I am working on this power supply PCB. It's double layer, 2.5 mm thick, and pretty much cooked to a crisp after decades in an inadequately ventilated case.
First cap that I desoldered, I ripped a trace off the top side of the PCB. I figured the solder wick method wasn't going to cut it this time, so, not wanting to cause more damage, I ordered a Hakko FM-301 tool which I had been wanting anyway (it's a fantastic tool, by the way).
Second cap I desoldered (with the Hakko this time)... I ripped a trace off the board again. But I noticed that the trace was pulled up by a repair eyelet that was stuck to the cap pin. Then I saw that the exact same thing happened with the first cap.
Next I desoldered the DB9 power connector, and 9 out of 9 pins came out with an eyelet attached. I realized the eyelets were on the pins pretty tight, and that they were being pulled up not necessarily by leftover solder but by mechanical friction.
Now, I'm pretty sure those Roedensteins and DB9 connector I pulled out are original, yet every single hole had an eyelet in it, so it looks like the factory built the PCBs that way instead of plating the through holes? Isn't that unusual?
Anyway it is what it is, and I have to deal with it. These eyelets and pads are incredibly fragile, so the less work the better. But, seeing how baked this thing is I think it's reasonable to do a full recap, test all components, put fresh thermal grease on the transistors, etc.
Where I have access both top and bottom sides I can apply heat to both sides at the same time, so I think pulling out diodes, transistors and resistors without breaking anything else will be OK.
Not sure what to do about the caps though. They're mounted flush with the PCB, so there's no room to cut the pins off. Maybe I can just tear them apart to expose the top of the pins
The PCB traces are .35 mm thick and they are super strong, so it shouldn't be too hard to epoxy them back down. Then I can replace all the eyelets and pads. Which will leave me with a severely cooked board that looks like absolute crap, but at least it will be in better shape than now.
I also thought up two other (creative!) ways out of this mess:
I don't know what else is there? What would you guys do??
Thanks in advance, all thoughts are welcome! 🙂
First cap that I desoldered, I ripped a trace off the top side of the PCB. I figured the solder wick method wasn't going to cut it this time, so, not wanting to cause more damage, I ordered a Hakko FM-301 tool which I had been wanting anyway (it's a fantastic tool, by the way).
Second cap I desoldered (with the Hakko this time)... I ripped a trace off the board again. But I noticed that the trace was pulled up by a repair eyelet that was stuck to the cap pin. Then I saw that the exact same thing happened with the first cap.
Next I desoldered the DB9 power connector, and 9 out of 9 pins came out with an eyelet attached. I realized the eyelets were on the pins pretty tight, and that they were being pulled up not necessarily by leftover solder but by mechanical friction.
Now, I'm pretty sure those Roedensteins and DB9 connector I pulled out are original, yet every single hole had an eyelet in it, so it looks like the factory built the PCBs that way instead of plating the through holes? Isn't that unusual?
Anyway it is what it is, and I have to deal with it. These eyelets and pads are incredibly fragile, so the less work the better. But, seeing how baked this thing is I think it's reasonable to do a full recap, test all components, put fresh thermal grease on the transistors, etc.
Where I have access both top and bottom sides I can apply heat to both sides at the same time, so I think pulling out diodes, transistors and resistors without breaking anything else will be OK.
Not sure what to do about the caps though. They're mounted flush with the PCB, so there's no room to cut the pins off. Maybe I can just tear them apart to expose the top of the pins
The PCB traces are .35 mm thick and they are super strong, so it shouldn't be too hard to epoxy them back down. Then I can replace all the eyelets and pads. Which will leave me with a severely cooked board that looks like absolute crap, but at least it will be in better shape than now.
I also thought up two other (creative!) ways out of this mess:
- Have the PCB cloned by one of those shops in China
- Toss the PCB and rewire the thing point-to-point
I don't know what else is there? What would you guys do??
Thanks in advance, all thoughts are welcome! 🙂
Do anything but use that PCB. Point to point might help with spacing out parts, but something tells me that there is no space to use as it is. In other words, you may not have the real estate for point to point. Cloning may be just as well, as long as you can get something heavy duty enough.
If it was me, I would 1. Definitely get a new clone PCB made. Problem with those old fiber PCB´s is, that when getting that black, they start conducting, and keep gettig worse. A new PCB will free you from future trouble. And 2. get rid of those "burgundy" colored ROE lytics.
This type is an absolute rotten type of cap´s. And the can´t withstand heat, The later orange types were far better in that regard.
This type is an absolute rotten type of cap´s. And the can´t withstand heat, The later orange types were far better in that regard.
The board is toast, literally. Use it to make a new board, same thickness, with 2 oz copper like you have now.
Use all new parts. This will be a lot of work unless you can find a company that does this.
Use all new parts. This will be a lot of work unless you can find a company that does this.
@lxnay You are correct, point-to-point would be too tight a fit so if I went that route I'd get a new case. The original case is ugly as sin anyway. It doesn't even look like the preamp.
@Boydk Of the 2 Roedensteins I pulled so far, one has really high ESR and the other won't even measure 🙂
@Boydk Of the 2 Roedensteins I pulled so far, one has really high ESR and the other won't even measure 🙂
If you can use a larger case and enlarge the board so you can space out the parts, so much the better.
I would replace every single thing, and I mean every single thing including the case. I wouldn't try buying another Krell power supply though.
This may give you some inspiration.
https://www.electronicdesign.com/ma...s-reverse-engineering-a-printed-circuit-board
https://www.electronicdesign.com/ma...s-reverse-engineering-a-printed-circuit-board
Along with a burnt power supply comes the question of why? Am I to assume that the preamp itself has no fault of its own?
@lxnay The preamp needs a volume control. I've got it but haven't installed it yet. The preamp PCB looks pristine, like it just left the factory. It's a beautiful piece of kit and I really want to bring it back.
I suspect the PS was overheating for reasons other than just the tiny poorly ventilated case. The Sprague caps all look pretty cooked, at least 2 out of 4 Roedensteins are bad and the rest are badly cracked. I'm not sure how far they were from exploding.
Another weird design choice Krell made is the DC-9 power connection. It's wired like this: +21V on 1 pin, -21V on 1 pin, ground on 3 pins. The other 4 pins are not used. Cable is a cheap RS-232, with 28AWG conductors as far as I can tell. That seems way undersized for the amount of power it's sending over an 8 ft cable.
I would prefer to upgrade the power connection, but interior space is very tight at both ends and both connectors are PCB-mounted. I'm thinking DB-3 connectors might do it, they would fit the existing case openings while allowing properly sized conductors to be used. More research is needed 🙂
I suspect the PS was overheating for reasons other than just the tiny poorly ventilated case. The Sprague caps all look pretty cooked, at least 2 out of 4 Roedensteins are bad and the rest are badly cracked. I'm not sure how far they were from exploding.
Another weird design choice Krell made is the DC-9 power connection. It's wired like this: +21V on 1 pin, -21V on 1 pin, ground on 3 pins. The other 4 pins are not used. Cable is a cheap RS-232, with 28AWG conductors as far as I can tell. That seems way undersized for the amount of power it's sending over an 8 ft cable.
I would prefer to upgrade the power connection, but interior space is very tight at both ends and both connectors are PCB-mounted. I'm thinking DB-3 connectors might do it, they would fit the existing case openings while allowing properly sized conductors to be used. More research is needed 🙂
Last edited:
A +/- 21v power supply is pretty simple. Does the board still work? You should figure out how much current the pre draws and proceed from there. I'm guessing a short in the PS caused extreme current draw. The case looks to have ok ventilation. I had a similar Krell pre, but it never ran super hot. I'm guessing a 60 watt power draw max.
@ulaby
The case has slots but only on top, so there is no convection. The PCB covers pretty much the entire area of the case, so the bottom of it cooks in a shallow, unventilated, essentially enclosed space.
Regardless, it probably shouldn't get that hot. Pretty sure you're right about a short in the PS causing it to overheat.
Here is the transformer rated 80VA
The case has slots but only on top, so there is no convection. The PCB covers pretty much the entire area of the case, so the bottom of it cooks in a shallow, unventilated, essentially enclosed space.
Regardless, it probably shouldn't get that hot. Pretty sure you're right about a short in the PS causing it to overheat.
Here is the transformer rated 80VA
Two things are on my mind about this problem.
1) What caused the power supply to overheat.
2) How much current draw does the preamp require.
The resulting failure has a cause and it hasn't been found yet. I wouldn't any possibility either with the power supply, cord, or trouble in the preamp itself. If it is possible to get a connector that would accept larger gauge wire that would be easy to recommend regardless. Same with the power supply itself. Those two things are obvious. Whatever happened will not fix itself. It must be discovered.
1) What caused the power supply to overheat.
2) How much current draw does the preamp require.
The resulting failure has a cause and it hasn't been found yet. I wouldn't any possibility either with the power supply, cord, or trouble in the preamp itself. If it is possible to get a connector that would accept larger gauge wire that would be easy to recommend regardless. Same with the power supply itself. Those two things are obvious. Whatever happened will not fix itself. It must be discovered.
Can you get a picture of the preamp board or a schematic? Looking to guess how much current is drawn from the +-21 volts.
If the preamp is drawing current within spec, I would guess they used fairly high voltages before the regulating devices causing alot of heat to dissipate in a poorly designed enclosure.
If the preamp is drawing current within spec, I would guess they used fairly high voltages before the regulating devices causing alot of heat to dissipate in a poorly designed enclosure.
Guys: No way that power supply is going to cook the board like that running normally. I've had a number of krell amps / preampss. They don't run that hot.
Unlikely the transformer failed. Shorted caps? I've never seen those little brown ones so not sure what their failure mode is. Even the best electrolytics only last 10k hours. This preamp is at least 25 years old.
Unlikely the transformer failed. Shorted caps? I've never seen those little brown ones so not sure what their failure mode is. Even the best electrolytics only last 10k hours. This preamp is at least 25 years old.
Of course, something gross happened there. But why? Any elcap near the heatsink is a red flag.
Edit: I'm not saying that's the case, but..
Edit: I'm not saying that's the case, but..
Last edited: