No one has commented about the drastic rise in DC offset with no signal input at my friend's place. At home with no signal input I get 40 mV - there it was 250 mV. By what mechanism is DC offset load dependent?
Jeff, I think you had problems with oscillation.
If it was input or output related, you would have to test again to find out.
You could also try a 300~330pf cap directly across the chip's input pins.
But you have to make the tests on that place.
If it was input or output related, you would have to test again to find out.
You could also try a 300~330pf cap directly across the chip's input pins.
But you have to make the tests on that place.
carlosfm said:
OK, but I'd like to know how an oscillation (whether input or output related) can increase DC offset.
More information is needed. Have you repeatedly measured increased offset using the same meter? Does 'no signal input' mean nothing connected to the input? What was the condition at the output then? Cables/speakers connected?
Seems like a case of stage-fright on behalf of your amp 🙂
analog_sa said:
In order:
No - I measured twice on the night, but can't do so again not having my friend's speakers here.
Yes - nothing connected to the input. With my pre connected I get 2 mV DC offset - at my friend's I got 250 mV with input connected to DAC or pre or with nothing connected, it didn't matter.
As I said in the post above - 40 mV with my speakers and cables, 250 mV with my friend's speakers and cables, both measured with nothing connected to the input
pinkmouse said:
Why?
That will happen if you use a volume pot direct to the chip.
A bipolar op-amp is not happy working that way.
I would always use a series resistor on the input (220R~1K), so that a short never happens.
Otherwise, measure with an input impedance that will be more or less what the amp will usually "see".
Btw, in NI implementation, a coupling cap on the input (before the resistor to ground), besides removing DC from the source, makes the amp more "immune" to output DC-offset variations depending on what's connected to the inputs.
carlosfm said:
Why not? It's a perfectly feasable state for any op amp to be in.
I have a shorting phono plug I use for testing all amplifiers. Plugged into the input, it ensures that the dc offset that is measured is purely down to the amp circuit itself. Using the amp to minimise offset from other parts of the circuit is not very good design.
pinkmouse said:
I wouldn't say so.
Or do you listen to the amp at minimum volume?
Do you have the inputs shorted while listening?
This is a power (op)amp chip, and the most likely circumstances it will be used it with an active preamp, which has a constant output impedance.
This is also how it performs best.
Sticking a pot directly to the chip is not ideal.
pinkmouse said:
I have phono plugs with 10k and 47k resistors.
pinkmouse said:
😕
The input resistor to ground is part of the circuit itself.
The short is not.
pinkmouse said:
😱
I don't think I ever mentioned a pot.
Every op-amp stage should be trimmed, (if required) for minimal offset in a proper design.
Every op-amp stage should be trimmed, (if required) for minimal offset in a proper design.
pinkmouse said:
Then I can't understand how in operating conditions you will have the inputs shorted.
You must design for operating conditions.
Otherwise, you are measuring one thing in your bench, and then you'll have a completely different thing on your system.
pinkmouse said:
How do you trim a non-inverting LM3886?
You choose the component values that give you minimum DC-offset, and if you want really 0V DC, use the Ci caps.
Now... you choose the resistor values that give you minimum DC-offset and then short the input, I can't see the point.
You short the input so both inputs (via whatever other components you choose to use), of the op-amp are at zero potential to each other. This means any DC offset is being produced purely by the amplifier module, and can be trimmed out with appropriate component choice. Is that clear now?
Any DC offset from a source or pre-amp should be trimmed out in the same way if required. You should not use the power amp stage to sort out the problems of the rest of the circuit.
Any DC offset from a source or pre-amp should be trimmed out in the same way if required. You should not use the power amp stage to sort out the problems of the rest of the circuit.
pinkmouse said:
No, not clear.
It seems that you are making experiences instead of knowing what you are doing.
That method will not give you proper results, specially with bipolar op-amps.
pinkmouse said:
That's where the input coupling cap comes in.
You can't trim the amp circuit for every preamp or source you connect there.
I have a DC-coupled pre, with minimal DC-offset of it's own.
I prefer to have an input coupling cap on the power amp, the final stage.
pinkmouse said:
And who's doing that?😕
Forget it, you didn't understand nothing of what I said...
The end.
My last comment on this matter.
I prefer to have good design, meaning I need no coupling cap.
carlosfm said:
I prefer to have good design, meaning I need no coupling cap.
pinkmouse said:
For a moderator that you are, you should know how to behave.
I don't need the coupling cap too, but I use it.
Because the amp can be used everywhere and connected to everything, without problems.
Portable devices (like MDs, cdps, etc.) are a good example, they have some DC-offset on the outputs, because most share a common output for headphone and line out.
This small DC is multiplied by the gain of the pre and the amp, and it quickly goes to near a volt.
I can't be measuring DC each time I connect something to my system, portable or not, mine or not.
The cap is there to stay.
And THAT is good design.
pinkmouse said:
Read the rest of the thread.
Now, can anyone explain why the DC offset would increase with different speakers and speaker cables attached, but with the input open (aside from the 22k to ground)?
carlosfm said:
The one in a chassis is, but the one in question is a prototype and is not connected to the safety earth. It floats.
Clutching a loose straw:
Did you check his power? Maybe he has a crossed wire in his power socket and Active and Neutral are swapped.
I don't know if that would cause the problem you see, but it would be worth checking along with all the other stuff 🙂 Might as well check he has normal power voltage too...
Michael
Did you check his power? Maybe he has a crossed wire in his power socket and Active and Neutral are swapped.
I don't know if that would cause the problem you see, but it would be worth checking along with all the other stuff 🙂 Might as well check he has normal power voltage too...
Michael
- Status
- Not open for further replies.