High DC offset with no source (LM4780)

[Tomten]

peranders, I'm gonna measure and see if I find something. Yes, I'm pretty sure that every resistor have the right value, I measured them all before installing them.

[pacificblue]

200 mV / 33,35 = 6 mV at the input. 6 mA across a 22k resistor corresponds to 0,27 µA. Above the typical value, but well within the LM4780's specs.

3,5 V / 33,35 = 105 mV at the input, which corresponds to ~4,8 µA. That is too much. The other way round, if 0,2..1 µA produce 3,5 V, there must be 3M5..17M5 involved, which would mean the 22k resistor is either open circuit or you need to check the solder joints.

[peranders]

Remember that the 22k goes to two LM4780 => double input bias current

[Tomten]

When I measure directly from input to the chassi ground I get 22k, shouldnt that mean that the resistor does its job?

[peranders]

How much voltage do you have across it?

[Tomten]

I have 6mV and 5mV on the other channel.

Another thing, I use 1k reistors on R6/R7 instead of the suggested 270R (http://www.audiosector.com/lm4780%20amp.pdf) because nationals PDF suggested it, if that have something to do with it (I dont think it should?).

[danielwritesbac]

Originally Posted by Tomten

When I measure directly from input to the chassis ground I get 22k, shouldn't that mean that the resistor does its job?

If the DC coupled amplifier has insufficient input load (at DC), then it will make offset.

Problem:
The spec is 22k resistor paralleled with 20k pot. So, if half the load is missing or if there's a capacitor in-between (problem if half the load is missing for DC) then offset is high.

Symptom: The amplifier acts like the potentiometer is missing.

Patch: You can use 12k instead of your 22k input load (R0/R4). This is a load at (upon) DC. It compensates for a missing potentiometer.
-or-
Patch: You can move your input filter cap from between the amp and pot, then re-install this cap between pot and rca jack. This repairs "cap in the wrong place" problem, and it also provides protection for the potentiometer.

[danielwritesbac]

Originally Posted by Tomten

I have 6mV and 5mV on the other channel.

Another thing, I use 1k reistors on R6/R7 instead of the suggested 270R (http://www.audiosector.com/lm4780%20amp.pdf) because nationals PDF suggested it, if that have something to do with it (I dont think it should?).

Why mix two different designs at random?
Audiosector design:
Maximum value appears to be 680R.
Since input load is half of feedback value then a more appropriate value appears to be 330R, and so 270R (in the directions that you didn't follow) is close enough.

[Tomten]

The 680R is for R3/R5 which i use. Or do you really mean R6/R7?

I dont use a potentiometer, its just a power amplifier. The problem is the huge difference in offset. I was thinking if I made a bad solder on the jumper, that one on the internal chips sees no load (R0) and then makes this high offset, but I cant see any problems on the solder. I have to check it out.

[danielwritesbac]

Originally Posted by Tomten

The 680R is for R3/R5 which i use. Or do you really mean R6/R7?

I dont use a potentiometer, its just a power amplifier.

Aha!

Here's some component values for you then.

R0, R4 are 12k for when/if you don't have a potentiometer.
(10k is workable)

R6, R7 are 270R (330R is workable)
A too-high value of R6, R7 will increase DC offset with that particular amplifier board.