HP 8662A power input protection devices - identification

[Bill Woodbridge]

Does anyone have any information please on two circuit protection devices on the HP 8662A power input, which aren’t shown on the schematic? 

The first, presumably an NTC thermistor for inrush current, is in series with the primary of the oven transformer.  In fact it and its leads form the connection between the transformer terminal and the 240V neutral output of the voltage selector switch.  It’s a disc about 10mm diameter, tightly enclosed in black heatshrink, solid ?nickel leads with clear insulation.  It’s 33 ohms out of circuit, room temperature.

The second sits between the live output of the filter and the switched 240V neutral from the voltage selector switch.  So clearly not a thermistor – maybe a MOV for spike protection?  Very similar appearance to the first, but the diameter is around 20mm.  Out of circuit it reads as open circuit on the meter, but presumably it does its stuff only when much higher voltages are applied.

I ask because I’m having to repair the power input section of my new HP 8662A after a corner smash in transit (two broken fuse holders (main and oven), a smashed main fuse just holding on by its wire, several distorted small covers and panels).  HPs legendary build quality, so evident everywhere else in the instrument, seems to have deserted them in this unglamorous corner - it’s a dreadful bird’s nest of mains wiring running between the rear panel and other sub-panels in every possible direction, several exposed terminals sitting just millimetres from various bits of the case and sub panels, and all crammed into too small a space.  I did all the replacements and repairs, powered up and everything seemed happy.  But I’d forgotten to fit the small internal metal cover that hides the mess of wiring behind the power panel. This cover was quite a tight fit due I think to a bit of residual distortion of the geometry around that corner, but in it went and I checked for no trapped wires or shorts.  Or so I thought – on the next power–up I got a flash/bang from somewhere behind the power panel, a little smoke and the house circuit breaker tripped leaving both fuses intact.  Investigations continue, but I want to be sure of what I’m dealing with for these two devices, and also to check they weren’t destroyed (they both still look perfect but I don’t want to disturb their heatshrinks if I don’t need to.

This is a UK-spec 240V operation unit, so I wonder if there was some official mod for the UK market that isn’t on the schematics I’ve seen?

Thanks in advance for any help, and apologies if they're mentioned somewhere else in the full manual, which I don't yet have a copy of.

[George Edmonds]

Hi Bill

It is a long time since I worked on a 8662A, but I do have the printed service manual for the power supply.

There are two NTC devices used, one designated RT2 is in the primary of the transformer, the second designated RT1 is in the input to the rectifier assembly it should measure 5 Ohms at 25C

In the past I have found that at times with the power supplies the 8663A service manual is the more accurate/relevant one when working on a 8662A power supply.

G Edmonds

[Bill Woodbridge]

Thanks very much indeed George – I think your valuable clue about the 8663A documentation has unlocked it.  In the 8663A Service Manual (once I’d navigated the multiple parts, changes etc) I finally found a power supply diagram which included both RV1 ‘varistor assembly’ and RT2 ‘thermistor disc 30-ohm TC=-3.9%/C-deg’ exactly as my real-life 8662A circuit has.  Plus a note under ‘Instrument Changes’ saying ‘At instrument serial prefix 2510A, RT2 was added as a recommended retrofit when fuse blowing to current surge transients during 230 volt operation occurs’.  I didn’t find RT1, but I think that’s probably further downstream in the power supply than I need to go.

That encouraged me to look again in the Manual Changes section of the 8662A Service Manual Part 1 (I haven’t got Part 2 with the schematics, which seems elusive online, but fortunately all of the enormous change list is at the front of Part 1).  After trawling through I eventually found ‘Change 24’ adding RV1 to the parts list and schematic, then ‘Change 41’ doing the same for RT2, with the same explanation.  So mystery solved and of course I should never have doubted the quality of the HP documentation.  Although it did make me wish briefly that I was still an engineer in a professional lab where junior technicians kept on top of the mountains of documentation changes like in the old days, rather than just a bloke in a garage ….

Thanks again, now to get on with actually finding the source of the explosion.

[George Edmonds]

Hi Bill

A PDF of the schematics can be found here:-

http://www.ko4bb.com/manuals/86.157.159.24/Agilent_HP_8662A_8663A_HQ_Schematics_collection_only.pdf

The 8662A/8623A power supplies are well known for failing,  They normally blow the input fuses, but the RT1 NTC device also gets destroyed.  As they are switch mode most attempts to fix them result in tears and a lot of hot copper and destroyed semiconductors.

At one time there was a considerable number of them for sale on ebay minus the power supplies.

It is just possible to fix them using Mean Well modular supplies.

I am well on my way to celebrating seventy years of fixing electronics.

G Edmonds

[Bill Woodbridge]

That’s exactly the source I have been relying on up to now, thanks – plus videos from the various 8662A Youtube stars, which sometimes fleetingly show schematics.  But the summarised schematics-only file in that link doesn’t show or mention RV1 or RT2 as far as I can see.

Just to finish the story: I was lucky as I didn’t fancy delving into the actual SMPS at all, as you say.  The problem turned out to be a dry joint on the small A10 board that sits just behind the enclosed mains filter – one of the blue neutral pads in the bottom corner.  It had arced (perhaps it had been doing this for a while) and I think then flashed across to the case, tripping the house RCD.  There was a bit of blackening (hard to detect when most corners are pretty filthy anyway) and a spattering of tiny solder globules on the back of the board.  Why it chose that moment to give up so I don’t know – I guess the impact of the smash disturbed the bad joint and I’d been lucky up to that point.

So I made that good, but failed to notice that the short PCB trace had also been vaporised.  So I had reassembled the whole intricate mechanical puzzle (A10 board & filter, oven transformer, mains sub panel, external connector sub panel, fan, HPIB mount and about five more covers and brackets) only to find the unit was completely dead.  Out everything came again to find and then make good the missing trace with a short copper jumper.  Then everything went back again, and all seemed well on preliminary checks until when wobbling the IEC lead to disconnect it, there was another flash and bang somewhere deep in the birds nest of wiring.  That was more difficult to find – it turned out to be one of the flying leads of RV1 which had fatigued away with all the constant prodding and poking.  I was getting thoroughly fed up by this point, and simply removed RV1.  Then on the next reassembly (this time after even more checks for continuity, earthing, insulation with a Megger etc) everything powered-up and finally seemed to be stable.  So the 99% of beautifully-engineered complexity in the rest of the instrument all worked fantastically well – there appear to be no faults whatsoever, much to my surprise – and the 1% of very simple but badly put together mains circuitry caused no end of trouble.

All in all a very frustrating corner of the machine to work on.  I found that more or less everything on the list above had to be removed, or at least loosened, to get any degree of slack on the cabling to allow any work to be done.  Every time I remade the connections and leads I tried to be as professional as possible with neatness, sensible routing and heatshrink where possible.  But every time the whole mess of wiring had to be put back, there was no alternative to pushing and pulling everything around quite roughly to fit, and just hoping that the connections and leads were healthy enough to withstand the treatment.  Which on the final attempt they were.

Thanks again!

[KE5FX]

Generally the Artek manuals will be better quality than anything you find for free online.  They are well worth the small cost.

If you haven't swapped out the line filter assembly itself, that'll probably be the source of your next explosion.   Typically the module is made by Corcom or one of a small number of other vendors.  The capacitors are epoxy-sealed, but the epoxy doesn't always hold up well over multiple decades.  When moisture eventually makes its way inside, it's just a matter of time until a loud/smelly failure.  Almost all HP equipment from this era is vulnerable, especially on 240V circuits.  We don't see anywhere near as many failures over here.

[Bill Woodbridge]

Thanks for that thought!  I did indeed wonder and worry about the health of the Corcom tin box, but I haven’t replaced it.  It looked to be a classic example of unobtanium, although a quick scan of Mouser just now brought up some fairly similar looking Corcom units.  I haven’t dived fully into the detail, but perhaps things aren’t as bad as I'd assumed if I did want a new one.  I can’t face another disassembly yet though, let alone a possible mechanical mod to get a new unit to fit, so I will probably wait for the next explosion if it happens.

And an Artek manual set is now on order, thanks for that link.  Usually I try to scrape by with free online documentation, but the 8662A doesn’t seem to be well catered for in that regard, and it certainly deserves to have the proper manual set rather than the bits and pieces I’ve managed to glean up to now.

I feel quite proud of getting my 8662A running again, although the repair was of the simplest possible kind (apart from the arduous mechanical wrestling).  I wondered if there were any more ‘iconic’ HP lab-grade instruments that stand out as trophies for tinkering amateurs these days?  The 3458 bench multimeter springs to mind, although that’s from a later generation and also prices are in a different league.  Also the 8566 / 8568 spectrum analysers, which might be next on the radar, although the UK choice is very limited compared to the availability in the US.

[KE5FX]

Yeah, the 8662A is the Ferrari of signal generators... specifically the F355.    Keeping it running well will challenge both your skills and your sanity.   The 8566/8568 boxes are easy to deal with in comparison, being both less complicated and less prone to intermittents.

A key advantage of the Artek manuals is their OCR quality.  You can run the free manuals through an OCR program, but subsequent searches will miss a lot of references because the image quality wasn't adequate to begin with.  When working on my 8663As I've always found both the Artek files and the paper manuals to be helpful, one for searching and the other for study and markup of the foldout schematics.