Headphone Amp Redux

You could probably do a decent headphone amp with one mosfet.

John

Yeah! Larkin! I've shown you mine, let's see yours ;-)

And it's 8V P-P into 64 Ohms.

Come on, Larkin, show us all!

What a bull-s*it artist... strike that... Larkin is no artist. ...Jim Thompson

You guys seemed to get along so well when critiquing my design; I have many more questionable circuit ideas in my head to present if it would foster engineering harmony!

Arguing is their idea of sports. :)

Feel free to ask. I'll always have an answer you can trust... you may not like it... but you can trust it. ...Jim Thompson

You're having a bad day. I'm surprised you aren't demanding stero and Bluetooth.

John

able!

FT

=A0 =A0 ...Jim Thompson

=A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 | =A0 =A0 et =A0 =A0 =A0|

=A0|

=A0 =A0 =A0 |

On your Bitrex amplifier, the bootstrap (R1,R2 C1) has always bothered me in general. Would a current source be better or worse?

G=B2

uite

able!

FT

You should insure the step response looks like the amp is under control. Here it is a matter of how much overshoot you want, if any. If the step response looks well controlled, you will find gain and phase margins are fine too.

If you want to stay class A, why not use a N-fet follower. I don't recall how much power you wanted, but earbuds can't take much. Say you wanted 100mW. P=3DV^2/R. v=3Dsqrt(5). That about 3vp for 50 ohms, or 60ma into the load. Now you don't want to vary the quiescent current of the output follower too much with signal swing since that will be a source of distortion. [Modulating the driver current will modulate the transconductance and of course the VGS, all which must be cleaned up with feedback. ]Work up your output stage with a N-fet and an ideal current source. Probably start at 100ma for the current source. Geez, that's like TO-92 range. Probably a N-fet in a bit better thermal package would be better, but don't over do it since the bigger the fet, the bigger the capacitance. Something in a TO-220 package. You might be able to get away without a heat sink. A device in a TO-200 would be say 50 deg C/ watt, at .2 watts that's 10 deg C rise. No big deal.

I've never seen a bipolar amp, well at least a serious one, without a zobel. MOS is way more forgiving with real life loads. MOS will isolate the output from the rest of the amp better. Bipolar drivers always looks better on paper until the amp is compensated.

As in life, the debate can get quite heated :-).

Have a look at this rough and ready sketch:

Note that there are no parts values, so you will need to ensure, for example, that there is is enough current through the first complementary pair to give adequate output drive, There's also no compensation shown, so you need to think about this to get the required response. Assume a vcc of +12 - 15Volts

This skeleton circuit is the basis for many transistorised audio amps and as a headphone amplifier, you should easily be able to get 0.05% or less thd at any frequency up to 20Khz and beyond.

When you have it working, post back your results together with the reasoning behind parts value choice :-)...

Regards,

Chris

Bootstraps can sometimes get your swing slightly closer to rail.

Also can be less prone to stability issues than current mirrors.

Cheaper ;-) ...Jim Thompson

JF, JL has a natural subterfuge personality ;-) ...Jim Thompson

stable!

Ah, enough said.

Nat Semi LM6171 in PDIP or LM7332 Dual in SOIC - doesn't seem all that hard and uses less than 100 components.

You could make a nice amp using a lot of small opamps in parallel, LM8261s or some such. Or some SO8 power-pad things.

John

I post original schematics all the time. I'm not surprised that you think it's impossible, and that you can't imagine a circuit on your own.

Google single mosfet audio amplifier

It's been done a hundred times at least. And there are lots of other possibilities.

John

Bitrex expounded in news:DsGdnX9XBZDebP_QnZ2dnUVZ snipped-for-privacy@earthlink.com:

FFT requires some care in its use. One huge factor is how finely stepped your simulation is. To exaggerate, you can't expect much accuracy if your minimum step time is 1 ms.

If you hope to see results in the 100kHz range, then you'd need a very small step time-- which I predict you won't be patient enough to wait for. ;-)

Some extra-curricular reading about the FFT algorithm may shed some insights.

Warren

what is the purpose of Q11, Q13 and associated components?