diyAudio (
-   Headphone Systems (
-   -   Five-BJT low power headphone amp (

linuxguru 1st August 2008 05:59 PM

Five-BJT low power headphone amp
2 Attachment(s)
Here's the LTSpice schematic of a minimalistic discrete BJT low-power headphone amp, designed for 2.5 mW output into 32 ohms, using a single 3V power supply (2 x AAs are fine) with a quiescent current of 20 mA. The simulated THD20 is 0.003% at 2.5 mW into 32 ohms.

The output buffer topology is derived from this post by MikeB, based on a Class-A JFET voltage buffer topology from Steven:

The rest of the circuit is a vanilla two-stage amplifier with voltage-series emitter-coupled feedback, with a closed-loop gain of ~4.

linuxguru 1st August 2008 06:00 PM

2 Attachment(s)
And here's the THD20 FFT at 200 mV into 32 ohms (0.625 mW):

Bonsai 1st August 2008 06:22 PM

20mA of Iq does not sound like a recipe for portable nirvana.

You need to get your Iq down to a few hundred uA.

linuxguru 1st August 2008 06:30 PM

It's Class A, with a rated output swing of 400 mV into 32 ohms - so we need to be able to source and sink 12.5 mA. Adding some margin to avoid abrupt clipping as well as to operate the output devices at sweet spots, we get to 18 mA in the output devices and 2 mA elsewhere. I can't see how it's posible to go much lower than that in Class A.

20 mA from a 2.8 Ah alkaline AA gives 140 hours of operation, which is plenty.

lineup 2nd August 2008 02:08 AM

20 mA x 2 = 40 mA

Even for operation with AA ( LR6 ) batteries
this is a bit much. So I agree Bonsai.
On the other hand, if we want Class A operation for 32 Ohm hphones,
I say that 18-20 mA can be a good choice.
Your circuit is interesting and I have saved a copy.
I would make another version. With regulated 12V supply.

I have made 2 such discrete 12 VDC headphones amplifiers for 32 Ohm.And I have published my schematics here at This way you know that I have done some research and study of myself.

Good amplifier, you have. A bit too high current consumption.
Using rechargable LR6 (AA) would still not be too unconvinient to try.

linuxguru 3rd August 2008 04:58 AM

2 Attachment(s)
> Using rechargable LR6 (AA) would still not be too unconvinient to try.

OK, I made a few biasing changes to allow operation with a 2.4 V rail, and reduced the output stage current as well. The tradeoff is that the output swing is now reduced to 250 mV only, limiting the output power to 1 mW into 32 ohms. (Headphones of sensitivity > 100 dB SPL / mW should be fine, but certain large can-type phones may not be sensitive enough for 1 mW drive.)

It will work with supplies of 2.4V to 6V and higher, and the THD20/sonics improve with higher rails. At 3V, the total quiescent current now simulates at ~14.5 mA, an improvement over the previous circuit. With 2 x 1.2V LR6/AA rechargeables, the quiescent current simulates at ~12.5 mA.

THD20 is more or less the same as the previous circuit, but sonics are slightly worse at full swing and 2.4 V rails. The closed-loop gain has been increased slightly to ~5.

Edit: R11 models the internal resistance of the battery for simulation. An optional small discrete resistance up to 10R can be placed in series with the battery for inrush current limiting, which will help to prolong the life of alkaline AAs. This can be omitted for NiMH AAs.

Elvee 3rd August 2008 10:17 AM

2 Attachment(s)
Looks pretty optimal:
According to an old engineer's saying: "it's always possible to get the same results with half the parts or half the power", but in this case, I didn't quite succeed.
I did halve the parts and the power, but the distortion remains a dismal 0.017%.
Here is the circuit, but don't take it too seriously: it is just a joke, the product of intellectual masturbation.....

linuxguru 3rd August 2008 02:56 PM

That's indeed a clever minimalistic circuit. I can't seem to go much below 0.05% THD20 at full swings - but that's good enough for a minimal design that runs off a single AA.

I'm trying to optimize it (without much success) so that it works reasonably well with a variety of BJTs. In the sim, it looks like only the bc337 and 2sd669 work reasonably well as the lower (active load) device. Sonics are very sensitive to the resistor in the emitter legs.

Elvee 3rd August 2008 04:20 PM

Unlike yours, this circuit is not serious: a proper design must be immune to normal variations in the Hfe of transistors, etc, and battery operation must be possible from 0.9V to 1.6V per cell, which is certainly not the case here. It is mainly intended to check how far the limits can be pushed.
And if it can stimulate the creativity of someone else, so much the better.

linuxguru 11th August 2008 08:21 AM

2 Attachment(s)
OK, I modified your 3-BJT circuit into a 4-BJT circuit by making the input stage a CFP pair, and slightly changing the biasing, gain, etc. The resulting circuit is shown below:

All times are GMT. The time now is 08:44 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright 1999-2016 diyAudio

Content Relevant URLs by vBSEO 3.3.2