Can wrong input termination cause NIGC oscillation?

[wes-ninja250]

I've been trying to figure out why my GC is buzzing:
Another Buzzing GC Query

It's a BrianGT LM3886 dual mono kit with common transformer. I'm thinking that perhaps my amp is just barely oscillating. It also runs quite warm, but not so hot that it stops working, even after extended idling.

The buzz is quiet, but audible from a few inches. Last night, I found out that adding chassis ground to the stars completely elminates the LOUD buzz that happens when there is no input connected. But the buzz is still there when I plug in an RCA cable, unless the cable is shorted at its ends. Note that this effect is identical on both channels, and one channel does not affect the other -- i.e. I can connect a single RCA input and only one have that channel buzzes.

So, the settling down when chassis grounded and the presence of buzz when there is an input wire is why I think I may have an oscillation issue. Maybe the input wiring is acting as an antenna and picking up stuff from the speaker leads?

My heat sinks are chassis grounded by design. I'll double check by meter soon. The inputs are sent through sheilded twisted pair, and are not adjacent to the output wiring (except where they meet.. amp board and backplane.. and they're as far apart as possible there, too). The input shield drain wire is connected at one end to the chassis.

Is it possible that the way I've terminated the input bus is wrong?

The attachment shows how Brian's manual says to wire up the pot, and the way I've done things (it's just a power amp).

Oh, I didn't have any problems on the bench, only when I threw it in my equipment rack. BUT -- on the bench I was using a battery-powered CD player and different speakers (same impedance, bigger drivers).

Here are some photos of the amplifier: http://flickr.com/photos/shellyandwes/sets/1291427/

Photos Of interest --
Back of amplifer; 1/4" jacks are in parallel with nearest other connector (RCA or binding post): http://flickr.com/photos/shellyandwe...n/set-1291427/

Side view of left channel; Power on green/black and yellowbrown/black pairs. Output on white/black pair. Input on red/black pair. Input shield connected to bolt at top. Orange Marrette are on CHG points of PCB; connectors in the photos are now meeting at the chassis. http://flickr.com/photos/shellyandwe...n/set-1291427/

Top view -- wires coming from the center of the backplane are inputs. Wires coming from the edge are outputs. http://flickr.com/photos/shellyandwe...n/set-1291427/

If you have a [free] flickr account, you can blow the pictures up even bigger.

Thanks for any input. In case you're wondering why I'm not doing more trial-and-error debugging -- this amp is REAL PAIN to disassemble. The cover's easy to open, but it's nearly impossible to get the amp boards out of the chassis. I should have used more quick-connects! So, I'm trying to "think out" the solution instead of just randomly disassembling/reassembling stuff.

Plus, I don't want to be without the amp for long. From my normal listening position, it's GRRRREAT!

Wes

[carlosfm]

Wes, put a resistor across each RCA plug (line in).
100K will do.

[wes-ninja250]

Hi, Carlos!

Just wanted to make sure we're on the same page here. Attached is a pretty good drawing of how each channel is wired up.

Does your suggestion for the ~100K resistor on the RCA plug still stand? (Does type matter here?) The blue one in my drawing is a 22k metal film.

Thanks,
Wes

[carlosfm]

Originally posted by wes-ninja250
Hi, Carlos!

Just wanted to make sure we're on the same page here. Attached is a pretty good drawing of how each channel is wired up.

Does your suggestion for the ~100K resistor on the RCA plug still stand? (Does type matter here?) The blue one in my drawing is a 22k metal film.

Thanks,
Wes

The pot confused me.
So, this is a power amp, right?
My suggestion of 100k across the RCAs was if you use a 10k~20k pot at some distance from the inputs.
22k is fine, then. Metal film is fine too.
Is that wiring short? It's not very clear on your pics.
Btw have you tried a small cap (200~300pf) directly across the chip's input pins?

[wes-ninja250]

Carlos:

The wiring is relatively short. Total length of the input wire is.. oh, 12cm or so. The unshielded portion is about 2cm at the amp end and about 3cm at the connector end. Basically, as short as I could make it and still have it fit. Where the shielding was taken off, twists were preserved as far as possible.

Wes