACV/DCV/ACI/DCI/R calibrators since.

What about bdunham7's example of the 5790A/B?

[alm]

What about bdunham7's example of the 5790A/B?

I already responded to that. The 5790 is basically a 540 AC-DC transfer standard replacement:

No, but that one could be used to calibrate a 5200A in a slightly overkill way. That's just a very fancy AC voltmeter. The 5200A was the last AC-only calibrator from Fluke. Datron continued longer with the 4200(A) and later offered versions of the 4708 configured for just AC volt/amps (no DC or resistance). It wouldn't surprise me if some of those were used with the Fluke 5440 since Fluke had no AC calibrator with the same specifications until the 5700. The Datron calibrators all had the 1 kV amplifier built in, sometimes optional. They have a better frequency range at high voltage than the Fluke 5700 without amplifier.

[Swainster]

Took me a while to realise that I have some experience of this model. Not the same symptoms but may be something to bear in mind. In my case, my 5200A blew out the magic smoke  a while back. The problem turned out to be a number of shorted tantalum caps, I think in the reference assembly. Anyway, it was the pcba with the small transformers linking the sections together. A number of tants had failed - enough that I wondered if they were a symptom of something else. However, once replaced everything seemed to work fine 

Anyway, the pics show the obviously popped caps, but there were a few more inside the assembly.

Rax's recent threads have inspired me to tackle my 5440B. I've noticed that it has suddenly drifted by 60 ppm. It's obviously not a happy bunny.

[bdunham7]

No, but that one could be used to calibrate a 5200A in a slightly overkill way. That's just a very fancy AC voltmeter.

I've not had one on the bench and I never realized that they didn't include a source.  For what those go for that would have been a pretty big disappointment!

[Rax]

I already responded to that. The 5790 is basically a 540 AC-DC transfer standard replacement:

I understand your point now. I was of bdunham7's mind, I thought the unit includes a source... That's a very odd unit, form factor is distinctively "calibrator-like," but it's essentially a very sophisticated ACV. Looks almost deliberately confusing.
Thank you for helping to clarify that!

[tridac]

Restored a 5100B a few years back. Power amp rebuilt, microprocessor, ram, the front panel push buttion switches and more. Like the hifi separates thing, all in one boxes always seem like a compromise, whereas dedicated function units like the  5200 and 335 seem far more complete. One tool to do one job really well...

[Rax]

Further observation and run time doesn't seem to reveal any of those OVERLOAD events. So it's been solidly working for a few days now.

But one thing that's off is that at 100V out and 1MHz setting, the OVERLOAD light may actually come on - the output is unstable.

I think it's just a matter of the relatively high frequency output, which both stretches the design, and makes things like cables a bigger factor (I just switched to 50 ohm BNC cables and things seem to be more behaved).

[Bill158]

But one thing that's off is that at 100V out and 1MHz setting, the OVERLOAD light may actually come on - the output is unstable.

I think it's just a matter of the relatively high frequency output, which both stretches the design, and makes things like cables a bigger factor (I just switched to 50 ohm BNC cables and things seem to be more behaved).

One point here.  The 5200A manual specs, pg. 1-4 under "MAXIMUM OUTPUT VOLTAGE".  120 volts @ 100 KHz is the maximum output spec.  From there increasing frequency requires lowering the output voltage to 8.33 volts at 1.2Mhz.  I have already PMed RAX with this information.  My 5200A is able to get to 200 KHz before going out of regulation.  RAX's seems to go higher.  I am surprised that FLUKE didn't implement some sort of logic to prevent this condition like turning on the "OVERLOAD" LED when in this area?  But in thinking about this it would be difficult without uP control of some sort.
Bill

[bdunham7]

120 volts @ 100 KHz is the maximum output spec.  From there increasing frequency requires lowering the output voltage to 8.33 volts at 1.2Mhz.  I have already PMed RAX with this information. 

Not to mention the input specs on most meters don't allow 10⁸ V-Hz.  Oops!

[Rax]

My 5200A is able to get to 200 KHz before going out of regulation.  RAX's seems to go higher.  I am surprised that FLUKE didn't implement some sort of logic to prevent this condition like turning on the "OVERLOAD" LED when in this area?  But in thinking about this it would be difficult without uP control of some sort.
Bill

I've read - and still do - that page spec differently - I assumed there's a built-in attenuation doing that linear decrease in output to the 8.33 at 1MHz. The OL actually kicks in most of the time in that part of the range, though not all of the time.