A4000 PSU - KABOOM!!!!

[dingebre]

IC1 in my Skynet A4000 PSU literally blew up in my face, thankful for eye protection So, two questions, please.

1. I found several copies of the schematic, but the part number for IC1 is not given. I'd rather not have to figure out and "best guess" what the part is. Does anyone know? I can't find any photos of the Skynet PSU board that are clear enough to read the part number.

2. The A4000 was working fine. It had been restored by a reputable shop and the work was excellent on close inspection. The PSU was cleaned but NOT recapped. The motherboard was cleaned and recapped. There was essentially no battery damage, yay.
     a. I installed new 3.2.2 ROMS and an A3660 CPU board. Nothing else changed.
     b. I turned it on, and there was no power LED nor HD LED.
     c. I disassembled some, tried again, same. I neither saw nor smelled smoke, nothing was remotely warm, much less hot.
     d. At this time, I noticed that neither the CPU cooling fan nor the PSU cooling fan ran when powered.
     e. I disconnected all the power connectors and tried the PSU on its own. Nothing.
     f. I tested resistance from all the power lines to ground on the motherboard and on the PSU connector and found no short circuits.
     g. I opened the supply and saw the fuse was blown.
     h. I replaced it with the specified 5A 250V fuse, turned it on and the fuse immediately blew and I saw a flash out of the corner of my eye near the two big filter caps and transformer. I saw no
         damage or discoloring anywhere in the area on the PCB and nothing looked visibly damaged.
     i. I removed the PCB, tested the caps in place with an ESR meter and ohm meter. They seemed "ok", but I noticed C6 looked like it was bulging a bit.
     j. I reflowed a couple of sketchy joints and snipped a couple of very long lead ends, but there was nothing visually amiss on the solder side of the PCB.
     k. I installed a new fuse and disconnected the fan (the shop who did the restoration installed a new quieter fan) just in case it was a problem.
     l. The fuse immediately blew and IC1 decided to exit this life in a blaze of glory.

Any thoughts as to where I should begin trouble shooting? I thought I'd start by replacing the polarized capacitors, but this problem just seems weird. I strongly suspect it wasn't anything I did on the motherboard, i.e. the A3660 and ROMS, but I don't want to fall victim to confirmation bias either. I just don't want to think or believe I've torched my motherboard

I have an ATX supply I bought for a future Amiga project and have an ATX to A4000 adapter cable on order, but until I get the adapter, I'd love to see if I can get this PSU working again.

Much thanks for any help or ideas!

David

[Castellen]

The main switching transistor (Q1) going short circuit would be a likely cause of the AC fuse blowing and IC1 being damaged.  The fault current may have also damaged the bridge rectifier as well.

These power supplies aren't that complicated, so should be repairable with a bit of time and effort.  I don't have one of those supplies apart on the bench at the moment, so not sure what the part number of IC1 is.  It's the switchmode controller and being a DIP8 package, it's certainly not anything common such as a TL494.  Someone might be able to supply a photo to identify the part.

If you can source a replacement IC1, I'd be inclined to replace both Q1 and the bridge rectifier as a start.  Also do a quick check across each diode and transistor to look for any obvious short circuits anywhere else.  I doubt if this would be a capacitor problem, but it would be worth replacing all of the electrolytics once you get the supply running.

Another tip for working on switchmode supplies is to use a series test lamp (100W - 200W incandescent/halogen) on the AC input.  That way if you have a catastrophic short circuit fault somewhere, worst case you'll just make the lamp light up instead of causing more damage.

 

[dingebre]

The main switching transistor (Q1) going short circuit would be a likely cause of the AC fuse blowing and IC1 being damaged.  The fault current may have also damaged the bridge rectifier as well.

These power supplies aren't that complicated, so should be repairable with a bit of time and effort.  I don't have one of those supplies apart on the bench at the moment, so not sure what the part number of IC1 is.  It's the switchmode controller and being a DIP8 package, it's certainly not anything common such as a TL494.  Someone might be able to supply a photo to identify the part.

If you can source a replacement IC1, I'd be inclined to replace both Q1 and the bridge rectifier as a start.  Also do a quick check across each diode and transistor to look for any obvious short circuits anywhere else.  I doubt if this would be a capacitor problem, but it would be worth replacing all of the electrolytics once you get the supply running.

Another tip for working on switchmode supplies is to use a series test lamp (100W - 200W incandescent/halogen) on the AC input.  That way if you have a catastrophic short circuit fault somewhere, worst case you'll just make the lamp light up instead of causing more damage.

Many, many thanks!!

Took a closer look around Q1 and sure enough, black marks on the PCB. They were a little obscured by R9. R9 also exploded I took out Q1 and just for fun, tested it, shorted. R9 was open. Best my tired eyes can tell, R9 is Red, Purple, Silver, Gold which decodes to a 0.27 ohm resistor. Seems like a weird value (not the order of magnitude so much but the 0.27), but ok. I've attached a photo that shows the bands if anyone wants to look and tell me what they see.

IC1 is a Unitrode SK-8085. Seems no one can find a data sheet or application note. But at the vendor where I found some, there was a description and it was described as a variable voltage regulator. not a PWN Modulator, does that make sense? Q1 is a Toshiba K2038 (SK2038) power MOSFET. I found a bunch of these, too.

The question I have now is, what fried IC1? Something on the MB or did something else in the PSU cause Q1 to explode?

Regardless, I'm on my way and again, thank you Castellen!

I'm attaching an annotated schematic of the PSU with a couple of part numbers as well as a couple more photos for fun.

(Edit: The bridge rectifier and other diodes in the area test ok...)

David

[Castellen]

Took a closer look around Q1 and sure enough, black marks on the PCB. They were a little obscured by R9. R9 also exploded I took out Q1 and just for fun, tested it, shorted. R9 was open. Best my tired eyes can tell, R9 is Red, Purple, Silver, Gold which decodes to a 0.27 ohm resistor. Seems like a weird value (not the order of magnitude so much but the 0.27), but ok. I've attached a photo that shows the bands if anyone wants to look and tell me what they see.

IC1 is a Unitrode SK-8085. Seems no one can find a data sheet or application note. But at the vendor where I found some, there was a description and it was described as a variable voltage regulator. not a PWN Modulator, does that make sense? Q1 is a Toshiba K2038 (SK2038) power MOSFET. I found a bunch of these, too.

The question I have now is, what fried IC1? Something on the MB or did something else in the PSU cause Q1 to explode?

A quick search suggests that the TI part UCC38085 is probably the same thing.  Looks as though it's obsolete and no longer available from the usual vendors (Digi-Key, Mouser), but you might be able to source something from Ebay, Ali, etc.

Yes, the value of R9 looks correct at 0.27 Ohms.  You can see in the circuit that it's used in a current detection arrangement.  I'd expect the controller IC (pin 3) measures the voltage across R9 to determine the current flow.  I would fully expect R9 to be open circuit, it would have been carrying a lot of fault current with Q1 shorted, so it's acted like a fuse.

To answer your question, Q1 has probably died first, which isn't uncommon with high voltage transistors.  So instead of silicon junctions in the transistor, you've now got a shorted mess.  That means you'll have around (110VAC/SIN(45)) = 156VDC on the drain terminal of Q1, and this is now connected directy to the gate terminal.  I don't know what the maximum voltage on the drive output pin of IC1 is, but probably something like 10V.  And you've now got 156V on it (via R6).  You've already seen what happens next....

Which reminds me, you should sanity check R6 as well to make sure it's not open circuit.  If it's a low-ish value (under 100 Ohms) then it might have been damaged.

The A4000 main board is probably fine.  Looks as though there's sufficient protection in the supply design to prevent anything too nasty from appearing on the DC output with this kind of failure.

 

[dingebre]

A quick search suggests that the TI part UCC38085 is probably the same thing.  Looks as though it's obsolete and no longer available from the usual vendors (Digi-Key, Mouser), but you might be able to source something from Ebay, Ali, etc.

Yes, the value of R9 looks correct at 0.27 Ohms.  You can see in the circuit that it's used in a current detection arrangement.  I'd expect the controller IC (pin 3) measures the voltage across R9 to determine the current flow.  I would fully expect R9 to be open circuit, it would have been carrying a lot of fault current with Q1 shorted, so it's acted like a fuse.

To answer your question, Q1 has probably died first, which isn't uncommon with high voltage transistors.  So instead of silicon junctions in the transistor, you've now got a shorted mess.  That means you'll have around (110VAC/SIN(45)) = 156VDC on the drain terminal of Q1, and this is now connected directy to the gate terminal.  I don't know what the maximum voltage on the drive output pin of IC1 is, but probably something like 10V.  And you've now got 156V on it (via R6).  You've already seen what happens next....

Which reminds me, you should sanity check R6 as well to make sure it's not open circuit.  If it's a low-ish value (under 100 Ohms) then it might have been damaged.

The A4000 main board is probably fine.  Looks as though there's sufficient protection in the supply design to prevent anything too nasty from appearing on the DC output with this kind of failure.

 

Thanks again very much.

R6 is just a jumper and it's fine.

Yes, I did witness close up what happens next 156 Volts on the drain really made for an awesome fireworks display!!

I appreciate the guidance. I took a lot of EE courses in undergraduate and graduate school and my dad had me soldering as soon as I could handle an iron. This only makes me dangerous when I touch electronics Power supplies are simply way outside my wheelhouse. Your explanations have really helped me understand much more about what is going on in this PSU.

Thank you!
David

[dingebre]

A quick search suggests that the TI part UCC38085 is probably the same thing.  Looks as though it's obsolete and no longer available from the usual vendors (Digi-Key, Mouser), but you might be able to source something from Ebay, Ali, etc.

Yes, the value of R9 looks correct at 0.27 Ohms.  You can see in the circuit that it's used in a current detection arrangement.  I'd expect the controller IC (pin 3) measures the voltage across R9 to determine the current flow.  I would fully expect R9 to be open circuit, it would have been carrying a lot of fault current with Q1 shorted, so it's acted like a fuse.

To answer your question, Q1 has probably died first, which isn't uncommon with high voltage transistors.  So instead of silicon junctions in the transistor, you've now got a shorted mess.  That means you'll have around (110VAC/SIN(45)) = 156VDC on the drain terminal of Q1, and this is now connected directy to the gate terminal.  I don't know what the maximum voltage on the drive output pin of IC1 is, but probably something like 10V.  And you've now got 156V on it (via R6).  You've already seen what happens next....

Which reminds me, you should sanity check R6 as well to make sure it's not open circuit.  If it's a low-ish value (under 100 Ohms) then it might have been damaged.

The A4000 main board is probably fine.  Looks as though there's sufficient protection in the supply design to prevent anything too nasty from appearing on the DC output with this kind of failure.

 

I took a look at the TI site and grabbed the data sheet. You are right, the ucc38085 is the TI equivalent to the Unitrode SK-8085. I read elsewhere on the net that TI took over the part from Unitrode and it is all making much more sense now. While obsolete, the TI ucc38085 is still in stock at some of the usual suspects, like Mouser.

David

[Castellen]

You should be good with that part from Mouser.  Not sure how available the SK2038 high voltage FET is, a quick search suggests it's not.  Won't be hard to find an equivalent part by comparing data sheets and using parameter based filters with Digi-Key/Mouser.  You're a bit spoiled for choice with what's available these days.

Datasheet for the original part is here: https://www.alldatasheet.com/datasheet-pdf/pdf/30600/TOSHIBA/2SK2038.html

Another suggestion is to replace capacitor C6 when you replace IC1 and Q1, before you do the initial run-up, as it might have been damaged at the same time as IC1 failed.  If you're unlucky, then the optocoupler (IC3) might have been damaged as well.  You could try it as it, or replace it if you wanted to be sure.

I'm not convinced that schematic is correct either.  It shows pin 7 (un-used push-pull output) connected to what's obviously the power supply for the PWM controller.  I suspect it should say pin 8 instead of 7 and you'll find that pin 7 isn't connected to anything on the PCB.

When you do the run-up, don't connect it to the computer, but instead connect a 4.7 Ohm wire wound resistor of at least 5W between +5V and GND (yellow and blue wires) to give the supply a 1A load, else you might get some unexpected results.  You might need the fan connected to the +12V supply also, some power supplies won't run correctly without at least a small load on the +12V supply as well.

If it all looks good, then you might as well replace the rest of the electrolytics while you have it apart, then call it done.

[Boing-ball]

Something to note here, which I wasn’t aware of until watch “Miketech” YouTube videos. The Glue used to hold certain components together after a time can harden up and become conductive. If you see any of that yellow glue on the circuit board. I would check with a multimeter or you may get another Kaboom! After you finished repairing.

[dingebre]

You should be good with that part from Mouser.  Not sure how available the SK2038 high voltage FET is, a quick search suggests it's not.  Won't be hard to find an equivalent part by comparing data sheets and using parameter based filters with Digi-Key/Mouser.  You're a bit spoiled for choice with what's available these days.

Datasheet for the original part is here: https://www.alldatasheet.com/datasheet-pdf/pdf/30600/TOSHIBA/2SK2038.html

Another suggestion is to replace capacitor C6 when you replace IC1 and Q1, before you do the initial run-up, as it might have been damaged at the same time as IC1 failed.  If you're unlucky, then the optocoupler (IC3) might have been damaged as well.  You could try it as it, or replace it if you wanted to be sure.

I'm not convinced that schematic is correct either.  It shows pin 7 (un-used push-pull output) connected to what's obviously the power supply for the PWM controller.  I suspect it should say pin 8 instead of 7 and you'll find that pin 7 isn't connected to anything on the PCB.

When you do the run-up, don't connect it to the computer, but instead connect a 4.7 Ohm wire wound resistor of at least 5W between +5V and GND (yellow and blue wires) to give the supply a 1A load, else you might get some unexpected results.  You might need the fan connected to the +12V supply also, some power supplies won't run correctly without at least a small load on the +12V supply as well.

If it all looks good, then you might as well replace the rest of the electrolytics while you have it apart, then call it done.

I found some supposedly NOS SK2038s on ebay. Yes, C6 should be replaced. It's definitely bulging and an easy one to swap out. I also have the NOS sk8085s on order.

Interestingly, pin 7 does go to C6, D3, R5, and the opto-isolator pin 5. Pin 8 is not connected. Maybe the TI part and the Unitrode part are not the same?

Hmmm, the smallest 5 watt resistor I have on hand are a few 500 ohm, left over from an O-scope rebuild. I'll see what I can find.

I found the PC111 and a directly NEC replacement, cheap enough that I'll just swap it out, too.


Thank you again, really, for your time and advice.

David

[dingebre]

Something to note here, which I wasn’t aware of until watch “Miketech” YouTube videos. The Glue used to hold certain components together after a time can harden up and become conductive. If you see any of that yellow glue on the circuit board. I would check with a multimeter or you may get another Kaboom! After you finished repairing.

Interesting. I'll have a look at the channel, too. No "yellow glue" that I can see. I will do more inspection before I power up again. I should have all the parts in a week or two. I'll post what happens.

David

[dingebre]

Here's my update. The PSU lives and the A4000 is up and thumping along happily. I don’t have the A4000 back together fully yet. I didn’t have time to finish today, but so far it all works great! The A3660 runs great, the ZZ9000 is wonderful, the whole system is like brand new.

I’m attaching an annotated schematic for the Skynet PSU with a few of the unmarked components marked.

In the end, here is what I did:
1.   Replaced the K2038 MOSFET which I think was the root cause. It was a dead short. Thanks for the tip Castellen. – Found some on ebay. It works so I’m not complaining.
2.   Replaced the 0.27 ohm resistor which exploded.
3.   Replaced the Unitrode SK-8085 which exploded. – I found some at UTSOURCE in China. I’ve had good experiences with them for at least a decade. They have a lot of very odd and rare semiconductors. I did get some counterfeit bucket brigade delay chips from them once (I had previously purchased some that were fine) and they were great to work with and I got the real ones I wanted.
4.   Checked the bridge rectifier, a KBL06 (x-ref to a GBU2506-G).
5.   Checked some other diodes and components.
6.   For good measure, I replaced the Panasonic P111 optocoupler (x-ref to a NEC PS2652).
7.   Recapped with low ESR electrolytics.

Tested it with a load on the +5 and +12 rails, then plugged into the motherboard. At first, it didn’t boot, but thankfully I remembered the daughter board on the ZZ9000 can come loose and cause problems after multiple insertions and extractions of the ZZ9000 into the A4000 daughter board. I reseated the ZZ9000's daughter board and everything came to life.

Thank you Castellen. Your guidance was invaluable. Thank you, too Boing-Ball for the tip about the "yellow glue."

Another Amiga is alive and breathing again!!

David

[Castellen]

Nice to hear you have the power supply working again.  At least it's a simple design that's easy to work on, unlike the A3000T power supply.

Someone was asking me the other day about how to fine tune the +5V output on the A4000D Skynet power supply, the same one as in this thread.  Not sure what the specification is, I'd expect around 3% (4.85V - 5.15V).  My notes on adjusting the output voltage:

Output voltage is set by ratio of fixed value resistors R12//R14 and R13//R15
R12 = 5.6k
R13 = 3.9k
R14 is not normally fitted
R15 = 220k
Increasing the R12//R14 total resistance increases the 5V output voltage
Increasing the R13//R15 total resistance decreases the 5V output voltage

Or to explain another way, leave the lower value resistors (R12, R13) as they are and add higher values (typically 100k or more) at R14 or R15 to fine tune the output voltage.


And just to state the obvious, it's not (practically) possible to adjust the +12V or -12V output independently of the +5V output.  What you see is what you get.

[dingebre]

Nice to hear you have the power supply working again.  At least it's a simple design that's easy to work on, unlike the A3000T power supply.

Someone was asking me the other day about how to fine tune the +5V output on the A4000D Skynet power supply, the same one as in this thread.  Not sure what the specification is, I'd expect around 3% (4.85V - 5.15V).  My notes on adjusting the output voltage:

Output voltage is set by ratio of fixed value resistors R12//R14 and R13//R15
R12 = 5.6k
R13 = 3.9k
R14 is not normally fitted
R15 = 220k
Increasing the R12//R14 total resistance increases the 5V output voltage
Increasing the R13//R15 total resistance decreases the 5V output voltage

Or to explain another way, leave the lower value resistors (R12, R13) as they are and add higher values (typically 100k or more) at R14 or R15 to fine tune the output voltage.


And just to state the obvious, it's not (practically) possible to adjust the +12V or -12V output independently of the +5V output.  What you see is what you get.

Thanks. I'm within about 0.05 to 0.08 volts on all rails.

David