[Juergen]

Last night I was listening to my chassisless (is that even a word) Dynahi ( http://www6.head-fi.org/forums/showp...&postcount=901 )while researching some chassis options when all of the sudden I see a puff of smoke rise from the power supply

. I powered down immediately and did a quick inspection. The only problem I could detect visually was a small spattering of brownish liquid at the base of the right OPA541:

The OPA541 is obviously blown. It was late so I just went to bed. I'll have to start searching threads for troubleshooting tonight after work.

 

[DigiPete]

Bummer dude

Looks like the bottom of your right hand 541 is lacking insulation.
Perhaps blown off, or the pad was too small and thus causing the short?

Pete

 

[Juergen]

Quote:

Originally Posted by DigiPete Bummer dude Looks like the bottom of your right hand 541 is lacking insulation. Perhaps blown off, or the pad was too small and thus causing the short? Pete

It sure looks that way in the picture. I did check for continuity (lack of) between the 541 and heatsink before applying any power. Maybe after heating up, something changed. The heatsink didn't seem too hot when it was working, nowhere near the temp of the toasty amp board heatsinks. I kind of wish that was the problem since I would know what caused the problem. Unfortunately at the moment I have no idea why/what happened.

I will go ahead and order another 541 today in the hope that the other one is ok. Does anyone have any suggestions on where to start? It will be a few days at least before the replacement 541 arrives. Can I do anything in the mean time to start the troubleshooting? Should I remove the burnt 541 first and then check the other side (with amp modules disconnected of course).

 

[Pars]

Juergen,

You might check this thread, as Dan Gardner has observe the OPA541s letting go in certain (unexplained) situations...

 

[Juergen]

Quote:

Originally Posted by Pars Juergen, You might check this thread, as Dan Gardner has observe the OPA541s letting go in certain (unexplained) situations...

On it, thanks

Here's the poor victim:

I also had a chance to look at what DigiPete noticed. Basically only the top 30% or so of the 541s are touching the heatsink

The 338s are touching evenly. I checked continuity and the back of the 541 has no short to the heatsink. Maybe it burned out just from plain overheating since there was very little contact with the heasink?

Can I turn it on now w/o the one 541 removed (and no amp boards connected) for testing the other one? Will I cause more damage?

 

[DigiPete]

Digi-Key OPA541AP-ND is around $16.65; heat pad BER118-ND $1.38

The 541s I got from them had the whole back plane as a heat sink contact area.

Perhaps yours did not have enough heat sink contact area and overheated?

Dan Gardner has a nifty spreadsheet that helps you optimize heat load on the regs,
also a 30v transformer like he suggests (I used ones from Avel Y236356 30v x 80VA) cuts down on the heat load.

Pete

 

[chris719]

I do have a massive heatsink, but my dynahi psu with a 35V transformer does not get very hot at all. Far less heat than the dynahi itself and I am using identical heatsinks; it would be hard for me to believe the opa541 overheated. Unless, of course, it was not making contact. I actually used arctic alumina thermal epoxy to attach them to the heatsink. First I applied a very thin layer to the component back and the heatsink itself, then after that dried I epoxied them on and made sure there was no short.

 

[dip16amp]

You can power it up and measure the output voltage of the LM338 regulators as well as the output of the one OPA541. If one of the LM338 fails and puts 50 volts out, then the 38 volts from the other one will result in 88 volts across the OPA541 which exceeds the absolute maximum voltage limit of 80 volts.

One concern of using a transformer higher than 30 VAC was that it might result in 50 volts input to the LM338 and exceeding the 40 volt absolute maximum voltage limit during power up. It may only be a short amount of time before the LM338 output comes up in voltage but it may weaken it some.

I think someone had a LM338 go bad and it also took out both 541s.

 

[chris719]

The rating is differential, not maximum. If you look on the internet you can see examples of people using LM3xx regs at hundreds of volts. I use Linear LT1084 in my dynahi psu which is rated for even less voltage than the LM338 (35 I believe) and it has survived hundreds of power-on cycles at ~45V.

 

[dip16amp]

Yes, it is a voltage differential limit but it depends on the load and how much time during power up that it can survive. It may be nanoseconds, milliseconds, or full seconds. Not sure if anyone has tested how long a LM338 would last at a 50 volt differential with a one ampere load. Some may last longer than others and not have a problem. A cold power up would probably be less of a problem than a hot power up.

 

[chris719]

I think that if you do not bypass the adj with a capacitor then the output can never see the full differential, can't remember exactly.

 

[Juergen]

I took a few readings:

From 80VA +/-30V transformer: +35V, -35V
LM338 Left: Vout=35V, across Vout/Vin=10V
LM338 Right: Vout=0V, across Vout/Vin=10V
OPA541 Left (pin#): (1)5V,(2)5V,(3)-35V,(4)-35V,(5)30V,(6)-13V,(7)30V,(8)30V,(9)-13V,(10)35V,(11) 35V
OPA541 Right - Removed
PSU out +30: 30V
PSU out -30: 13V

It looks to me that everything besides the burnt out 541 is OK? Am I off base here?

 

[dip16amp]

The +/- 35 voltages from the LM338s seem good. The white lettering on the bad 541 looks burned off indicating over temp and not thermally bonded to the heatsink good enough. The other one looks like it still has white BB (R) lettering.

 

[Juergen]

Quote:

Originally Posted by dip16amp The +/- 35 voltages from the LM338s seem good. The white lettering on the bad 541 looks burned off indicating over temp and not thermally bonded to the heatsink good enough. The other one looks like it still has white BB (R) lettering.

Look to the right of the 'BB' in post #5 and you will see the little hole that erupted.

Here's a side pic of the remaining 541:

It's barely touching the heatsink. With the new 541 I can play with the leads a bit to get it closer to the heatsink but the existing one isn't going to move since it's soldered in. Removing it may prove to be extremely difficult. I had to clip the leads first on the burnt one to remove it. Moving the heatsink over a mm or 2 is impossible since the 338s are flat against it. Maybe I could put some sort of heat conducting shim between the 541 and heatsink to fill the gap?

 

[sft]

I suggest you to lock the 4 ICs (2 lm338s, 2opa541s) to the L bracket first, then attach to psu pcb and solder them last.
Its a not bad idea to lock part on HS first, then solder it to PCB to make sure perfect contact between part and HS.