Help verifying LM3875 GC implementation

[kneadle]

Hello,

I'm building a P2P GC based roughly on the BrianGT kit. One important variation is that I have some uninsulated chips.

So now I've verified that the power supply is good to go. And I'm pretty sure I have the chip well-constructed. However, when I powered it up--without any signal input or output connectors, only power--I found a DC offset of roughly 7 VDC.

Now, since the thing is uninsulated, the heatsink is carrying a nice charge of 37VDC, the rectified power supply voltage.

So, now, I need to learn how to diagnose and verify what I'm doing. My best first guess is that I need to re-verify the P2P connections on the chip's pins. The second best guess is that I have a grounding problem.

What, my brain wishes to resolve, is the most likely reason for a 7 volt DC offset? It isn't 34 VDC, nor is it 60mV. So, I have roughly seven stray volts.

Orderly advice is greeted with thanks.

Dave

[Russ White]

I would start by isolating your chips.

[leadbelly]

Originally posted by kneadle
Now, since the thing is uninsulated, the heatsink is carrying a nice charge of 37VDC, the rectified power supply voltage.

Well, considering your comments, the uninsulated package does raise alarms. If you are using the chassis as ground or just have connected your input RCA connectors to the heatsink through the chassis, that would certainly explain the offset.

How are you planning to implement the uninsulated chip? Do you plan to leave the heatsink at V- and just insulate the whole heatsink from the chassis? Or do you plan to install insulators between chip and heatsink and have skipped that step for a trial assembly?

[kneadle]

Originally posted by Russ White
I would start by isolating your chips.

OK, how?

Originally posted by leadbelly


Well, considering your comments, the uninsulated package does raise alarms. If you are using the chassis as ground or just have connected your input RCA connectors to the heatsink through the chassis, that would certainly explain the offset.

How are you planning to implement the uninsulated chip? Do you plan to leave the heatsink at V- and just insulate the whole heatsink from the chassis? Or do you plan to install insulators between chip and heatsink and have skipped that step for a trial assembly?

Right. First, my case is non-metal (PVC, to be exact). I have a star ground connecting the decoupling caps and the AC mains. Thus, the heatsink is completely isolated, yet carries the charge.

In addition, there are no RCA connectors yet. Should I go ahead and install them for testing? I'm not quite there yet when it comes to understanding the NFB chipamp structure. I figured that having the circuit built would be fine for testing the power stuff. I didn't think that the signal circuit would have to be complete as well.

Dave

**edit** I guess my real question is how the uninsulated package is causing some sort of voltage problem, especially if the circuit is otherwise all OK.

[leadbelly]

Originally posted by kneadle
**edit** I guess my real question is how the uninsulated package is causing some sort of voltage problem, especially if the circuit is otherwise all OK.

It should not. And if you want to build your amp so that the heatsink sits at V- the whole time, that is perfectly acceptable. Just make sure that it is very solidly mounted so that nudging or bumping it accidentally will not cause a short circuit. Similarly, it should not protude from the enclosure to prevent it from coming into contact with something outside and shorting.

So if you are telling us that your heatsink is at V- but it is not touching any part of the circuit, then all I can say is to keep checking your connections.

[kneadle]

A photo attached

black is V+, red is V- and white is ground.

I'll double and triple check the connections and report back. Perhaps I'll build the other circuit to see if I have the same problem, eh?

Dave

[leadbelly]

I see 2 resistors not connected on one end. I assume 1 is the input resistor, what is the other?

[kneadle]

Uh...That is Ri, to connect the FB resistor to ground.

I'm just going to keep going down this road...

So whaddya think? Should I connect that to the ground somehow?



Dave

[leadbelly]

So, you are building this picture:



If I had to take a guess, I would think that maybe you are not connecting SGND to GND? They need to be connected by a wire.

[kneadle]

Ok. That's helpful. I have been under the impression that signal grounds and power grounds should forever separate; I don't know where I got that idea, though. Now that I think about it, it doesn't make sense to keep them separate completely. Is there a rule of thumb where along the power paths to bring the two together?