[dip16amp]

I noticed the hole eruption but wasn't sure if it was caused by overvoltage or overtemp. Since the LM338s are since good, it didn't have an overvoltage. The burned off white lettering is a good overtemp indicator. All four of my OPA541s still have all the white lettering after more than a year of daily use.

Maybe you can put a notch in the heatsink for the board or trim the edge of the board so that it doesn't touch the heatsink and will let the 541 mount flush on it.

 

[chris719]

The white lettering was burned off does not necessarily suggest they were killed by heat in the first place, although possible. I've killed one before and in the failed state, applying power to them causes extreme heat and smoking.

 

[Juergen]

Quote:

Originally Posted by 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.

Actually I did that. I placed the 338s & 541s in the pcb holes and then attached them to the heatsink with the bolts then soldered them in. Unfortunately I just pulled the 541s to the heatsnk from the bolt which is near the top, but the bottom of the 541 didn't move. The 338s and 541s are not in the same plane. The 541s are further back. I should have noticed.

Quote:

Originally Posted by dip16amp Maybe you can put a notch in the heatsink for the board or trim the edge of the board so that it doesn't touch the heatsink and will let the 541 mount flush on it.

The part of the pcb under the heatsink has several traces so it couldn't really be trimmed plus the issue of different planes would still be there.

I tried desoldering the remaining 541. I got 99% of the solder but that darn 1% on 11 pins is keeping it in place. I will probably shim that one when I figure out what I can use. With the new 541 I will try bending the pins first.

 

[sft]

Quote:

Originally Posted by Juergen Actually I did that. I placed the 338s & 541s in the pcb holes and then attached them to the heatsink with the bolts then soldered them in. Unfortunately I just pulled the 541s to the heatsnk from the bolt which is near the top, but the bottom of the 541 didn't move. The 338s and 541s are not in the same plane. The 541s are further back. I should have noticed.

sorry there are some misunderstandings due to my poor english,let me explain more in details.

1) attach 338s and 541s to the L bracket, so now all contacts are perfect.

2) combine the (L bracket + ICs) unit with PCB.
2.1)you have to bend some IC's leg a little to put them all into pcb holes. make sure all conteacts are perfect still. It's easy.
2.2)Lock the L bracket to PCB

3) solder

 

[Juergen]

Quote:

Originally Posted by sft sorry there are some misunderstandings due to my poor english,let me explain more in details.

Your english is just fine, it's my understanding that is poor

 

[Juergen]

Here we go again

!

!

I got the new 541 today and soldered it in. Got good contact with the heatsink. Plugged it in. Started taking readings (no load connected)

PSU out +30: 30V
PSU out -30: -30V (so far so good)

From 80VA +/-30V transformer: +34.8V, -34.8V

LM338 Left: Vout=35.8V, across Vout/Vin=9.6V
LM338 Right: Vout=0V, across Vout/Vin=9.5V

OPA541 Left (pin#): (1)5V,(2)5V,(3)-35.8V,(4)-35.8V,(5)30V,(6)0V,(7)30V,(8)0V,(9)0V,(10)35.7V,(1 1) 35.7V
OPA541 Right - (1)0V,(2)0V,(3)-35.7V,(4)... pop...smoke!

There is now a hole the size of a small pea in about the same spot as the previously blown 541. The 2-338s and 541s were barely above room temperature. I can't see how this could be caused by overheating. I was getting +/-30v so all seemed good, for about 10 minutes. Do any of the above readings point to anything?

 

[dip16amp]

There are four holes where traces on the top of the board connect to traces on the bottom of the board. Should those have a wire soldered in them to ensure a good connection?

 

[amb]

Quote:

Originally Posted by dip16amp There are four holes where traces on the top of the board connect to traces on the bottom of the board. Should those have a wire soldered in them to ensure a good connection?

If the board is properly plated-through (as they should be), then it wouldn't be necessary to solder those vias. I am often tempted to solder them anyway.

 

[amb]

Quote:

Originally Posted by Juergen Here we go again ! ! ... I got the new 541 today and soldered it in. OPA541 Right - (1)0V,(2)0V,(3)-35.7V,(4)... pop...smoke!

Did you check how much current the amp is pulling from the PSU? If your amp boards are running the usual ~300mA each, then the dissipation on the OPA541 should be 4.8V * 0.6A ~= 2.88W. Not bad at all. The LM338 on the other hand is dissipating 9.5V * 0.6A = 5.7W which makes them hotter but should still be safe if the heatsink is adequate.

Sorry about the loss of those OPA541s... they are pricey buggers.

 

[Juergen]

Quote:

Originally Posted by amb Did you check how much current the amp is pulling from the PSU?

The amp boards were not connected yet this time. The right LM338 had a Vout of 0V, while the left was 35.8V. Does that make sense?

 

[amb]

Quote:

Originally Posted by Juergen The amp boards were not connected yet this time. The right LM338 had a Vout of 0V, while the right was 35.8V. Does that make sense?

I think you mean the left LM338 outputs 35.8V while the right LM338 outputs 0V (both relative to ground). If that is so, then yes it's correct. See the schematic -- the second LM338's output pin is tied to ground, this is because it's a positive regulator and is being used in an "inverted" manner to derive a negative voltage rail.

If your amp boards are not connected to the PSU and there is no load at the output, then the LM338s and OPA541s shouldn't even get warm as there is very little current flowing through them. As for why your second OPA541 popped, could there be an accidental short circuit somewhere? Maybe as you were mucking with it something touched something it shouldn't have? Solder bridge? Loose wire? Inspect the board carefully...

 

[Juergen]

Quote:

Originally Posted by amb I think you mean the left LM338 outputs 35.8V while the right LM338 outputs 0V (both relative to ground). ...

You're right....corrected...thanks

I was hoping that some of my readings were off so that they might give a clue as to where the problem is. Unfortunately I wasn't able to get all of the readings for the right 541 before it blew. What's really weird to me is that I was getting a perfect +/- 30V on the output.

 

[ScubaSteve87]

smoke is a bad thing i would guess. The thing that kills me is the smell of burnt out electronics. Its just plain rank

 

[JahJahBinks]

a moment of silence for the small chip.

 

[dgardner]

I have been meaning to post something helpful to the victims of OPA541 smokage. I'm not sure that I can. This same thing happened to me several months ago when I built up my first Dynahi PSU board. At that time, I was using a 38VAC transformer which was way too high, and I was also using a relatively small heatsink. I was using a thermocouple to watch all of the temps on all of the tabs of LM338's and the OPA541's. As expected the temp got really high. The tab got to around 90C on the LM338s in this bad setup. OPA541's were a few degress cooler.

One of my OPA541's smoked. I thought for a moment that I may have shorted something with the tip of thermocouple, or that one of LM338's let go due to overheating. Upset, I removed the blown OPA541, which is a challenge in itself. When I put in a new part, along with a much lower transformer, it blew again! My immediate thought was that something else in the circuit (possibly upstream from the OPA541) was damaged. I am positive that my insulators from the sink were done properly and my solder joints were perfect.

I am mad now that I did not track down the failure to a specific root cause. To solve my problem, I used conservative brute force. I started with a new PSU PCB and moved all the big caps, precision resistors, rectifier diodes, and connectors from the blown board. Then I put in fresh LM338's and OPA541's and even a new REF02. It worked.

So the moral of my story is that if you blow your OPA541, all the IC's are suspect. You may very well blow the OPA541 replacements, unless someone can help track this down to a specific root cause. Blowing $16 parts is truly depressing.