HP 54520A PSU repair

[sv3ora]

An HP 54520A scope has a SMPSU that fails.
The scope powers up and stays ON for a minute or so and then the fuse in the PSU is blown.
The PSU is very similar to the reverse engineered attached PSU.

I have also read this thread https://www.eevblog.com/forum/repair/oscilloscope-power-supply-burnt-and-strange-clicking-noise/ and the pictures of the PSU seem exactly like mine. https://www.eevblog.com/forum/repair/oscilloscope-power-supply-burnt-and-strange-clicking-noise/
https://www.eevblog.com/forum/repair/oscilloscope-power-supply-burnt-and-strange-clicking-noise/?action=dlattach;attach=959176;image;PHPSESSID=3qcb91vb3mdch8tlvkq629c9k7

I have replaced the primary big electrolytics and the gas tubes but no luck yet. Shall I replace ALL the electrolytics of the PSU? I see no signs of liquid but some of them are covered in silicome at the sides.

There must be a short circuit somewhere caused after a while and the fuse is blown. Can this behaviour be a leaky electrolytic?

[Tomorokoshi]

I have some data I collected when I repaired the power supply in my HP 54542A.

What is the part number on your power supply? Mine is 0950-2369.

History of switching power supplies
http://www.righto.com/2012/02/apple-didnt-revolutionize-power.html
https://www.electronicdesign.com/technologies/boards/article/21795586/robert-boschert-a-man-of-many-hats-changes-the-world-of-power-supplies

Boschert maintenance manual
http://www.bitsavers.org/pdf/boschert/Boschert_OL25_Single-Stage_Power_Supply_Maintenance_Manual_May79.pdf
https://archive.org/details/bitsavers_boschertBogePowerSupplyMaintenanceManualMay79_3604356
https://archive.org/details/bitsavers_teleraytel10SeriesCRTDataTerminalReferenceManualAu_24345013

HP 54503 Boschert power supply
https://www.eevblog.com/forum/repair/hp-1660as-xl130-3604e-psu-problem/?action=dlattach;attach=355506

Related discussions
https://www.eevblog.com/forum/repair/hp-53310a-power-supply-troubles/
https://www.eevblog.com/forum/repair/hp-1660as-xl130-3604e-psu-problem/
https://www.eevblog.com/forum/beginners/fixingreplacing-bad-transformer-in-logic-analyzer-smps/
https://www.eevblog.com/forum/testgear/hp-54540-dumpster-scope-166240/

Do you have an LCR meter that can report D and Theta when testing at 120 Hz? Here is what I measure at the output connector:

Code: [Select]

# Function μF D Θ
1 1 6160 0.757 -52.9
2 1
3 1
4 1
5 Ground
6 Ground
7 Ground
8 Ground
9 2 968 0.101 -84
10 Ground
11 3 990 0.102 -84
12 Ground
13 4 609 0.120 -83
14 Ground
15 5 95 0.043 -87.5
16 Ground
17 6 611 0.128 -83
18 3 990 0.102 -84
19 7 970 0.135 -82
20 Ground


[sv3ora]

Thanks so much!!
Mine is also PN 0950-2369. The 54520a fits exactly the same PSU with yours.
I do not have an esr meter, I just have a capacitor meter and a multimeter.
I was thinking of recaping the PSU completely but this won't be cheap. But I do not have any other means of troubleshouting this, cause it works and then suddenly fails.
Can a bad capacitor cause this blown fuse, after a minute or so of correct operation?

[Tomorokoshi]

It's possible that a capacitor could do that. Removing the capacitors on this supply is difficult. The silicone adhesive is difficult to cut through, and the capacitors are soldered on to very heavy traces. It's difficult to remove them without damaging the board.

The HP 54503 schematic is very close but not identical. However, you can definitely use it as a guide.

Note that a shorted diode can also do it. See the secondary schematic on page 6 of 8, the equivalent to CR31 to CR34, CR25, and CR24. Check both directions in "Diode" mode with a multimeter.

Also check page 5 of 8, the equivalents to CR11, R16, C20, and C21 in the switching circuit for Q3.

You can at least check the capacitance values at each connection at 120 Hz. If they are within perhaps 10% then they may be okay.

My power supply failed such that it was always in "hiccup" mode. I had to work through all the circuits. I removed several parts for measurements. I finally tracked it down to a weak diode on the controller board. It caused the protection circuit to activate as if the output voltage was too high. There was a hint to this in one of the postings. I never had the fuse blow out, so you have a different problem.

Be very careful if you probe on this supply when powered up. There isn't any isolation on the primary side. The secondary side is also difficult.

I slowly ramped mine up through a Variac. It would start to chirp at maybe 30 VAC.

[sv3ora]

Thanks a lot for the Hints.
Ok checking the diodes as well with a multimeter won't be so difficult, desolder one pin and check them.
The caps are tricky but I can manage it.

What is strange, is that it only blows the fuse after a while. so The diodes and caps could be measured ok, but under load, something may be heating. This is why it seems to me so difficult to find the problem.

Can it be a faulty Q3, that shorts after a while (or it's drive circuits in the base). Cause this would explain a blown fuse.

I will do these checks first you suggest, thanks a lot. If you have any other ideas let me know.

[Tomorokoshi]

You can check the diodes in circuit. I recorded those values, so later I can track them down. Then you don't risk damaging the board. They are also really heatsinked in.

Here are some values from my notes:

Code: [Select]

Part Ref Details Instrument Reference Units
Fuse F1 Ohms 0.03 Ω
Thermistor RT1 Ohms 12 Ω
MOV ? Ohms Open Ω
Bridge Rectifier CR1 Hot to Pos Diode mode 0.528 V
Bridge Rectifier CR1 Neutral to Pos Diode mode 0.520 V
Bridge Rectifier CR1 Neg to Hot Diode mode 0.536 V
Bridge Rectifier CR1 Neg to Neutral Diode mode 0.532 V
Bridge Rectifier CR1 Any Reverse Ohms ~10 M Ω

I have other notes but I will need to transcribe them before I can post. I should be able to get that done tomorrow. Anyway, just measure the diodes in circuit for now. Should be around 0.4V to 0.5V in one direction, and something else in the other. Shouldn't be low. For instance, on page 6 of 8, in the -12V ANALOG circuit in the forward direction you might get 0.5V, while in reverse it would be something that doesn't make sense in diode mode. Switch to Ohms mode, and you might get around 10K + whatever equivalent leaks through the LM350.

[sv3ora]

I should say, that the scope does not only blow the fuse but also destroys (really cracks and blows) the mains thermistor!

Ok, here are all the diodes tests that have been done with them connected IN CIRCUIT.

Diode labels are as printed onto the PCB. A few labels that I have checked, seem to agree with the hp54503 pdf you sent me, but I have not checked all of them.

I measured the diode in the diode test of the multimeter, in forward and in reverse and I wrote down the values in the first 2 columns. Then I measured the resistance of each diode in circuit in forward and in reverse and I wrote down the values in the next 2 columns.

I hope this tells you something... ?

[DeepLink]

Working on such SMPS that explodes in a minute, you might want to build youself a timer (like 30 sec)

When you press a button a relay will supply the mains to the SMPS and shut off before it explodes again

Remember to let the SMPS cool down in between (inrush current limiter and so)

Plan your measurements, attach your probes, turn on and observe/record

Work your way through the SMPS

To repair this particular supply, you must first understand it's working principle
And you need a scope to repair your scope's supply

My guess is that the output transistor is not regulated properly and therefor explodes and kills the inrush current limiter and fuse

I might have a schematic which are drawn in Orcad

[sv3ora]

If you have a schematic, it will be very useful of course, so please attach it.

You mean the power transistor in the primary side of the transformer?

I thought that too, cause if this shunts to ground there will be a short which would end up in blowing the fuse and thermistor.
But what could cause such a short in this transistor? Do you suspect a faulti transistos, can it behave like that?
Do you suspect a component on it's base?
Could it be just not enough heat sinking doe to thermal grease dry throughout the year? Could non good thermal conductivity cause it to overheat and short? Then back in life again?

[DeepLink]

Schematic attached (I have done the primary side so far)

My suggestion
1. build a timer (or you be the timer - remember to shut down in time)
2. build a propper load (do not run the supply without load)
    I've attached a suggetstion for a load (total of 44W)
    but build one with what you have laying around
3. attached one or more DMM to the outputs
4. turn on and observe the output voltages
    maybe one output is faulty, so this might give some clue where to look
5. look at the drive of the output transistor Q3
6. Depends on the above points...

[sv3ora]

Thanks so much!
If I do not manage to find the solution I may end up in buying a variac (useful for other projects) . I doubt it is an electrolytic, all seem optically fine and no signs of leaked substances. Unless internally they are bad. Since I cannot test this (it will blow on my face), a list of voltages/currents and pins on this edge connector would be useful if you know them, so as to possibly build a linear psu. It is too pity for the whole scope to go bad just for the psu.

What could cause a Q3 power transistor short after a while? (assuming it shorts).

Because it's base and it's collector ate the only ones connected to the HV side. To blow the fuse there must be a problem there.
CR10, CR12 and CR28 are the components I suspect. If they fail, the base of Q3 may make it to conduct much more, perhaps to the point of short?

[DeepLink]

Yes, a Variac is very usefull (remember to get one that is galvanic isolated)
Electrolytics can look fine, but still be defective. And in these old instruments, you might want to renew them all while you are workin on it

If you drive the output transistor fully on (for a "long" time), it will blow the fuse. But this will not destroy the transistor

So for the transistor to explode (it will short internally)

Either
1. the transistor gets very hot
 - not driven fully on / off (remember that the transistor is used as a switch)
 - overload

or
2. it is subject to high voltage (flashover/breakdown inside)
 - defective snubber
 - wrong input voltage

This SMPS is a flyback type and the transformer kicks back every cycle
a snubber is attached to "eat" this spike
CR16, R16, C20, C21

The connections are shown in the schematic on page 6 (but do check on yourself, as the picture is hard to read)
J2
1 +5,15V
2 +5,15V
3 +5,15V
4 +5,15V
5 GND
6 GND
7 GND
8 GND
9 +3,5V
10
11 -5,2V Analog
12
13 +12V Analog
14
15 -12V Analog
16 GND
17 +12V Dispaly
18 -5,2V Analog
19 +15,5V
20 GND

No need to build a new supply. These old SMPS can be repaired, but it can take a while. At least - you learn a lot

Components to investigate
Snubber resistor R16
Capacitors C17 & C46
Thermistor on B-E Q3 (51 Ohm NTC)

But your best approach is to measure the thing with a scope, while it is running
First measure all output voltages with some DMM's, to see if one supply is shorted

[sv3ora]

All right thanks. I do not have a second scope
I will check these components, and maybe replace them, apart from the thermistor which I do not have and see how it goes.
I will let you know

This 51R thermistor is NTC or PTC?
MTS90A is the one onto the PCB

[akimpowerscr]

I took a look at the diagram and I think I understood the principle of operation of this power supply

It starts with the 120K resistor R12 which supplies a base current to Q3.
As soon as Q3 begins to conduct, a positive voltage induced in the YE winding of transformer T1 confirms the conduction of Q3 by strengthening its base current.

The control circuit of Q3 is then also supplied with a positive voltage by the winding 2 of T1.

With a certain delay adjustable by the potentiometer R7 and by the optocoupler CNY21, the unijunction transistor 2N2647 generates a blocking pulse for the transistor Q3.

The voltages at terminals YE and 2 of transformer T1 are then reversed and Q3 is blocked.

Normally, in case of strong overcurrent, Q3 should be damaged at the same time as the fuse is blown.

That is not the case here.

This seems to indicate that there is over intensity but that it is not very high.

In my opinion, what can cause this is the optocoupler which has lost its gain ... So, first of all, we must try to replace the CNY21

NB: replace also C46 (33µF) and C7 (470µF)

[George Edmonds]

Hi

This is a standard type flyback SMPSU, they are very dangerous to work on without proper precautions and knowledge as there is some 360VDC at high current in many places.  NEVER use a standard class one scope to work on one, all of the primary side, including the circuit ground is sitting at some 180VDC above true ground.

They suffer a  damaging runaway condition if the feedback loop is defective, this includes the opt-coupler.

The best way to prevent damage whilst testing them is to put a 60W to 100W Incandescent bulb in series with the incoming phase, most need a minimum load on the +5VDC supply to work correctly

George

[akimpowerscr]

All SMPS configurations are dangerous, not just flybacks.

Of course, you have to know what you are doing and know the security rules that have often been repeated on this forum.

The use of differential probes or a good quality isolation transformer are essential.

Another precaution that we often forget: do not be alone when working on dangerous circuits.

As for the 100W lamp in series with the AC, all switching power supplies, and more especially with PFC or self-oscillating power supplies such as this one do not accept to operate under these conditions.

[George Edmonds]

Funny,  I have repaired many thousands by using a series bulb to limit the mains current that can be drawn

Fly back types are a particular problem because they go into a run away condition with out the correct feed back loop operation, they then can spay molten copper  if not current limited by a series bulb.

I repair them not talk about doing so.

George

[akimpowerscr]

Funny,  I have repaired many thousands by using a series bulb to limit the mains current that can be drawn

Fly back types are a particular problem because they go into a run away condition with out the correct feed back loop operation, they then can spay molten copper  if not current limited by a series bulb.

I repair them not talk about doing so.

George

Funny...

So far I know, SM power supply with PFC do not work with a lamp in series.
The PFC requires a low impedance AC power supply.

In the case of sv3ora, what could be the point of putting a lamp in series?

With a lamp in series, nothing will happen and this does not help in the diagnosis of a fault as described.

In short, you are recommending sv3ora to do something totally unnecessary.

You cite your experience in repairing switching power supplies.

I will not answer you on this point by quoting my experience because it is not appropriate to use this argument on a world wide forum where there are hundreds of specialists in all areas, including project and repair of switching power supplies

[sv3ora]

It took me a bit long to respond as I was reverse engineered the primary side. Here it is attached.
It is quite accurate with the PDF, but there are some points marked A, B and C which differ.
Particularly the point A, there are no 2 parallel 1R resistors at that point. Maybe the thick PCB track can replace the 0.5R series resistor, maybe not. The fact is that there is no current limiting at all in case the transistor sorts.

The opto isolator diode is connected to the vertical regulation PCB, which is harder to reverse engineer. How can I test this opto? Maybe I can compensate it's supposed loss by trimming R7?

I have not yet investigated the next components for correct values.
Snubber resistor R16
Capacitors C17 & C46
Thermistor on B-E Q3 (51 Ohm NTC)

And I also remember that at a time, after replacing the big electrolytics on the primary and the surge tubes (It might be relevant, it might not), the scope operated once for about 10 minutes before blowing the fuse again.
I thought I have fixed it and closed the enclosure. Then at some time, the scope fuse was blown away.
I just remembered it, it might mean something to you, or not.

[akimpowerscr]

Simply first replace ISO1 (CNY21), C7 (470µF) and C46 (33µF)....

And let's see if it fix the problem or not.....

[SoundTech-LG]

Hi, worked on a dozen or so of these Boschert supplies in 53310A, 1652B, 545xx HP series. Most were easily repaired by the leaking caps that are so common. Unfortunately a few are tough dogs to trouble shoot and repair. The worst case scenario I have seen was continual blowing of the TiPL755A power transistor. If the leakage of the caps is bad enough it travels the circuit board. If left long enough it will actually wick up into the transformers. Measuring leakage across the primary to secondary reveals it. Removing the transformers (not easy) and measuring them, confirms it, along with the soaked connections, and windings. Replacing the transformers cured the issue. Unfortunately it also wicks up into the pots (as mentioned earlier) this can also be difficult to trace. If it was not for the leaky cap issue these supplies would be pretty reliable.

[SoundTech-LG]

Schematic attached (I have done the primary side so far)

My suggestion
1. build a timer (or you be the timer - remember to shut down in time)
2. build a propper load (do not run the supply without load)
    I've attached a suggetstion for a load (total of 44W)
    but build one with what you have laying around
3. attached one or more DMM to the outputs
4. turn on and observe the output voltages
    maybe one output is faulty, so this might give some clue where to look
5. look at the drive of the output transistor Q3
6. Depends on the above points...

GREAT JOB!!! Looks very accurate. The later versions have a short wire across R13 and R14. These would blow and also take out the TiPL755A anyway. Save money modification.

[akimpowerscr]

All right thanks. I do not have a second scope

Since sv3ora does not have an oscilloscope, it does not make sense to ask him to check the base signals of the output transistor Q3.

5. look at the drive of the output transistor Q3

I try to find solutions compatible with the resources at his disposal.

[SoundTech-LG]

I have not had the need to use a scope to troubleshoot or repair these supplies. Visual inspection, and a good DVM is all I have needed. A good supply to compare with helps as well.

[sv3ora]

Thanks I will replace replace ISO1 (CNY21), C7 (470µF) and C46 (33µF). I bet you mean C17.
For the CNY21 I will have to wait since I do not have it.
I will let you know.

[akimpowerscr]

On the schematic of DeepLInk, I see C7 (and not C17) in serie with R4 (22R 5W)

[sv3ora]

Ok we are talking about the same capacitor thanks. Note the schematic I have reverse engineered at the previous post.

My S1 reads 1K by the way in circuit.

Can I temporary remove RT1 thermistor from the mains of the PSU? I am running out of them, so I want to place them there when the PSU finally works.

[SoundTech-LG]

Shotgun replacement of CNY21 (if you find one) is possible (but unlikely) to solve the issue. What did you do to test the part, and determine it was bad?

[sv3ora]

I did not test the opto isolator, I just read what other members here suggested about possible fault.

I replaced C46 and C17 with the same type and value of capacitors (one of them is tandalum). I also temporarily removed the mains inrush thermistor, which was blown every time the fuse was blown, because I am running out of them.
I powered up the scope and it worked for half a minute without blowing up anything. Good that the inrush current is not much to blow out the fuse without the thermistor in place.
Then I checked immediatelly for very heated parts in the PSU and I could not find any. The Q3 and the R16 are heated a bit though, but nothing worrying.

The fan on the scope passes air through the PSU enclosure to cool the heatsinks, so I do not want to operate the psu a lot of time without enclosing it and place it as it should inside the scope to allow for air flow. But I can point a fan onto it as it is now (without any enclosure, just the bare PCB and when/if the fuse blows again, I could check with my finger for overheated parts.

Will this help?

[SoundTech-LG]

No need to be concerned with heat. It can run a long time cool, without forced air before it gets very hot at all (not an issue) at least when it is operating properly. The CNY21 is easy to test. Remove it (carefully) and use your diode tester on your DVM. The LED will conduct one direction with a voltage drop 1-2 volts or so. The photo transistor will test open unless you light up the LED. You should also find an open ckt. between the two components (no leakage). If these conditions are found, it is unlikely it is bad.

[sv3ora]

Ok after the changes I did, mentioned in the previous post, and with an external fan blowing air into the PSU, the fuse was still blown after 4:34 mins, of course without blown thermistor, as there is no such now.
The Q3 is cool ok.

I am going to check the optoisolator the way you proposed, although it seems to me unlikely that this causes the problem after such a delay.

[SoundTech-LG]

Here are some photos I snapped of the creeping Electrolyte under, and into the main transformer. This was a real dog to find, and I almost gave up. It was not so readily apparent with it mounted to the board, but certainly tested leaky.
Oh, and it had the same symptom as yours, but instead, blew the TiPL755A after about 4-5 minutes.

[sv3ora]

Ok you convinced me. I am going to remove the transformer and check.
Wait for my update.

[SoundTech-LG]

Can't hurt to have a look. BIG soldering iron. Metal strips coming off transformer to ground plane. Also plastic insulators on the pins standing off the transformer to board, easy to lose. Before you remove it, you should test leakage across primary to secondary. Should be practically unmeasurable resistance. You may have to suck the solder out of the connections to really be sure. Good luck.

[sv3ora]

It's true that the board is very clean and I cannot see any signs of fluid or corrosion by looking veeery carefully in different light angles and in backlight under the PCB. I look carefully underneath the transformer and I cannot notice anything. and the board by the sides of the transformer is very clean and shiny.

How did you fixed yours?

[SoundTech-LG]

I replaced both transformers. Both had leakage across primary to secondary. You need to confirm little, or no leakage. If tests good, then no need to remove them.
Got the transformers from a donor board.

[sv3ora]

I am not sure, but I checked with the multimeter in the auto range all the pairs of the primary/secondary pins and I find no leakage (high resistance value) in both transformers.

The only place I can't check for liquid, is at the base of some secondary capacitors, cause these have silicone glue on their base.
I quess I have to remove it or them and check.

I will also check this optoisolator the way described tomorrow and let you know.

[SoundTech-LG]

You are ok if you measure no leakage. You certainly want to replace ALL the 1000 uF 35V caps at the very least. They are the ones which really leak out. Sometimes it is hard to see the electrolyte on the board. It depends how long it has been there. I have seen it clear and really hard to notice (until it starts corroding everything).

[akimpowerscr]

No need to be concerned with heat. It can run a long time cool, without forced air before it gets very hot at all (not an issue) at least when it is operating properly. The CNY21 is easy to test. Remove it (carefully) and use your diode tester on your DVM. The LED will conduct one direction with a voltage drop 1-2 volts or so. The photo transistor will test open unless you light up the LED. You should also find an open ckt. between the two components (no leakage). If these conditions are found, it is unlikely it is bad.

It is quite impossible to detect a loss of gain of an optocoupler in this way.
However, it is precisely this defect that could cause the problem that sv3ora notices.

Anyone with a bit of experience repairing devices with optocouplers would start by replacing the optocoupler without hesitation.

[akimpowerscr]

I am not sure, but I checked with the multimeter in the auto range all the pairs of the primary/secondary pins and I find no leakage (high resistance value) in both transformers.

The only place I can't check for liquid, is at the base of some secondary capacitors, cause these have silicone glue on their base.
I quess I have to remove it or them and check.

I will also check this optoisolator the way described tomorrow and let you know.

The insulation faults of a transformer must be measured with a megger, under voltage of 250V or 500V, not with a multimeter.

The multimeter measurement only makes it possible to detect obvious faults, such as a clear short-circuit between primary and secondary and is therefore totally insufficient.

[sv3ora]

I suspected it so, but it's too dangerous to measure when live!
I will order a new optoisolator and replace it just to make sure. If it isn't at least we would have excluded this to be the fault.
I will keep you updated.

[akimpowerscr]

If it is an earth leakage due to a faulty transformer insulation, the earth current protection of your electrical installation should cut off the current to your electrical network.

Do you have such protection (10 or 30mA) against earth leakage currents?

Try powering your oscilloscope with an isolation transformer and check if the fuse blows again or not.

[sv3ora]

I am not sure I have, but if regulations here specify a must have, I should.
Problem is that I do not have an isolation transformer of that power. All these tests (isolation transformer, variac, scope) require instruments that I currently don't have. Perhaps it is a good investment to buy some in the future.
so simple things like the multimeter, capacitance/inductance meter, frequency meter is all I have now. I used this scope for measurements, but since it failed...

I am ordering another optoisolator.
Oh by the way, can I use another type of optoisolator temporarily as a test, like for example a 4N25, or I have to stick with this particular model?

[akimpowerscr]

It doesn't seem logical to burn out fuses and NTCs and not damage transistor Q3.

The overcurrent therefore seems not to pass through Q3.

There remains one point that has not been verified: the VR1 and VR2 spark gaps.

If an arc bursts on one of the spark gaps, the other will also strike and there will therefore be a dead short circuit of the dc power supply, which would immediately burn the fuse.

Replace CNY21 by the same model.

[sv3ora]

I have changed these spark gaps, they are of the same type, brand new. I have also changed the big electrolytics after the rectifier bridge. All brand new.

[akimpowerscr]

If we add your lack of measuring instruments and equipment with everything that has already been replaced, we gradually evolve into a problem without solution.
 

[sv3ora]

There are always solutions.
One will be replacing everything, although the cost would be high
Even if this can't be fixed, there may be a member who has a failed scope and wants to sell his PSU.
Anyway I will try this opto isolator to see if this is the problem.

[akimpowerscr]

When I talk about a problem with no solution, it refers to finding the fault in this power supply and fixing it.

Apart from this, of course, there are always solutions, for example, to buy another oscilloscope.

Replacing everything is not that easy, for example, the transformer certainly should not be easy to find.

There is essential information that you are not able to provide.
1) is the overcurrent which causes the fuse to burn a current between phase / phase or phase / neutral or a current between phase / earth?
2) how is your distribution network configured in Greece?
It is, I imagine, 220V or 230V, but is it 220V between phase and neutral or 220V between phase and phase?
3) Does your installation have protection against earth leakage currents? Find out more or go see your electrical panel.
4) Have you checked your mains voltage? Are there no transient overvoltages?
This information would be essential in trying to understand what is going on ....

Intermittent failures that occur in random times are very difficult to diagnose.

And C2 and C47, are they really X2 capacitors? Have they never been replaced?

Remove them to see if the fault continues or not without these capacitors.

[akimpowerscr]

I still don't understand how it is possible to burn out multiple fuses without damaging transistor Q3.

A fuse does not normally protect a transistor.

The only semiconductors that can be protected by a fuse are diodes, thyristors and triacs, and special, ultra-fast fuses are required with an I²dt at least 30% lower than that of the semiconductor to be protected.

[DeepLink]

Replacing components from one end - is not the proper way of repairng

Have you measrued the output voltage of any of the outputs?

I suspect that it could be an overvoltage on the output, which triggers the SCR (A2) and kills the +15,5V
As the supply runs for a few minutes before it blows the fuse, it could be that the output voltage creeps slowly up
which could be a worn out opto coupler (CNY21) with low CTR, or an overload on the +5,15V from the scope itself

Again the best way to repair this is to do it out of the scope (this way you dont destroy the rest of the scope when something goes wrong)

Plan your measurement
Attach your DMM
Turn on
Record/observe
Turn off
Evaluate
Plan your next measurement
....

[SoundTech-LG]

Good plan...  measure everything you can, otherwise you are flying blind.

The CNY21 will be difficult, expensive, or both. Take a look at Vishay CNY65, or 66 series. Measure for correct dimensions. Cheap to try.

[akimpowerscr]

The CNY21 optocoupler is neither hard to find nor too expensive.

https://www.ebay.fr/itm/CNY21-Optocoupleur-T-H-T-OPCNY21-/170798599099

If you want to fix your oscilloscope, don't behave like a homeless man ... don't try to save a few euros on spare parts.

You risk introducing new problems that could make the diagnosis even more difficult

Buy only the original part, not the so called equivalents.

[sv3ora]

Hi,
today I removed a few electrolytics on the secondary side and I noticed that the pcb traces near them have signs of leaked electrolyte (attached). There is also electrolyte in other places near some other electrolytics on the secondary and it stinks fish when desoldered, this is electrolyte. In some caps the electrolyte can be seen below the cap body at the black rubber.

So there is definitely some leakage under the silicone. I bet these have to be carefully examined and replaced, what do you think?
Do you think this could cause this 2 minutes on, then BOOM problem?

[mistamidget]

I had exactly this problem this weekend with my HP 54510B DSO. I was using it to look at an rs232 data stream and it threw an error message and started to smell very badly.
I turned if off immediately and having a sniff around I found the power supply was the cause. Measuring the supply voltages I saw the -12V was not present and the lm240t (A8) was very hot.
Using the reversed engineered circuit from Thierry Magis referred to previously I found that -12V and +3.5V had 10ohm resistance between them which turned out to be exactly the same corrosion under C27 that your red arrow points to.
So, I scrapped out the circuit board between the two tracks and removed all the white silicon and bingo the carbon I dug out was the 10 ohm resistor.
I reassembled using all the original capacitors and the power supply is now perfect and my scope is fully functional again.
I think the electrolytic's are not leaking but the silicon causes condensation to be trapped and cause the corrosion.
I removed all the silicon where it touches the pcb so it doesn't happen again.
Hope this helps.

[sv3ora]

Finally, I fixed mine by replacing all the electrolytics in the PSU. Now the PSU seems to be working fine.
Lesson to learn, even if this is not always obvious, you should first check all the electrolytics UNDERNEATH for leaked electrolyte before you try anything else in these old SMPSUs, unless there is a visible trace of damage in another component.

Thank you all for the help, I leave the thread open for other people to post if they have similar problems with this PSU.

[wje]

I wish I had found this thread a week ago. I had exactly the same problem, -12 reading only 1v or so, some mystical low-ohm reading between -12 and ground even with every cap that could have shorted removed. Then I found the bad cap on 3.5, same carbon buildup, bingo. I sure spent a lot of time tracing, removing, replacing. At least it was educational. BTW, my final diagnostic tool was to use a delicate procedure... back-feed the -12 with a bench supply capable of supplying 5A at 12v and waited to see what caught on fire, or at least got hot. Seconds later, that pesky cap identified.
Bill