Headphone amp rating

[Brisance]

So I am designing my second headphone amp(first was a simple LM386 one), this time I think with an Op amp preamp/buffer and possibly a class A NPN output stage.

The question is what Power rating would be good to design for headphones? One watt? Less?

[antonis]

Depends on YOUR headphone power rating...

[Brisance]

I want it to support a wide variety of headphones, so more of a rule of thumb question

[antonis]

Although I think that there isn't such a rule of thumb, 50-60mW should be OK...

(you may calculate it for your A class amp design..)

[merlinb]

Average listening level is around 500uW. Max power of 50mW is good enough for casual purposes, but you might prefer 100mW for total flexibility. Don't go over 500mW or you may accidentally expose your ears to dangerous SPL (or burn out the phones...)

[Brisance]

Oh, that's a lot less, than I thought. I might even use one of my vintage soviet Germaniums then for mojo, such as this one http://alltransistors.com/transistor.php?transistor=41798

[JFace]

I've built several headphone amps, the first one using the LM386. After being very disappointed with the noise of that chip, I sought some Hi-Fi solutions.

One method is to use opamps in parallel, like in this article: http://www.ti.com/lit/an/sboa031/sboa031.pdf
Some chips to try in this arrangement are NE5532 and RC4580

I have been very pleased with using the NJM4556 as a one chip stereo solution. Just use each opamp as a non-inverting buffer, one for left and one for right, use a 10 ohm build out resistor with a large capacitor (hundreds of uF) to block DC, and it will run any pair of headphones with a 9VDC supply (you probably need about 100mA and good power line reservoir capacitors for any of these implementations). I use this chip to run a 20 ohm pair of in-ears at unity gain, and the quality is excellent. I would suggest breadboarding your design to get your levels right...these chips can easily damage your hearing if you aren't careful.

[bool]

I don't think that you'd like any of "soviet germanium mojo" with a headphone amp ...

But wrt. your plan to use an opamp with a NPN class-A output stage - if your power needs aren't excessive, a lowly MPSA13 darlington follower will hapily run headphones on smaller power supply voltages. Just don't overspec your current-sink so it doesn't overheat.

(Or google for some "solid state white follower" variants to improve load drive ability).

[Brisance]

Thanks everyone, some nice ideas here, although JFace, I do not need stereo, it's for guitar so I can practice without pissing off the girlfriend.

I already build one, with a LM386, but I do not like it's lack of headroom and the fact, that if it gets pushed even a tiny bit into overdrive, the clipping sounds crap. Both channels of headphones were just connected to a single output.

This time I will either use 2 9V batteries or a charge pump, now especially since the headphones need a lot less power, than I thought, which fit well into the limit of negative 100mA on the pump. And I won't run it off an adapter, since I like it to be portable and not depend on power availability, so I can practice anywhere.

EDIT: also I do not want any existing schematics, because that will take away the fun of learning and experimentation

[JFace]

Thanks everyone, some nice ideas here, although JFace, I do not need stereo, it's for guitar so I can practice without pissing off the girlfriend.

I already build one, with a LM386, but I do not like it's lack of headroom and the fact, that if it gets pushed even a tiny bit into overdrive, the clipping sounds crap. Both channels of headphones were just connected to a single output.

This time I will either use 2 9V batteries or a charge pump, now especially since the headphones need a lot less power, than I thought, which fit well into the limit of negative 100mA on the pump. And I won't run it off an adapter, since I like it to be portable and not depend on power availability, so I can practice anywhere.

EDIT: also I do not want any existing schematics, because that will take away the fun of learning and experimentation

Your input may not be stereo, but your output is (left ear, right ear). I messed around with a charge pump, but I noticed distortion when I pushed the volume a bit. I found 9V is more than adequate, so long as the current capacity is there. Nothing wrong with using an existing schematic as a starting point. You can tweak/add from there to make it your own. No reason to re-invent the wheel 

[Brisance]

Good idea about the stereo, what I did was just run the both earphone speakers in parallel, on my favourite headphones(Sony MDR 7506), they give about 30 ohm impedance, which I deemed to be enough, but on some cheap low impedance ones, this might not be the case.

[reidk1960]

If you do decide to go for stereo, consider adding crossfeed.  At first
the idea of reducing channel separation seems strange, but with
headphones it can be a real improvement.  Ever listen to the first
two chords of the Rolling Stones' Brown Sugar through headphones?
They do some fast panning that sounds bizarre without crossfeed.

Here's an article about it: http://headwize.com/?page_id=403

I'd also suggest a good look at some of the other articles Headwize.com.

Have fun.

Reid

PS Have you thought about tubes?

[Brisance]

Hmm, that's a good idea, maybe I will make it with multiple inputs (one for mono guitar and maybe dual RCAs). That way I will probably add panpots for each channel(trimmers, too many knobs is not, what I want, but stereo separation looks like a good thing to have a possibility to tweak.

Tubes could be cool, but I want to run it from battery and the heater current draw will eat them too fast. Maybe I will make 2 versions, one portable battery version and another adapter powered tube version one day

I want it to be functional and nice sounding, but experimentation with different stuff is at least as important for me, maybe if all else fails, I will start with an exsisting circuit

[JFace]

A stereo input jack would be a nice addition...that way you can play along with mp3s. I like using 1/4" stereo jacks, and use a 1/4 to 1/8 converter if necessary. You can use a dual gang 100K audio to control the volume of both channels simultaneously.

[Brisance]

Okay, got the rudimentary circuit down, needs tweaking of course, but what do you think?

click to enlarge

[PRR]

> what do you think?

Can't think nothing without some numbers around the transistor.

Seriously: 2N1893? You have some? That's an 80V part. You have 18V, and in an essay which I did not post (because you want "the fun of learning and experimentation"), I showed that +/-5V is usually ample for any <300r headphone.

Topology problem: what happens when you turn R2 all the way? Dead-short from +9V through Base-Emitter junction to -9V. You rarely find much more than 0.6V or 0.7V across a B-E junction. Base current is an exponential function of voltage. Current will be very nearly infinity. Something will burn. With luck, the pot. Maybe the battery. Maybe your 2N1893.

An eerily similar issue at the un-named 100K pots at the first op-amp. Turn them all the way, gain (signal and DC) goes to infinity. Nothing will smoke, but you do not ever want infinite gain. *Usually* at this point in an audio path we will use a straight Volume potentiometer followed by a fixed-gain stage.

Back to the back-end: *What value is R3?* Even a clue? The DC load resistor must be selected in terms of the audio load impedance and the stage dissipation. Tip: the efficiency of this type coupling is never more than 8.6%, and usually much less. So the Q1+R3 dissipation must be MUCH more than your desired power to the headphones.

A hard-grounded Emitter amplifier has low input impedance, high distortion (~~26% THD) at high current swing, and the maximum un-clipped input is maybe 50mV. It is also very hard to bias-up stably, but stable bias is critical to power stages. Especially with a "mere" 800mW transistor which is hardly ample for the job. There is, IMHO, a much better way using the parts you show. If you insist on split supply (not optimal for heavy resistor-loaded coupling), you could even eliminate the output cap.

The inefficiency of a single-ended resistor-coupled amp is SO bad that I think battery operation is folly. For rare 50mA peaks you must suck a *constant* 1 Watt from the battery. Choke-loaded or A push-pull is 45% efficient, only 0.11 Watts of waste/drain. Push-pull class B is 70+% efficent at full load and its drain *declines* with signal level; average speech/music drain might be under 0.01 Watts. A 1:100 difference of battery life.

Why the split supply? In general a single-ended R-Q stage with a heavy load should be biased significantly offset from "half supply". And experimentally the output cap gives some safety against breadboard accidents. Then it could as well be done with a single supply.

A side-quibble-- surely there is a simpler cross-feed? Maybe two resistors and one pot?

[Unlikekurt]

Try something like a TDA2822
Class AB.  Can run happily off of a 5v supply (use a regulator).
Mono or Stereo, set it up how you like it.
And the datasheet pretty much tells you how to make it.
It'll smoke any LM386 type circuit that you cobble together and the cost is probably the same or less.

[merlinb]

Okay, got the rudimentary circuit down, needs tweaking of course, but what do you think?

Yikes, that ain't going to work! All those opamps and you're trying to drive phones with a grounded emitter transistor?? This is how you clasically drive headphones in a hybrid configuration:


Or you can do it with parallel NE5532 opamps (you really only need two per channel, not three like in this circuit. Reduce the 300R output resistors to 33R or 47R):


You're gong to need 470uF output caps if you want any bass.

Also, four pots just to do crossfeed?? Are you made of money? See here:
http://sound.westhost.com/project21.htm

[boppy100]

+1 on TDA2822M