SMPS repair attempt strategy advice sought

[jaunty]

recently bought a pretty pricey (well over 2000 USD) piece of gear from ebay ... the owner warned me there were probably some electrical issues with it despite it having had service previously but it's a very complicated piece of gear  - a '79 era studio digital delay unit made by AMS. At any rate - when i first got it in I decided to fire it up to see what it was doing  - it ran FINE for the first several minutes - then POOF!! one of the PS resistors flamed out just like a struck match.

Now unfortunately SMPS' are a bit confounding to me  - I've read lots about them and watched tons of repair videos on youtube but it still seems like a black art to me and I've only just got my head around linear PS's to the point where I'd have a good shot at designing one.

Since i had a resistor flame out - i'm a bit scared to try to fire it up again lest i do any more damage to the boards inside (not so easy to disconnect the PS from the rest of it since half the SMPS is on the 'motherboard' to which 10-12 large 'daughtercards' are attached to.

But preliminary investigation shows that the SMPS controller IC (SE5560) has the current limiting signal pin shorted to the GND (big surprise right? haha) - the tantalum (1nf IIRC) attached to that pin is also shorted and there's a bad diode attached to THAT. I'm assuming maybe the IC controller had a spike that shorted the tantalum and took out the diode and resistor - OR the diode started the problem... at any rate - I think i have a decent start on this and should replace the SE5560, diode, tantalum and the resistor  - but I am wondering if maybe there are other places I might look for damage (?) like in the rectifiers or whatever (or are those mosfets?)

ALSO the flamed out resistor is now open and since it burned up I have NO WAY of knowing what the value of the resistor is. There are NO schematics or service manuals available unfortunately!! and the the only person who does repair on these units refuses to release info on them but WILL look at defective units for a flat fee of USD 1400

can ANYONE help??? i would be very grateful.

note: I've already gone over the whole board and found no other obvious problems - at least in-circuit but I do plan to replace all the electrolytics pictured as well - i would just like to put together a bill of materials for this repair.

John T


overall board view

flamed out resistor
AC reservoir (?) side of things on lower board

[Andy Watson]

1983 - that's an oldie! I think you're going to have to reverse-engineer the schematic. Since the resistor needs to be replaced you've got nothing to lose by removing it from the board - it might reveal the colour of the third band. Also, the 2k2 resistor three places along has a suspicious black spot on it  - have you checked it?

Also, post the model number of the AMS unit - somebody might be sitting the appropriate knowledge.

[oldway]

AMS NEVE still exist, try to ask them for schematics of this the power supply.....
AMS Neve Ltd,
AMS Technology Park,
Billington Rd,
Burnley,
Lancashire,
England,
BB11 5UB

T | +44 (0)1282 457 011
F | +44 (0)1282 417 282
E | info@ams-neve.com
W | www.ams-neve.com

Without schematics, it will be very difficult to repair.

[jaunty]

AMS NEVE still exist, try to ask them for schematics of this the power supply.....
AMS Neve Ltd,
AMS Technology Park,
Billington Rd,
Burnley,
Lancashire,
England,
BB11 5UB

T | +44 (0)1282 457 011
F | +44 (0)1282 417 282
E | info@ams-neve.com
W | www.ams-neve.com

Without schematics, it will be very difficult to repair.

they are NOTORIOUS for being tight fisted with their schematics - everyone else says the same thing sadly. and they insist you send it to them or their american 'agent' for repair (absurdly expensive) - though I thought i might try to take a crack at it.

[jaunty]

ok - it's an AMS DMX 15-80S

well i figured that finding the failed components that I did was a pretty good step forward - i suppose i can just replace those and see what happens - perhaps use a lightbulb current limiter - but failing having schematics perhaps it's not such a terrible strategy to replace most of the silicon components as well as the capacitors? it can't be THAT complicated ... I was just wondering if there was a good troubleshooting guide somewhere online so i can understand the problems with these and how they work better ...

also - i'm trying to get someone to send me a pic of the same board in theirs - maybe i can get the resistor value from that - good idea about pulling it to see if there's a hint on the other side though - but i'm not sure i would expect the paint to survive ...

[Paul Moir]

A photo of the bottom side of the board would help a lot.  We need to be able to trace wires to work out the schematic so clear shots of the overall board are more helpful than closeups.  I'd start with a wager that the burnt up resistor is 4.7 ohms, same as the nearby one.  May even be there to protect the tantalum.
Fuse blow?  If not the BUZ10 MOSFET is likely OK. 

[jaunty]

A photo of the bottom side of the board would help a lot.  We need to be able to trace wires to work out the schematic so clear shots of the overall board are more helpful than closeups.  I'd start with a wager that the burnt up resistor is 4.7 ohms, same as the nearby one.  May even be there to protect the tantalum.
Fuse blow?  If not the BUZ10 MOSFET is likely OK.

Fuses seem fine - it looks like the resistor WAS the fuse in this case ....

ok sure i would be more than willing to do a shot of the bottom if someone thinks that might help ...

[Seekonk]

Have you actually checked to see if that resistor is bad?  I suspect a tantalum capacitor is bad somewhere.  Look to see if one side of that resistor is shorted to common. 

[lordvader88]

For the brunt 47??ohm I guess start really really high and work down if u don't really know.

[jaunty]

well the resistor measures open (in circuit anyway) ... that much i know

[Shock]

Isn't that resistor the same as the one next to test point 3? Yellow Violet Gold? There looks to be some gold in the picture. When you remove the resistor carefully look at the underside for more color.

The other way is crime scene investigation time, if the resistor is broken in two then carefully clean in the break to take measurements of both halves and add them together. Since the burned out section might account for some resistance add a percentage appropriate to the unrecoverable area of the resistor.

To get an idea of current flow with the resistor out of circuit you can measure resistance to the source and common or ground. You can also trace out and draw what is connected to that resistor to get an idea, reverse engineering a schematic makes it easier to understand the function of connected components/nodes.

Once you have ruled out everything else once it's running you can measure voltage drop across it to determine current this may show up an over voltage/current/ripple problem which will cause a premature failure.

Check for dry joints as well, along with heat comes dry joints. There looks like there is a dry joint on the gate of that FET and another on that voltage regulator?

[Armadillo]

The power supply board makes following voltages;
+5v @ 5A
-5v @ 500mA
+15v @ 1A
-15v @ 1A
+12v @ 500mA
You can get alternative supply boards to replace it.

Alternatively, as Paul already mentioned, a complete clear TOP and Bottom photo taken upright would be helpful to trace the circuit.

Edit: I saw another post under Bitsmith asking David for reply.....   

[anachrocomputer]

Lots of tantalum capacitors on that little PCB. They have a tendency to fail as a dead short circuit, and will frequently burn out a resistor as a consequence. First step, then, is to check all the tantalum caps.

[jaunty]

here is the underside of board pic - thanks for your patience. Looks like the area under the MOSFET has been getting NICE AND HOT!! hmmm ... the plot thickens.

anachrocomputer - yes - checked all those tants out first thing - I think they tend to get a bit of a bad rap - as long as voltages are kept within spec they do a great job - but i guess it's a lot to ask of a component on a flawed or failed design

Shock - thanks for the tips - and yes i guess i would probably already have reflowed all the solder joints on this board before a preliminary test anyhow

[jaunty]

The power supply board makes following voltages;
+5v @ 5A
-5v @ 500mA
+15v @ 1A
-15v @ 1A
+12v @ 500mA
You can get alternative supply boards to replace it.

Alternatively, as Paul already mentioned, a complete clear TOP and Bottom photo taken upright would be helpful to trace the circuit.

Edit: I saw another post under Bitsmith asking David for reply.....   

yes that was on the GS forum ... not sure if i'll get a reply ... we will see.
alternative boards? i don't think generic ones are going to work very well given the design of this thing (the other half of the supply - the AC end?) is on the motherboard below

[jaunty]

i wonder if i should scratch back that resist over the MOSFET and maybe build up the traces under it a bit with more solder to improve the current carrying capacity of the traces and help act as a bit of a thermal sink too (?) - doesn't seem like a terrible idea

[Andy Watson]

The power supply board makes following voltages;
+5v @ 5A
-5v @ 500mA
+15v @ 1A
-15v @ 1A
+12v @ 500mA
You can get alternative supply boards to replace it.

Alternatively, as Paul already mentioned, a complete clear TOP and Bottom photo taken upright would be helpful to trace the circuit.

Edit: I saw another post under Bitsmith asking David for reply.....   

yes that was on the GS forum ... not sure if i'll get a reply ... we will see.
alternative boards? i don't think generic ones are going to work very well given the design of this thing (the other half of the supply - the AC end?) is on the motherboard below

I think those values are all the supply voltages from the whole power supply. I believe the PCB in the photo is responsible for converting either the +12V or +15V down to +5V . The pads on the PCB seem to confirm this, i.e. there's a ground, an input and an output, plus one other (as yet) unidentified trace - could be feedback. In short, if you can't repair this PCB, it should be easily possible to find (as Armadillo says) an alternative +5V substitute supply.

[james_s]

Or build a replacement. If you just need to get 5V form 12 or 15V there are single chip solutions that need only an inductor, diode and a couple resistors. Or you could buy a ready to use module from China for not very much.

[jaunty]

well - i don't really see the difference between building a new one (more work) vs just replacing enough components on this one ... given what a pain it is to order and wait for components via mouser maybe i should just replace all the silicon while i'm at it (?) - might be more expedient than replacing a few parts and having that fail (?)

[james_s]

Well for me building a new one is 5 or 6 parts and takes 20 minutes to deadbug, while repairing the old one could take considerable effort. I do prefer to repair the original whenever possible though, I was just tossing another idea out there.

[jaunty]

OUT OF CURIOSITY - does anyone know what causes the wrinkling in the underside traces/ground planes? is that just the resin over top chemically aging? or is it the copper wrinkling from overheating etc?

should I assume the MOSFET needs to be replaced with something else  given how badly the board is getting baked below it?

[jaunty]

James - i'm not sure i want to mess with the design in terms of current draw and ripple - it seems like the components on this board were chosen for a reason - if anything I think this supply's probably been under-spec'ed ... and maybe needs to be improved upon or at least meet original spec - whatever that might be

[Shock]

OUT OF CURIOSITY - does anyone know what causes the wrinkling in the underside traces/ground planes? is that just the resin over top chemically aging? or is it the copper wrinkling from overheating etc?

Lets ask Dave

[jaunty]

oh interesting - thank you shock. i was kinda wondering - i have a few other pieces with a similar problem

[Paul Moir]

The yellow "tantalum" capacitor marked 103 and pin 11 of the NE5560 are shunted to ground by that low value 4.7 ohm resistor, so they may not be shorted.  The capacitor marked 103 is likely ceramic which are not nearly as susceptible to shorting out as the orange tantalums.  I see you have removed one of the orange tantalums in the path of the burnt resistor:  I is this one  shorted?

Sorry, I'm still trying to figure out what the burnt resistor does.  It almost appears to be shutting down the regulator via one of those two transistors next to the tantalum you removed by taking away the gate drive, but that's just a half-baked idea.  Since there's a 35v tantalum (the one  of the two near the edge of the board by the adjustment trimmer) in line with it, it shouldn't have more than 35v applied to it.  That should limit it to being 470, 47 or 4.7 ohms or else it shouldn't have burned out.  But that's assuming something very bad didn't come up to it through the BUZ10's gate.  There's no continuity between the gate (connected to that jumper right in front of the BUZ10) and either of its other pins?