help with op amp headphone amp hum and distortion

[gregas]

Hey Rick,
Excuse my ignorance, most of this is way over my head.
What exactly does this latest circuit do? To adjust temperature of the chip?
What do you mean by zero output offset ?

[Rick PA Stadel]

No need for excusing -- it's on me, I should've done some 'splainin'.

As long as you have the 220uF non-polar capacitor between pin 7 and the headphones, this is Not Needed.

But there are folks hereabouts (and elsewhere ) that aren't too fond of capacitors in the signal path, especially inexpensive aluminum electrolytics! This is a way to eliminate that.

The designers of of the AD817 were expecting the surrounding circuitry to be low impedance -- achieving the gain and phase linearity for NTSC video requires it. The beginning of the offset trouble was choosing the 100k: The part has an input current in the 3uA range, and they are NPN transistors. Not terrible, but it means that the '+' input will rest at about -300mV with no signal. That's too much for headphones, thus J-P's solution using the 220uF to protect them.

So, just adjust the trimpot to minimize both AD817's offset at pin 7. Both devices won't come to 0 at exactly the same point. But it should be possible to get within a few millivolts, which is low enough to be OK. And it should be stable over about a 20°C range.

The diode is there to partially correct for the change in AD817 input current over temperature -- about 3,2uA a little below normal indoor room temp; ~2,7uA at the upper end of the range this circuit can correct.

Keep asking good questions!

Cheers

[gregas]

Hey Rick,
Thanks for the explanation.
So this circuit connects to pin 7 of both ad817 chips.
Does the orginal +15v still connect at pin 7?
So now where would be a good place to put this circuit?
Also should I move the power supply for the headphones?
Hard to move forward without a good plan.

[Rick PA Stadel]

Nope, nope, nope -- my bad. I gave the wrong pin # for the output pin: That phrase in the earlier post should read ".. AD817's offset at pin 6." That's where to measure the offset. Pin 7 still needs its +15V.

The best place to put it would be a little separate perfboard (potentially much more compact than Vero which forces orthogonal routing), reasonably close to the AD817s -- if possible close enough to 'share temperature'. The important thing to keep in mind about this little ditty is, it is specific to the AD817. Any of the op-amps better suited to this duty will not use it.

If you haven't already moved the power supply, save that for later. It probably should be done eventually, but 1) isn't likely to be the cause of the main current symptom, and 2) can't be evaluated for success until the offset problem has surrendered.

Are the 220uF non-polars in place on the outputs?

Ya know, perfboard is usually much cheaper than Vero board. If you use 1/4W resistors and a non-huge trimpot and capacitor, this little bit of circuit ought to fit easily on 2½ by 2½ cm of board. And be light enough to simply piggy-back over the AD817-area of the HPA, supported by its own connecting wires.

Below is a little VCO example from 30-some years ago, complete with dust and pet hair. It has more and larger parts, but about the same 9 nodes as your bias circuit. About 1 square inch.

Regards

[gregas]

Hey Rick,
Ahh i see the circuit connects to the outputs.
Thanks for the tip on perf perf board. I don't see your picture.
I've only used veroboard so far. This is my third audio project.
So I'll have to learn how to use perf board. I recall watching a video about it a while back.
So I have an idea.
Any recommendations on brand or type and where to buy? I found that BusBoard Prototype Systems made good stripboard.
I haven't made any changes yet. I have to order parts first and figure the details.
Btw the muting circuit works perfectly. The sound of those relays firing is very satisfying.

[jean-paul]

It would be more standard and more efficient to start designing with a PCB program. Your devices will very likely perform better with a decent PCB.

[Rick PA Stadel]

Originally Posted by gregas

Ahh i see the circuit connects to the outputs.

Umm .. no, the 68k resistors are to provide the INPUT bias, and will alter the output quiescent voltage.

Sorry about the missing pics -- no doubt more *pilot error*.

Big 'Yay' on the mute relays!

Cheers

[gregas]

Thanks for the clarification. I'm going to put in an order. I don't have any 1/4w resistors. I've always purchased 1/2w. Any 10k potentiometer will work? I think i have some 5k ones.

JP perhaps on my next project I can try a pcb.

[Rick PA Stadel]

If you're still doing or planning mostly valve work, staying with ½W as a 'house standard' might be a reasonable choice -- otherwise I don't see them much anymore. I didn't mean to suggest changing to 1/4W. The convenience of size is compelling for me, especially since the power levels rarely exceed a few milliwatts or 10's of 'em. And the leads plug into the proto-boards I have a lot better. Power supplies are the only place I fit bigger, as necessary.

Yes, any 10k pot will work fine.

Without re-doing the arithmetic (took me too long the first time), I'll guess there's a halfway decent chance the 'sweet spot' will be within range. If not, just add in a 4k7 on whichever end you need to move it (the sweet spot).

Regards

[Rick PA Stadel]

Sorry -- the line, ".. just add in a 4k7 on whichever end you need to move it (the sweet spot)."

should instead read, ".. on the opposite end from the one where you're hitting the end stop."