HP 54510A Power Supply Smelling

[Paul Ramsbottom]

I understand that the power supplies on these 'scopes are troublesome. Mine is actually working but is giving off an unpleasant burning smell. Information about these units seems scant but there is a reverse engineered circuit diagram here http://www.radiocollection.be/images/restaurations_img/HP54503/supplyHP54503.pdf

The track between CR11 and R16 seems burnt.

My immediate thought was that leaking electrolytics were causing the problem. It isn't easy to isolate the secondary side sections except for -12v one starting with C32 as this has a lead connecting it to the transformer. I disconnected this but the smell persisted so I went on to replace C7 to C14 as there were more of them. This didn't help either and there is no sign of distress in any of them.

I really don't want to proceed further with a blanket replacement of caps and I wondered whether anyone here has any experience of these or any suggestions for a more scientific approach.

I repair valve radios as a hobby so have some experience with electronics but the limit of this is circa 1949! I certainly haven't tackled anything like this.

Any suggestions would be gratefully received.

Thank you

Paul

 

[73's de Edd]

Sir Paul Ramsbottom . . . . . . .


GEE-OH-GOOD-GOSH--GOLLY-DAMNED-ALL-A-MIGHTY ! . . . .if THIS is what you are using on feeble OLD tubed radios . . .don't you DARE let us see what instrumentation you are using on Tran-sist-um-oh-ferrized and Interly-Grated-Circuitious designed units .

You are in a completely different type of power supply from the rest of the units linear based units, when you are . . . .on page 5.
If it seems that the R16 / CR11 pair are running hot you just might have a drive situation that is producing less than a fully saturated condition on Q3.
That would cause a decline in driven power output from the windings on that T2 power transformer.
Then the outputs feedback loop, via that A9 optical isolator, would tell the Q1 unijunction to up the power drive to the Q3 power transistor.
Initially, see if any or all of the derived voltages off T2's secondary supply windings are up to spec.

ANALYSIS:

Pick a given run time in minutes and evaluate the touched temp of both R16 / CR11 . . . .make mental note.

The two most suspect parts in that base drive supply loop would be a t . . .i . . .m . . e . . . related capacitance decline or ESR onset in electrolytic caps.
( Remember . . . . . those little puppies have acquired some age on them. since day one of circa 1991. )
In this immediate circuit, that would be the coupling provided by C12 . . .470ufd / 35VDC cap and the slightly less suspected C46 . . .33ufd . . .just by virtue of being a tantalum unit.
The same conditions on C22 and C23 could be at fault.
(However. . . .if those happen to be some ORANGE units they are being some premium units, that tend to be a bit more bullet-proof with age onset.)

Now if you don't have them on hand, the sub of a shunting pair of 100 ufd units across them should give you some answer, or you could also read the AC ripple . . .or any decline . . . across them after being put in circuit.

Thassssit . . .for now . . . . .

Sub in some other units and do that same run test and then compare those R16 / CR11 parts temperatures.

MARKED UP SCHEMA REFERENCING:




73's de Edd

 

[Paul Ramsbottom]

Wow, thank you! The 'scope is actually my son's but it seems to have taken up residence here since it started to smell.

Just one thing before I start - is it safe to power up the power supply whist it isn't connected to the rest of the scope? If I powered up a valve radio with the output valve removed the smoothing caps would remain charged which might prove a bit surprising if I touched them. There are an awful lot of electrolytics in this power supply and I would rather not take any chances.

Here is a photograph of the component side of the unit by the way.

Thanks

Paul

 

[73's de Edd]

Sir Paul Ramsbottom . . . . . . .

It would definitely be best to leave the PS connected and be getting loaded down by the circuitry.
You know about the high voltage associated with Q3 from your "OLD tube" know how.
I make connections with clip leads or solder tacked on wires and thereby, that lets me " remote " those meter readings, for added safety.


73's de Edd

 

[Paul Ramsbottom]

Ok, thank you. I think I will order in some caps before proceeding. The tantalum has markings of "33u 10" on it. Does that suggest 10v rating?

Paul

 

[73's de Edd]

yep

 

[Paul Ramsbottom]

I have to admit I didn't quite follow your instructions as I was too scared to power it up again with the fault. Anyway the capacitors arrived today and I started by replacing the 470uf cap which I think is actually C17. That seems to have cured the fault. I say "seems to" because I haven't tried it back fully in its case yet. My son has a thermometer which you just point at whatever you are taking the temperature of and the area around R11 is steady at around 31C which seems ok. I have also replaced the tantalum cap with a higher voltage rated one.

I ordered 560uf caps for C22 and C23 as that was the value shown in the circuit diagram. Those fitted are actually 680uf I now see. Should I just go ahead and replace them with the new lower value ones?

Very many thanks

Paul

 

[Paul Ramsbottom]

Just poking around after I had turned it off I have found that R10 is getting pretty hot. Is that to be expected?

 

[73's de Edd]

.,

Sir Paul Ramsbottom . . . . . . .


I ordered 560uf caps for C22 and C23 as that was the value shown in the circuit diagram. Those fitted are actually 680uf I now see. Should I just go ahead and replace them with the new lower value ones?
Since you probably don't have another scope to check a scope with , instead, use your VOM in AC mode to measure any AC ripple level across each filter to even see if they are " complaining" any ?
You just might have an old Triplett 630 series of analog meter, which you would use in the OUTPUT input, to have DC isolation afforded by the series input capacitor used in that mode. Or a Simpson 260 says ME TOO !


Just poking around after I had turned it off I have found that R10 is getting pretty hot. Is that to be expected?
It and the companion 200 ohm resistor up above it, take the combined power pulse train from the Qi unijunction
in that power loop.
Measure the voltage across it, down to floating ground and use "Ohmega" law to compute the power level it is actually being subjected to, and now actually being encountered.
Our quixotic touch sense sometimes makes us think that something is running hotter than it actually is.
This particularly shows up on newer spiral surface carbon resistors; in their being compared against the old carbon molded composition resistors, where you have that dense shroud of insulative bakelite surrounding the central carbon billet..

73's de Edd



.

 

[Paul Ramsbottom]

Thank you Edd. I have to admit I am a bit scared of probing about with the restricted movement it has. I certainly don't want to cause any damage now it is working so well. I have temporarily refitted the power supply board back into its case and I cannot detect any smell. Examining it again I don't think that R10 is getting over hot, it was just a bit of a surprise when I first touched it.

With regard to C22 and C23 I would really like to replace them as that would more or less have done them all in the primary side. My enquiry was whether the 560 ones are likely to be suitable in view of the large tolerance in electrolytic capacitors. 200v 680uf ones with leads rather than snap in and of a known make seem fairly rare in this country.

 

[Paul Ramsbottom]

I decided to try the lower value caps yesterday and discovered that the old ones are snap in types. I have therefore ordered some Vishay replacements of the correct value together with a new filtered power input socket as I believe that the original ones can self destruct.

I am very grateful indeed for the invaluable help I have received here. The 'scope had come very close to a one way trip to the local recycling centre and now it will hopefully give many more years of service.

Thank you!

Paul

 

[73's de Edd]

What would I do . . . . .
I personally would go with the 680 ufd units and I would be for the 105 C more beefy and ruggedized internals.
I definitely prefer the snap in type due to the infinitely superior seal afforded by them as compared to the leaded types.
I also prefer the Mil spec / NASA / Aerospace 5000 hr rated ones with their "wetter" electrolyte loading of the dielectric.. . . . . . BUT those are typically special order.
However, some show to be stocked by suppliers down in the 2000 hr offerings.

73's de Edd

 

[Paul Ramsbottom]

The ones I have ordered are Vishay/Philips 3000 hour 105 c ones which seemed ok. It actually belongs to my son who imported it. It has a sticker on the side saying it was a demonstration unit from HP Great Lakes area.

Thank you again

Paul

 

[Paul Ramsbottom]

I wonder whether I could have some more help with this. It worked fine when I left it on a soak test for 5 or 6 hours so I gave it to my son for a project he is doing. After he had used it for a few hours it started to cycle on and off. I have only just had a chance to look at it again and I noticed some arcing at the base of one of the old caps. I assumed that was because of electrolyte which had leaked out of it so I ordered a complete set of replacements. I replaced this one first of all and cleaned underneath it before |I replaced it. There was still some arcing so I scraped around it to remove any carbon deposits and that stopped it. It still cycled on and off so I set about replacing all the caps. I tested it every time I had replaced a few and the problem remained. I have just replaced C19 and now all that happens is that the fan turns very slowly. I am getting only low varying voltages on the output pins.

I have checked my soldering and cannot find any mistakes so I am assuming that I have disturbed something when I have been handling it. Are there other places I can test voltages to try to pin down the problem? Having come so far with it, it would seem a shame to give up with it now. I can post the actual readings on the output pins fi that helps.

Many thanks

Paul

 

[Paul Ramsbottom]

As a matter of interest I have taken voltage readings on the output pins with no load as follows:

1-4 5.17v (5.15)
9 3.44v (3.5)
15 -12.2v (-12)
17 1.5v (12v)
19 1v (15.5)
13 12.08v (12v)
11,18 3.5v (5.2)

All the values were with respect to earth, but pins 11 and 18 don't seem to be earth referenced so maybe the value there is ok. How can I check this properly? I cannot find any problems with C 30, 31, 33 or 34 but there cartainly seems to be a problem with that line. Any suggestions would be very helpful.

The values in brackets are those according to the circuit diagram.

 

[Paul Ramsbottom]

Under load the values drop a lot and become very unsteady. Pins 1-4 become 2v. 3 also 2v. 15 = 10v 17= 1.5v. 13 = 10.2v. 11, 18 = 0.3v. Those are average values as they constantly vary.

 

[73's de Edd]

.

Sir Paul Ramsbottom . . . . . . .

I am placing on some initial Marked Up references to be consulting , but you query:

All the values were with respect to earth, but pins 11 and 18 don't seem to be earth referenced so maybe the value there is ok.

So my very first question / action would be for you to confirm . . . . .

We can see that on J2 connector that using an OHMMETER in lowest . . .ohms. . . range should be revealing that pins 5 thru 8 should be interconnected.
Reaching over to measure pin 16 and pin 20 should also reveal a like interconnecting .

NOW . . .we figure out the grounding question . . . . by initially noting that the +12 VDC analog supply as well as the MINUS 5,2 VDC supply are using an isolated T3 transformer for their power derivation.

The derived voltage from CR26 plus C40 and C41 is generating a straight forward supply with only a slight
variant in the respect of using Q13 / BUZ71A series power FET as a series switch in delivering a slight delayed timing of full powering up of that supply..

Lets revert back to the ohmming function and doubly verify that the C42 final filter cap of that +12 VDC analog
supply supply is having its NEGATIVE lead making a connection to our earlier cluster of grounds.

If so, then we should expect that the other MINUS 5,2 Analog supply should also be tied into that common buss.
BUT here is its difference, it is using CR27 plus companion filter capacitors C15 and C16 for power derivation
and has the created NEGATIVE voltage flowing on down from C15 and C16 to the next C? filter (get that number for us) and finally down to the last C50 filter.

So you now want to test the POSITIVE connection of that C50 filter to see if it also is grounded to all of all of our other grounds.
If so, that clears up all of the ground interconnect verifications.
The only variant now is the design of that last negative 5,2 Analog supply, if normal, the cathode of the CR27 and the shared interconnects to the POSITIVE leads of C15 and C16 would go straight to ground.
Not so in this design, as they are wanting a current sample of this supply and are using series inserted R43
( 5 watts, no less . . .babycakes ! )as a current sampler, with its shunt voltage take off fed to the base emitter junction of Q11 and then feed back to the control board thru Q11' s emitter.

That was the first slight disruption of supply flow, getting to ground , with the next one being the Drain -Source
disruption at Q14 /BUZ71A series power FET, which is dependant upon gate drive from the control board at P3-12 for making that final ground connection of this circuit loop.

I really think that this supply drop is associated with the fault at the pin 17of the supply +12 VDC Display source supply voltage , being now reduced to 1.5 VDC and the MAIN fault, is with the +15,5 at pin 19 of J2 now only being 1 VDC.

I think that a voltage or current sensing fault from a supply level has been detected and fed to the control board and it has activated the gate of the A2 SCR and thereby crow barred that supply.

Most suspect would be a long time related decline in capacitance of filters.

I forget if you have capacitance reading equipment or ESR evaluative capabilities.

Also I think that your voltages might be stable with no loading placed upon the unit, less our major 2 supply voltages and the minor decline of 5,3 to 3,5 VDC on the last one.

On the bottom schematic you can see 3 BLUE ARRROWED mark up routes that can avalanche their respective zeners and feed gate drive thru R11 to activate A2 crowbar shutdown.
Its imperative that the YELLOW marked cap C29? has not had signifiant overall capacitance decline or its
reduced time constant will make that leg of circuitry hypersensitive / erroneous.
Same situation over at C28, associated with a clean stable reference supply for the LM339 comparators .

The GREEN ARROWS mark ups are related to the other voltage lines and current samples that are fed into that A2 SCR gate circuitry via the Q7 transistors C-E junction.

Can you supply R11 and R42 resistive values.

I am most suspicious of the four orange electrolytics in the unit.

Are your power semiconductors, that are being heatsinked, still having silicone wetted unions/bonds to the
heat sinks . . .OR . . . has the moist / wet silicone element, long since migrated laterally, and left the devices
sitting high and dry on a thermally isolated, bone dry layer of tin oxide ?

Will continue on TS procedure . . . after your info fill ins.


TECHNO ASSEMBLY REFERENCING . . . . . # 1

TECHNO SCHEMA REFERENCING . . . . . # 2

73's de Edd

.

 

[Paul Ramsbottom]

Thank you Edd. I will look at it and let you have a proper reply over the next couple of days.

Paul

 

[Paul Ramsbottom]

Thank you once more Edd. I am sorry for the delay in replying. My replies to your questions are on the quote below. I think you may be asking me to get some new tantalum caps next.

Paul

So my very first question / action would be for you to confirm . . . . .

We can see that on J2 connector that using an OHMMETER in lowest . . .ohms. . . range should be revealing that pins 5 thru 8 should be interconnected.
Reaching over to measure pin 16 and pin 20 should also reveal a like interconnecting . Yes they are connected

NOW . . .we figure out the grounding question . . . . by initially noting that the +12 VDC analog supply as well as the MINUS 5,2 VDC supply are using an isolated T3 transformer for their power derivation.

The derived voltage from CR26 plus C40 and C41 is generating a straight forward supply with only a slight
variant in the respect of using Q13 / BUZ71A series power FET as a series switch in delivering a slight delayed timing of full powering up of that supply..

Lets revert back to the ohmming function and doubly verify that the C42 final filter cap of that +12 VDC analog
supply supply is having its NEGATIVE lead making a connection to our earlier cluster of grounds. Yes that's right

If so, then we should expect that the other MINUS 5,2 Analog supply should also be tied into that common buss.
BUT here is its difference, it is using CR27 plus companion filter capacitors C15 and C16 for power derivation
and has the created NEGATIVE voltage flowing on down from C15 and C16 to the next C? filter (get that number for us) and finally down to the last C50 filter. C19 100uf 10v

So you now want to test the POSITIVE connection of that C50 filter to see if it also is grounded to all of all of our other grounds. Yes it is
If so, that clears up all of the ground interconnect verifications.
The only variant now is the design of that last negative 5,2 Analog supply, if normal, the cathode of the CR27 and the shared interconnects to the POSITIVE leads of C15 and C16 would go straight to ground.
Not so in this design, as they are wanting a current sample of this supply and are using series inserted R43
( 5 watts, no less . . .babycakes ! )as a current sampler, with its shunt voltage take off fed to the base emitter junction of Q11 and then feed back to the control board thru Q11' s emitter.

That was the first slight disruption of supply flow, getting to ground , with the next one being the Drain -Source
disruption at Q14 /BUZ71A series power FET, which is dependant upon gate drive from the control board at P3-12 for making that final ground connection of this circuit loop.

I really think that this supply drop is associated with the fault at the pin 17of the supply +12 VDC Display source supply voltage , being now reduced to 1.5 VDC and the MAIN fault, is with the +15,5 at pin 19 of J2 now only being 1 VDC.

I think that a voltage or current sensing fault from a supply level has been detected and fed to the control board and it has activated the gate of the A2 SCR and thereby crow barred that supply.

Most suspect would be a long time related decline in capacitance of filters.

I forget if you have capacitance reading equipment or ESR evaluative capabilities.

Also I think that your voltages might be stable with no loading placed upon the unit, less our major 2 supply voltages and the minor decline of 5,3 to 3,5 VDC on the last one.

On the bottom schematic you can see 3 BLUE ARRROWED mark up routes that can avalanche their respective zeners and feed gate drive thru R11 to activate A2 crowbar shutdown.
Its imperative that the YELLOW marked cap C29? C24 actually has not had signifiant overall capacitance decline or its
reduced time constant will make that leg of circuitry hypersensitive / erroneous.
Same situation over at C28, associated with a clean stable reference supply for the LM339 comparators .
C24 and C28 are tants. Very difficult to get to the back of that board so may have to replace from the front.
The GREEN ARROWS mark ups are related to the other voltage lines and current samples that are fed into that A2 SCR gate circuitry via the Q7 transistors C-E junction.

Can you supply R11 and R42 resistive values. R11 989ohms (1000) R42 359 ohms (360)

I am most suspicious of the four orange electrolytics in the unit. I have replaced all the electrolytics on the main board

Are your power semiconductors, that are being heatsinked, still having silicone wetted unions/bonds to the
heat sinks . . .OR . . . has the moist / wet silicone element, long since migrated laterally, and left the devices
sitting high and dry on a thermally isolated, bone dry layer of tin oxide ? The silicone paste seems quite moist but I could have a go at replacing it if necessary.

Will continue on TS procedure . . . after your info fill ins.

 

[Paul Ramsbottom]

Well, I was hopeful that the new tantalum caps would solve it but they have made no difference. I have noticed a regular clicking noise like that of a relay coming from around the mains transformer. I had assumed it was the fan which slowly turns when switched on but if I put something to stop the blades from turning it continues. I cannor see any reference to a relay in the circuit diagram. though.