Fluke 332D Calibrator repair

[cncjerry]

I was lucky to pickup a somewhat functional 332D yesterday.  Initially I had a problem with the reference not providing the correct voltage but after reseating things in that area it came to life.  It maps well against a known-good Geller (thanks Orin, someday I'll return it) and other standards I have with the exception of the second decade from the left. 

Looking at the schematic, R30 in parallel with R29 when combined with R4, r20 and r13 you would think should be around 10k after adjustment.  As near as I can tell, one of the resistors either R30 or R29 could have dropped in value as the combination comes to around 4.7K.  So in the dial settings where S2 isn't shorting that part of the string(0,2,4,6,810), all is well within a few PPM; with that section of S2 open (1,3,5,7,9) I get about .47 of the additive value, consistent with my expectation of one of two 19.985K resistors dropping in value.  It looks like taking that section of the assembly apart will be tedious so before I do, I was wondering if anyone has a part box with those resistors?  The are low tempco, 0.01% and of course, part of a matched set.

Setting aside the parts box, anyone have recommendations for resistors?

Thanks

By the way, thanks again Dr. Frank for sending the manual, etc.

Jerry

[Dr. Frank]

Jerry, you're welcome!

It would be quite difficult to replace these resistors, as they are matched as a pair to a T.C. of about 0.5ppm/°C, or even better.
A Vishay VHP 101 of 9.9925k would be required.

I doubt, that one of the ww resistors is gone defect.

Maybe, there's contamination on the PCB, or the the switch produces a short in the open condition.
Latter may be caused by abrasion particles from the switch contact.

To avoid heating R29, R30, I would at first carefully desolder R4 and trim pot R8, to measure the ww resistors R29, R30.
Maybe, it's easier to desolder the appropriate connection from switch S1 to the PCB.

Anyhow, to find the root cause, I think you have to disassemble the switches and the whole PCB first.

Frank

[cncjerry]

yes, Frank, after I wrote that post I started thinking about the odds of the resistor going bad.  I wonder if there is something on the switch preventing it from opening all the way?  I am going to see if I can scope that area.  It is very consistent though.

It's a simple circuit.  If I didn't have to do all the disassembly I would have it down to the component.

Thanks again.

[cncjerry]

Frank, does that front panel assembly fold forward just by removing the 4 or 5 bolts on each side?  I can't tell if I have to disassemble the terminal connector or not?

[dacman]

You also need to loosen the set screws on the power switch shaft and pull the front panel straight forward until the shaft can be removed.  The shaft will otherwise prevent tilting the front panel.

[cncjerry]

I pulled the front out a little this afternoon and I still can't see in that area very well. 

I pult a meter across the resistors in question (they are in parallel) and they read 4k7575.  If I press on the board in that area, the reading drops a couple hundred ohms so I'm starting to think like Frank in that the resistors are OK but something is creating a high resistance short in that area.   When I press on the board it gets worse.  There are some standoffs in that area but they look insulated.  I found some silk-like material in there and I was thinking, "a ha" but no dice.  it looked like spider silk, quite a bit actually.  I can't see it, but maybe some shaft grease got on the board.

I was thinking about flooding that switch and area with contact cleaner.  Thoughts?  I know most would say, "don't do it".  Flood with IPA?

So enough said, when I get it down to my lab I'll have more room and will try to lift some components to figure it out.

Interesting unit but the problem is "the guy that has 2 voltage standards does't know the voltage".  I have four or five now (until I send Orin's back).

[dacman]

If you use IPA (or contact cleaner) on the switches, then a switch lubricant would also need to be used to prevent destroying the switches.  My switch cleaning method (on these types of switches) is that I first just use the lubricant (or something like DoexIT-Gold) and if that doesn't work, I then use contact cleaner (or IPA) followed by a lubricant.

[cncjerry]

Thanks for the hints on how to get the front board to tilt.  I should have thought about taking the power shaft off, duh.

Unfortunately I have a problem with the two resistors R29 and R30. One is measuring 17.377K and the other, 6.590K.  Combined in parallel they are 4.778k.  They should be 19.985K each for a parallel total of 9.9925K.

The 5 resistors in this string are supposed to be all matched.  These are the only two that are anywhere near being off from what I can tell.

I am attaching a picture of them as well as my oven.  The oven has taken a beating over the years.  You can see where it melted on the top side where most of the heat would form.  Overall, the unit is incredibly stable and seems to be calibrated pretty close to other standards I have.

The question is how to repair?  Search hopelessly for the same component?  Replace with 'good enough for home use'?  I have some very good 10k resistors, vishay type, that I could select from.  I could build-up 4 of them in a two series/two parallel pair for now.

I find it strange that these two resistors would go in that circuit and the other 3 of the set be ok.  These two are in parallel where the other three are in series, not that it makes a difference.  No heat in that area, I'm at a loss.

Anyone have a parts unit?  I would take resistors from a parts unit, the hell with matching them after all these years.

Thanks

[cncjerry]

I don't know if it means anything productive is happening, but I took one of the resistors and gave it a few good bangs.  The resistance jumped from 6.95K to 10.3k.  I assume these are wire wound?  I guess something is shorted inside?

[Johnny10]

Hey Jerry,
You know what I always say... I hit it a few good blows with a hammer and now I know exactly why it is broken !

Good project.

[doktor pyta]

During repairs of precision stuff three times I had problems with wirewound resistors with shorted windings (intermittent). They showed two values of resistance depending on temperature or mechanical shocks.
This might be Your case.

[dacman]

What is the size of the resistors?  (About 1/2" x 1/2"?)

The schematic has 7 of those resistors, R29 through R35.  The ones in series would have 20 Vdc across them on the 1000 Vdc range (if inserted) and the two parallel ones would have about 10 Vdc across them.

If I had a customer that was willing to spend upwards of $400.00 for parts, this is how I would get new parts:
1) I would contact PRC (Precision Resistor Company) in Florida USA and ask if they could make an HR508N resistor, which is not in their catalog.  (50 = 50/100 or 1/2" diameter.  8 = 8/16 or 1/2" Length.)
2) I would ask for 7 resistors, 3 pairs, each pair haveing TCs of zero when added, and the other single unit with a TC of zero by itself.
3) Parameters would be 19.985 kOhm.  Tolerance 0.01% or 0.005%.  TC pair inverse match of 0.5 ppm max. TC of 1 PPM max for the single unit.  I would also state that the TCs should be with a voltage of up to 20 Vdc.
4) The cost I would expect for 7 resistors would be from about $175.00 to $450.00, which is a guess.  It would depend on whether or not they considered these one resistor series or two.  It would also depend on if the TCs were considered to be different (which they are).  For slightly less TC accuracy (and less cost), you could ask for a TC of 1 PPM for all 7 without any matching.

[cncjerry]

I'm running a simple test right now.  I took a 10k resistor and paralleled it with an 10M and a 3.3M, all three selected by hand so that the resulting resistor combination came out to 9.9925K at 68 degrees, which the room holds pretty well.  The 10K is a vishay 1/4 watt fat resistor, can't remember the type, but nothing that special, maybe $2.50 per?  Can't remember but I bought 10 a while back and I could be off by  a dollar.  The 10M and 3.3M are 1/4W 1% metal film.  I installed them in the slot thinking I didn't want to have nothing in there if I switched over that range.  I then ran a stdev test using one of my HP 3457a meters and it came out to 6.1E-6 over about 90 minutes.  The room it is in is very noisy.   TV is on, lights, all on the same circuit.  I than switched this resistor out of the circuit (it is shorted during the 0,2,4,8,10 ranges for that decade) and ran the same test over 90 minutes.  It came out 6.04E-6, slightly better but not that much.  The best I've seen on this test with this 332D is 5.0XE-6.  This meter generally has a stdev of 7E-9 (or better) with leads shorted. One thing I have to check is the average voltage reading because it looks like when I switch this resistor on, that the last digit as in 10.0000X goes up 1 count but it could be 1/2 since the meter is only 6.5 digits unless you use GPIB or statistics.  So the voltages are 10.00001, switching in the decade 10.10002 and switching it out again 10.20001. I think I can tweak this but will wait to see if someone has a part. 

Given that the best meter I have is a 3457A (setting aside the 3456A meters I have), I would be happy with it the way it is.  I plan to get a 3458A sometime this year or a high-end Keithley, calibrated, and then immediately calibrate the 332D.  Then I can calibrate the other 20 meters I have in my spare time.

Thinking about the string, if only one decade is out of calibration, can't you just tweak that resistor, in this case by using R8, so that the 1uV range matches the other decades?  I don't know much about these dividers, as you can tell from that question, but if switching that decade in (which has its own adjustment) and the last digit changes as well as the decade, of course, wouldn't that allow me to keep it in sync with the others?

Still waiting for someone to say, "hey, I have dozens of those resistors, here, have a handful!"

Thanks for the contact in Florida.  I don't think I'll go that route.  I was kicking around winding my own though as I have some of the wire.

[Dr. Frank]

It's a pity, that both resistors are defect, obviously.

The reason, that there are two 19.985k, instead of one 9.9925k, is the T.C. matching.
One has a positive, the other has the same, but negative T.C., so that the combination is matched to less than about 0.25ppm/°C.

It's very hard to impossible, to get such matched pairs.
Maybe Edwin Pettis may produce such a matched pair for a decent price.

The other possibility would be to get a guaranteed low T.C. type from Vishay, like the VHP 101, which have about 0.3ppm/°C, specified as 10ppm Change over a 30°C window. Costs around 50$, but at least 3 month delivery time.

For linearity calibration, read the Manual first. In short, you calibrate the unity resistor of the third deck against the "10" value of the fourth, and then consecutively work upwards.
The calibration for the 332/335 is quite complicated, as it does not feature an internal Bridge.
The linearity can be calibrated to better than 1ppm.

This requires a 720A, or a 3458A.
The linearity of the 3457A or other 6.5 Digit DMMs is also about 1ppm, if you use it on a fixed range (10V), but that's not sufficient to bring it to the full specification.

Frank

[Edwin G. Pettis]

A bit off the main subject, but most of the PWW resistor failures that have interwinding turn shorts was caused by bad enamel, one major wire manufacturer in particular was found to have this problem and it caused a lot of trouble years ago.  Unfortunately it was mostly covered up in that the cause was never mentioned publicly.  The root problem was that the solvent that was used to thin the enamel was not being completely removed from the enamel and the residual eventually caused pin holes and in some cases even short lengths of enamel to eventually fail thus causing intermittent or fixed shorts between adjacent turns.  It was left up to each PWW manufacturer to 'discover' this flaw and figure out what to do about it.  As noted, using a hermetic seal did not help in this case.  These resistors when discovered can not be repaired and must be replaced, as mentioned here, the 'short' is variable and attempting to fix it by whacking the part will not fix this problem.

[Kleinstein]

If this is a kind of series fault, chances are that the other matched resistor also have a higher risk of failure.   So I would not spend very much on a repair if thus unit.
A first thing would be a test with an of the self resistor (e.g. 0.1%), just to see if there is an other defect.

Than one might consider checking out the adjustment procedures for that resistor. If this easy and accurate possible find a suitable, no so expensive PWW replacement even this means a more frequent check / adjustment of that resistor and maybe a limited temperature range.

[cncjerry]

That's what I did, made up a resistor and did some testing.  I also checked the others in this batch and they all measure to the limits of my 3457a.  The resistor in question has an adjustment pot just for that decade and it hits the odd settings only.

I have an extra 3456a that I had never really used much so I hooked it to the 332D and calibrated the DC volts.  I only had to tweak two ranges as the rest were exact.  This doesn't say much, just that the 3456a and the 332D agree.

I then hit a couple of settings for the heck of it. I'm going to keep this meter next to the 332D to see what happens.  I only paid $75 for the 332D so even if it turns into parts, I still think it was a deal.  It seems very, very stable even the decade that I had to bodge-up, at least good enough for transfers from one 3456a to another or even my 3457s.  I have 3 or 4 3478a meters and other Keithleys that this is more than good enough to use for adjustment.

If nothing else pops on it, I can wire up three .01% resistors (10K in parallel with a combination of 2 making 13.32M).  That will get me close enough to adjust it by the next decade.

[cncjerry]

Update on the 332D.  I've had it running now for a week and find it to be stable.  Tonight I brought a 3456A meter up and put it on 3.00000 output and hit the Stat button on the 3456.  After 2,000 samples at NPLC 100, the variance (stdev) is 00.0000.  The 3456 will drop into mV so if there was some jitter or drift it would read something like 0.000001E-3.  I've run the same test using a 3457A and 3mV and the Stdev was 190.xxxxxE-9 after several hours.  I had started another thread about noise in measurements and found that using shielded twisted pair cable really does help with noise, lowering the stdev about 10-15% over 24 hours.

So the 332D is stable  but dangerous!  I've been zapped by 300V several times.  I had the external cover on without the D-bolts fastened which leaves the cover at high voltage, up to 1100V.  I leaned on the cover at 300V and got quite a shock.  I then grabbed the banana jacks once and this morning, I unplugged the bananas with 300V dialed in.  When the +V jack came out it bumped against the side on a 3478a meter I was checking as that meter has inset jacks.  The spark started and as I pulled away, the arc grew to about an inch and there was enough current at 50mA peak to cause a piece of the jack to fly off and hit my face.  Needless to say, I now move the meter into standby before touching anything.

I don't understand how, and I haven't looked at the manual, or why the external shell is at high voltage when not bolted down.  The external case has small insulating pieces of nylon that prevent the cover from touching the boards when you slide it on but I guess the insides are floating above ground and must be touching the case or something.  Anyway, 300V gives quite a shock.  I had dialed it up to 1KV into the 3456A and now I am really glad this didn't happen then.  We get used to working with low voltages and I think I was complacent.  Guilty as charged (or shocked).  As someone once said, "thought you were an engineer and you're only a conductor."

Frank, thanks again for sending the manual and help.

Jerry 

[rastro]

... So the 332D is stable  but dangerous!  I've been zapped by 300V several times.  I had the external cover on without the D-bolts fastened which leaves the cover at high voltage, up to 1100V. ...
Jerry

This is very concerning.  I'm going to refurbish a Fluke 335A in the near future and is definitely something to verify and wring out any problems before proceeding further.  As I understand the 332 and 335 have a lot of similarities - at least on the voltage output section.

Have you checked the power plug's earth ground to outer case on the unit when it is unplugged?
- My 335A shows a short as I would expect.
- I did notice the plug prongs are brass/copper and had developed some corrosion/oxidation.  It seems these need cleaned or could cause connection problems.
- I checked the case continuity at one of the rivets where the paint was worn off;  However measuring the plug to the metal handles it showed a flaky connection - it should have been also shorted to the case.  However this may be the screw contacts with the handle.

I also did the outer-case-to-earth-ground test with the 2 large zuse fastners on the rear of the unit disengaged.
- It shows intermittently open.
- Moving the case slightly out a crack made the outer case float from the plugs earth ground.
- The zuse fastners are quarter turn to lock.  If the metal bar they engage with is broken this will probably leave the return path open - causing the case to float.
- I noticed at least one of the zuse fasteners on my unit needs repaired.  These are easy to damage if someone treats them like a regular screw.

-rastro

[cncjerry]

With the connectors screwed-in, the outer case is dead.  I used a meter to confirm it.  There is a label in the unit that warns if the outer case is not screwed on then it could be at the full 1,100V potential.  There are nylon insulators on the inside of the outer cover that should prevent it from touching the inner cover.  Also, the front panel has insulators I believe from the inner cover.

On the removable panel (hiding the adjustments) there is a label:

Calibration Access
Warning:
Do not remove this cover.  Refer
servicing to qualified personnel.

This cover must be removed
prior to removing the outer shell.

Shock hazard exists if
outer shell is placed on
earlier instrument that has
inner cover without four
rectangular clearance holes
for access cover fasteners.

The inner cover must act like the guard for a triax where it is at the same potential as the output connector.  If you have the outer shell not fastened and it touches anything on the inside (like hanging connectors, maybe) then it will be at full potential.  I was getting sparks from the outer cover to the front panel prior to screwing-in the connectors when I had the voltage about around 1K. Since bolting the entire thing together and measuring everything, it is all at ground potential now.  The advice I would give you is to never unplug the output connections unless you have the unit in standby.  I would also be very, very careful using banana plugs that could pop out (spring loaded types) or those that fall out.  These units are lethal above 300V and will easily go to 1100V at enough current to kill you.  I got it twice at 300V and in neither case did it trip the current sense or voltage sense circuit.

[rastro]

It sounds like you resolved the shock issues with the outer case by engaging the cover fasteners.  I suspected this was the causing the cover to float to a high potential. 

I'm glad you posted.  It highlights a potential problem with these instruments.  Those 2 fasteners do much more than hold on the rear cover.  They are critical for electrical safety.  This design is a very flimsy way to insure the safety ground connection to the cover!
Fluke should have also put a caution label on the outer case to insure that the 2 rear fasteners are engaged before applying power. 

If the outer cover is properly connected to earth-ground (through the 2 Dzus cover fasteners) it should never have any voltage potential on it.  Even if the guard voltage from the inner case became shorted to the outer case the earth-ground would short out any potential guard voltage present. 

I like to use good quality banana jack jumpers (Pomona/fluke) to insure dependable connections.  Also, I certainly wouldn't unplug any "Live" leads/jumpers with anything even approaching 100V's.

-rastro

[cncjerry]

There is a small label, about 2CM^2 on the back that says you must have the power line ground connected.  I wonder if the case would float without it?  People like me were dumb enough to use the three prong to 2 prong adapters years ago with a cheap extension cord.

And by the way, you know Rastro was born Tralfaz?

[rastro]

I think the ground from the slotted prong is tied to the earth-ground back at the breaker box.  It makes a big loop but they are at the same potential.  If the earth-ground is disconnected through an adapter it would defeat this connection.  So you would again have the outer cover floating - not so good.