1984 power supply repair

[connorwk]

I'm trying to repair this power supply from an old 1984 computer it seems to not give enough power on its output.
Its a 36v-5v DC-DC power supply branded ASTEC, there was a metal plate screwed to the back of the board to protect the solder joints from the metal case of the computer and on it it says ASTEC AA7281D (think the D is a rev letter as on another one of these power supplies it says AA7281 rev <letter>) 65-00029.
When it is in the computer its 5v output reads only 0.5v. When out of the machine hooked up to 36v and nothing on the 5v output it reads about 5.8v.
Nothing on the board seems obviously bad. The electrolytes seem good and everything else seems ok. Nothing obviously wrong.
Any ideas as to what may be wrong? 

[dr.diesel]

Somebody stole R28!  Normal to see unpopulated parts, a bit unusual to see snipped off ones!

Anyway, you can verify the 5v rail isn't being shorted by the computer?  Or perhaps can you load it while out of the PC and see what happens?

[JoeO]

Capacitors - If you have an ESR meter, use it on the largest capacitors.  If not, just replace the largest capacitors.

It is 29 years old.  The capacitors are probably "dodgy".

[Tita]

Switch mode power supply repair video:

View the entered video if you want to learn something! O0

[connorwk]

dr.diesel:
I can verify the machine isn't shorting it. I have two of these computers (1984 Burroughs B25s) and I put the supply from one of them into the one with the bad supply and it worked fine. So obviously it is the supply itself.
JoeO:
I do not have an ESR meter. Guess I should get one if I plan on doing more electronics in the future. Really all I have to test at the moment is a multimeter. I looked up and some people have said you can test the caps by using the resistance mode of a multimeter? Anyway I guess I could replace them although to the fact that they are dodgy, yes they are nearly 30 years old yet I have 3 more of these power supply boards with original caps that say otherwise to being dodgy. (3 more because I have two of the computers and each computer requires two of the power boards)
Tita:
Thanks for the video.
I guess I know what I shall be watching for the next hour!

[dr.diesel]

I looked up and some people have said you can test the caps by using the resistance mode of a multimeter?

You can get an idea, but nothing more unless you want to calc it, by knowing the meter source current etc.  If you remove the caps from the board and test their resistance it should read a dead short and rise to infinity/open.  How quickly it goes from short to open depends on the meter and size of the cap.

Make sure they are fully discharged before removal/testing.

[connorwk]

I removed the two right most big electrolytes the left blue one went up in resistance from short so I assume its good the other bigger black cap went from short up but then I retested it and it was open. So I guess that cap is bad. I will replace it and try it I guess.

[dr.diesel]

If you didn't short it back out the retest would still show open. 

Make sure it's completely discharged before each test.

[connorwk]

Silly me your right it is fine then. I guess I will check the smaller caps then and see if they are good.
If its none of the electrolytes then I am unsure of where to go next.

[fluxcapacitor]

The solder doesnt look good in or around the capacitor thats just below IC1 (LM393P comparator) .Its worth replacing all the caps if you can and take out LM393P and clean its legs up .

[connorwk]

I took out all the caps they all seemed fine.
I replaced the one cap under the lm393p chip since under it was a bit corroded.
Unsoldered, cleaned the pins, and resoldered the lm393p and still it gives the same results. I would replace the caps but I do not have a good supply of caps  I will need to buy a bunch later on but for now I don't have them to replace unless I used other values, but as I said they all seemed to test fine.
I'm assuming its going to be a transmitter next then? maybe a diode? I guess I will have to unsolder them all and test them.
I do have two of these boards so what I did before all this was go around just testing resistance between a bunch of different points around the board and compared it to the same points on another working board and everything seemed fine. I know that's not the proper way to test but I figured of something was way off I could know that somethings connected to those points where wrong.
I'm not the best at repairing and testing electronics thus posting it here in the beginners, so thanks for all the much needed help.

[Paul Price]

From the odd angle of the yellow transformer, it looks its been physically damaged.

[connorwk]

I don't think so the holes on the board are at a bit of an angle and on the other power boards it is angled as well but not as much.
Looking at it all that's actually slanted is the ferroid core, like I said the plastic holder of the wire is slightly slanted because the holes in the board for it aren't square.
But I don't think it is damaged the other ones look the same and they work fine.
Probably just an 80's design thing.

[JoeO]

Try moving the caps from the good supply to the bad one, if you feel confident enough in your soldering abilities.

[connorwk]

Hmm I guess I could do that.
I guess capacitors would cause me not to get enough power out of the board?
I'm sorry for my ignorance of electronics at the moment but I just don't understand how bad caps would cause just a power drop and not cause it to fail all together.
Like I said the only problem is its not supplying enough power as I get 5.8v when not hooked up to the computer but only 0.5v when it is in the computer trying to power it.

[fluxcapacitor]

Its a good idea to try the caps from the good board , try the big caps first,then the caps at C20 and C21 . 

[connorwk]

I did every electrolytic capacitor on the board and those two ceramics you said. still doesn't work.
So no its not the 30 year old electrolytes causing the problems or those two ceramics.

[Paul Price]

If you are getting almost no voltage out of this troubled power supply, the reason is not likely to be a capacitor, something more like a bad connection, shorted diode or transistor or bad IC or even a turn on/turn off part of the power supply circuit.

There is a myth that all power supply trouble is caused by capacitors, but a good deal of is.

 If you follow some of this advice given here you will waste hours and money replacing good capacitors or else swapping capacitors(totally ridiculous! advice). 

If a capacitor dome top is flat and there is no sign of leakage of electrolyte, the cap is probably ok. If you test all the capacitors, in circuit with an ohmmeter, and see any indication the capacitor is charging, the caps are probably OK. Capacitors can be fine even after 36 years, especially if this device has not used much during  this long period. Let an oscilloscope tell you if a cap is good or bad. 

if an ohmmeter measurement or an oscilloscope reading or even an AC DVM reading gives me reason to suspect a capacitor, I usually "bridge" a new capacitor across a  a suspected capacitor (or even bridge several suspected caps at once) to troubleshoot.

 If I can easily doe this, I sometimes disconnect just one lead of a capacitor, leaving the other soldered, to apply a ohmmeter test, in the case I suspect a leakage current through the capacitor or just to see if it will charge, in the instances where coupling capacitors could be causing problems. For bypass filter capacitors, bridging and ohmmeter measurements in circuit is all that is needed to check the capacitor.

I've been repairing electronic devices for a zillions years and have never needed or used an ESR meter, really not needed. If you suspect a capacitor is open  or has an high ESR, simply bridge it with a capacitor of a similar or a somewhat larger value to see if you can notice a change in operation. Use an oscilloscope or DVM to measure noise voltage across a suspected capacitor, even a AC DVM can be well used to detect excess AC across an electrolytic capacitor.


Try to ohmmeter test all the diodes, zeners, transistors, regulator chips and/or  MOSFET. etc. for shorted or open junctions.

The best tool for this troubleshooting job after resistance checks is your oscilloscope. If a capacitor has opened or has a high emf, you can easily see the high ripple (noise)voltage across the capacitor.

 Try comparing the good circuit board with the bad as far as voltages or wave shapes and this should show the fault.
 Try comparing the good circuit board device measurements of the diodes, transistors, etc with the bad board.

If you can find a point of difference, I can help you to proceed further.

What is the input voltage to this power supply? Why use 36V? Is this the voltage into the power supply from the power transformer post rectifiers?

[fluxcapacitor]

connorwk said he didnt have any of the larger capacitors,and there wasn`t that many to swap and try.I`m not doubting your experience,but caps can be bad even with a flat top and not spewing its guts out.Only trying to help

[connorwk]

The computer has an external 115v AC to 36v DC power supply then that is fed into the computer and this board turns it into 5v thus it being a 36v to 5v DC-DC converter.
I'll start testing all I can with my multimeter now. Sadly I do not have an oscilloscope other then a 1960s one that I think one of the vacuum tubes in it are bad, but thats besides the point.
fluxcapacitor:
I did swap all the electrolytes from the working board and those two ceramics as you told me to and I even put the ones from the broken unit onto the working one and it still didn't work and the one that did work worked with the caps from the broken one so it is NOT the caps. I know caps commonly fail now adays but most of the older stuff I have worked on say 20 years or older the caps never seem to be the problem.

[Shock]

Wouldn't he need to apply a load to the circuit to test the output voltage outside of the computer?

[JoeO]

connorwk said he didnt have any of the larger capacitors,and there wasn`t that many to swap and try.I`m not doubting your experience,but caps can be bad even with a flat top and not spewing its guts out.Only trying to help

We have nothing to feel sorry for.

You have to use what you have on hand. 

Since he had known good working caps, it is logical to try to swap them out to determine if they are the problem.  It should not take him "hours" to do something like that. 
The problem now has been narrowed down to "a bad connection, shorted diode or transistor or bad IC or even a turn on/turn off part" or " the diodes, zeners, transistors, regulator chips and/or  MOSFET".   

[fluxcapacitor]

Is there 1 external 115v AC to 36v DC power supply or 2 If theres 2 then.................

[connorwk]

No there is one and inside the machine there are two 36v to 5v converters and in the HD and FD box there are two more power boards 36v to 12v and another 36v to 5v. This computer has quite a few power boards all powered by the one huge external 115v AC to 36v DC power brick (the brick is roughly the size of two old xbox arcade supplies, a bit bigger then that in-fact).
Also I think I've only made matters worse I ended up making a mistake and blowing up a few more things on the board.
So I'm going to order all the transistors and a new lm393p from digikey and replace all those and all the diodes as-well. I'm going to assume it isn't any of the resistors or ceramic caps but unless someone thinks I should replace them too then I may order some of them as-well (all I'd really need are the ceramic caps as I already have a ton of resistors).
Hopefully that will get everything working if not I shall be at a loss for words.

Oh also this is what the machine looks like working. Well this is the one I have working.

As I have said I have two of the main CPU units and it has a bad power board thus me fixing it.
The computer has a modular design making it so you can add HDs, FDs, modem units, ect. onto the side of the machine if you wanted.
Anyway I'll be making an order up in digikey and getting what I need to basically rebuild the board.

[Paul Price]

It is just too much work and money and time to try to fix this thing by replacing every component on the board..there are easier ways to get your computer working. It's just bad advice and a poor strategy to repairing anything that has this many components this way.  Suppose you accidentally  create a solder bridge or insert a new component into the wrong place or the right component backwards, then you will end of spending a dozen hours for a short fireworks show at worst or a dead do-nothing supply after all that work.

If the total amp output of the power supply is <=5 amps, you could easily just pull out the whole defective board and replace it with a DIY LM2678 switcher power supply that has only 10 components. It could be constructed in an hour or less and take less time than replacing every part on the whole board. If you need more than 5amps, the LM2678 circuit could be duplicated and supply another 5 amps. And yes, I have built these on perfboard and they work just fine.