How much DC is to much?

[roadkill]

Well, what I tested was how much DC is going to my speakers under normal operation.
So the system was playing music, it was not without inputs or under no load. I think this is a fair test. Am I wrong?

The music may have instantaneous peaks which are picked up by the DMM and shown as offset. Also, some music results in an asymmetric waveform (ie, it's positive more than it's negative or vice versa), causing a false offset reading. Please do the test with *no* input, keeping the input shorted to ground. The connection should be through a capacitor if your amp does not have a coupling cap, since if there's no coupling cap, the input bias currents can get amplified and cause an apparent offset.

This, incedentally, is a good reason why any amp without a DC servo must not be operated without a coupling cap.

[roibm]

Originally posted by carlmart

The tweeters should be protected, as long as they have a blocking capacitor in the xover.
What gainclone circuit are you using? IGC?

There is a cap in the xover.
The gainclone is from Brian(non-inverting, no coupling cap)

Originally posted by roadkill

The music may have instantaneous peaks which are picked up by the DMM and shown as offset.

That is not the case if I would use a sine wave, or?

Please do the test with *no* input, keeping the input shorted to ground. The connection should be through a capacitor if your amp does not have a coupling cap, since if there's no coupling cap, the input bias currents can get amplified and cause an apparent offset.

Who should I measure through a coupling cap? is that fair? I mean, doesn't the cap remove the DC component?

This, incedentally, is a good reason why any amp without a DC servo must not be operated without a coupling cap.

Well, I know a coupling cap is recommendable, but since I don't want to use one, I want to know how much DC is to much. That's all.
Peter Daniel said he is not using coupling cap in any of his gainclones, but I think he is to busy to participate in this thead.

[djmiddelkoop]

Well, I know a coupling cap is recommendable, but since I don't want to use one, I want to know how much DC is to much. That's all.

Well, the reason nobody answers this question is that it depends on what speaker you are connecting without a coupling capacitor.
i.e. every offset causes the connected speaker to move a little bit from the neutral rest position.
So the distorsion from the speaker will increase a little.
How much, and the sensetivity to this depends on the speaker used.

I for myself use 20mV as a margin to be on the safe side.

Dick.

[jackinnj]

What is being discussed is the DC Output error. Offset voltage is the no signal INPUT to the operational amplifier -- this arises because some current flows into or out of the INPUT pins. The DC offset voltage is the input bias current multiplied by whatever resistance attaches between the input pin and ground. The DC Output Component you are experiencing is this DC Offset multiplied by the gain of the opamp.

For the LM3875 the input bias current can be up to 0.2 uA but is specified as 0.01uA typical. Let's say you have a "WORST CASE" LM3875 with 0.2uA bias current and 11K resistance between ground and IN+, the offset is going to be 2.2mV on the input pin and 44mV on the output (gain of 20).

A really high quality opamp is going to have input bias currents in the hundreds of picoamperes (10e-12) range by comparison -- see this interesting Bob Pease link for measuring femotamperes! http://www.national.com/rap/Story/0,1562,5,00.html for comparison, but we aren't using gains of 10e5...

To tackle the problem the first thing is to make absolutely sure that there isn't some DC component on the output of your preamplifer! The second thing to do is make sure that the resistors are really soldered well. If this doesn't solve the problem you may have to resort to injecting a little DC bias into the IN+ pin of the NIGC. Since we are talking millivolts here, you would use a multi-turn 100K trimpot with the ends connected to the +/- rails and the wiper connected to the IN+ through a 10K resistor.

[roibm]

I have just measured the DC falling on the tweeter: 0.2mV
The DC falling on the midrange driver is about the same as at the outputs of the gainclone, so quite high.
Are are only the tweeters very sensitive to DC?

[Matttcattt]

On my bridge amp, the outputs are at 1v and 1.05v in relation to ground, so I only have 0.05v across the actuall output. I need to sort this sometime...

[carlmart]

Originally posted by Matttcattt
On my bridge amp, the outputs are at 1v and 1.05v in relation to ground, so I only have 0.05v across the actuall output. I need to sort this sometime...

It's too much.

Are you using the circuit on Fig 4 on your site? Put a 220 K resistor or less from every non-inverting input to ground. That should cure the offset.


Carlos

[Matttcattt]

Originally posted by carlmart


It's too much.

Are you using the circuit on Fig 4 on your site? Put a 220 K resistor or less from every non-inverting input to ground. That should cure the offset.


Carlos

Technically im using Fig 5, but its the same pinciple.

Why would this make a difference? Should I use variable resistors so that I can zero the offset?