Strange Power Problem, no negative rail?

[Adam M.]

I was trying to finish my amp tonight, and everything was going well. I attached the heatsinks permanently using sillicone caulk to the chassis, with a layer of electrical tape on the bottom of the heatsinks to insulate it. (the heatsinks are painted, but I did this as an added security measure). I added an LED as a power indicator, connecting it to + on the bridge and power star ground.

When I went to test the amp before I closed the case(it did work earlier this morning), I got no output. After some troubleshooting, I found that I was not getting anything on the negative rail, with reference to ground, but the first time I touched the negative rail with the power off, I felt a small shock.

I verified all connections, and checked for shorts anywhere I could. All +output tabs real about 21k to ground, all ground are less than .2 ohm. All input tabs are between 48k and 52k to ground, input ground is less than .2.

Removed the LED, no difference.

I think the problem is the heatsinks not being isolated properly. (I am not using insulated chips)

THe metal tab on the lm3886 to ground were all above 3Mohm, but only 2 were off the upper limit of my meter.

Does anyone have any ideas before I take the heatsinks back off, I'd rather not do that.

THanks,
Adam

[thomas997]

Is silicone caulk what you have between the chip and the heatsink? Is that a good enough thermal path?


Although even if you have the chips not heatsinked at all, it should still work.

Is there voltage from ground to the heatsinks?


Have you measured the AC voltages from the transformer? Its possible the rectifier is messed up.


hm

[Adam M.]

The caulk is between the bottom of the heatsink and the chassis, the chip is held to the heatsink with a screw.

there is .9V between the heatsink and ground, where it should be around 35.

Both secondaries on the transformer put our 27V (25v toroid, so it seems correct) Does that lead you to believe the problem is the bridge? How could it have been damaged?

[lazyfly]

Wont a 25V put out 37.5V rails?

Just a typo or I'm lost!

[Tim__x]

25*1.414=35.35-1.2 (diode losses)=34.15

[Nuuk]

You measured the resitance from the voltage pins of the chip to ground but you should be measuring the voltages on those pins.

[Adam M.]

Voltage from pin 4 of each chip to ground is the same as I get from the recifier to ground, about .9v I only measured the resistance from the heatsink to ground to make sure they were isolated from the chassis and each other.

[Nuuk]

Voltage from pin 4 of each chip to ground is the same as I get from the recifier to ground, about .9v I only measured the resistance from the heatsink to ground to make sure they were isolated from the chassis and each other.

So you have 0.9 volts coming off the rectifier instead of the 35+!

Looks like you have a problem with your rectifier or the connections to the transformer.

Do you have a discrete regulator or ready-made variety?

[Adam M.]

I assume you mean the bridge? It is a ready made, attached to the bottom of my case to keep it a little bit cooler, even though I never noticed it running hot. Its rated for something like 400v/25A (I don't remember exactly, but it was overkill)

[carlosfm]

On the LM3886 the tab is connected to V-.
I presume that, although you used electrical tape , you don't talk about plastic spacers to isolate the screw from the tab.
This would still not be drastic.
What's bad is that I presume that the heatsink is not really isolated from the case.
What's more, have you grounded the case?