AB100 Class AB Power Amplifier

This was my build, a really great sounding and very solid amp. Thanks Mr Pass for the design and osscar for the help. !!
Still doing its job on big PSB bookshelve speakers of my son in law.
And thanks for Prasi for his PDF files ..


  • Pass AB 100.jpg
    Pass AB 100.jpg
    449.8 KB · Views: 167
  • Like
Reactions: 5 users
HRDSTL: do you replace the original output Darlingtons by the MOSFETs 'just like that' or do you modify something else in the circuit accordingly? How many pairs do you use at what rail voltage? What is the bias current per MOSFET? What are the specification requirements against the MOSFETs? SOA compatibility is obvious and I also see people arguing for lateral type MOSFETs here and there for low-bias AB purposes (without giving an explanation). Is there anything else? Thanks, Miklos
I’ve been reading about optimal biasing points for bipolar output stages in Class AB…if the goal is just lowest overall THD. I admittedly have not looked at harmonic spectra info.

The consensus amongst some of the cogniescenti is to measure 22mV across the emitter resistor. This is irrespective of the emitter resistor value. As such, values at or near 1 ohm would be biased at around 22mA per BJT. And for values considerably lower, i.e. 0.22 ohms, bias would be 100mA per BJT transistor. The advantage with larger emitter resistors seems to be more stable bias current. The advantage with smaller emitter resistors seems to be lower distortion (but the BJT devices have to be more closely matched).

I’m curious with regards to Papa’s perspective (as well the rest of you reading) on this given that he has tremendous experience especially with his Threshold designs.

This is a stark difference for me compared to vertical MOSFETS where I would just crank the bias as much as my heatsinks could handle.

Is it true that overall THD can INCREASE at higher bias (but only at higher amplitude levels) with BJT complementary output stages such as the AB100? However, if the output transistors are tightly matched, will one be able to do away with the emitter resistors (or use small values), as such bias currents can be cranked much higher?

Thanks again for the enlightenment,

Joined 2011
Paid Member
Lots of people worship Barney Oliver's technical paper in the Hewlett-Packard Technical Journal. Worship it without bothering to read it.

Bob Cordell's book takes a somewhat more nuanced view: Oliver was a good man and Oliver had a good idea ... but ... we now have better equipment and better mathematical models.

Douglas Self's book, I will let you read for yourself. It seems muddled to me. Maybe Self's writing is crystal clear and my reading of it is muddled.

Both of them talk about Barney Oliver, GENUFLECT GENUFLECT, twenty-something-millivolts, yadda yadda; but what they say afterwards is very different.


  • Barney_Oliver_Distortion_In_Complementary_EF.pdf
    959.5 KB · Views: 30
  • Like
Reactions: 1 user
We recently had a long discussion on my article. https://www.diyaudio.com/community/threads/class-ab-biasing-article.393566

Distortion increases with overbias because the voltage drop across the emitter resistors is an error term - it goes from zero to some maximum on every cycle in class AB. More bias = greater error. Only when the bias is made so high that the voltage error becomes a small percentage of the signal amplitude at which cut-off occurs does distortion fall. This is where MOSFET amplifiers operate. Bipolar transistors can operate that way too.
Last edited:
  • Like
Reactions: 1 users