Bias Rules of Thumb??

[Zero Cool]

I remember reading an article Nelson pass wrote where he made some statments about bias, ill be danged if i can find it. granted he was talking about class A etc but there were some percentages given etc.

So, Im working on a amp module and i have no idea what the factory speced the bias at. and, i am only running it with 2 pairs of output devices instead of 3 pairs. BUT, i intend to run it on less voltage they they did as well, for a bit less power... All that aside.

Are there any general rules of thumb about Setting Class A/B Bias??

IE, 200 watt Amp, 4 output pairs set at X, or Set Xma for each output pair, etc etc...


It seems, with 2 pairs of output devices, and +/-70VDC rails, my bias range on this module is 30-60ma which seems very low to me??? i was expecting more like 200-400ma


Point me in a direction here...
Zc

[Stocker]

Rule of thumb:

Turn up the bias until it no longer sounds poor. If the amp is poor already, this will possibly result in an explosion, so wear goggles.

Scientific way:

Play into a load with an Oscilloscope on the output and adjust bias for lowest crossover distortion.

[Centauri]

Normally I would think around 20 to 25mA per transistor pair, ie. around 50mA for 2 pairs (4 transistors) etc.

Cheers
Graeme

[sam9]

Depends to some degree on output stage topology. Transistor CFB stages appear to need a lower bias, around 5mA while 47mA is suppossed to be optimal for EF designs.

MOSFET' seem to be a different story all together. Apearently you can just keep increasing the bias and move closer and closer to Class A provided the heatsink is big enough. For more ordinary Class A or AB operation you will find people who swear by everything from 10mA to 100mA. I think you are trading diminishing lower distortion for higher operating temperature.

[Zero Cool]

Originally posted by sam9
Depends to some degree on output stage topology. Transistor CFB stages appear to need a lower bias, around 5mA while 47mA is suppossed to be optimal for EF designs.

Ahh thank you!

This is a traditional Emmiter Follower design using bipolar transistors.

Currently it has 2 pairs of the D2155's, and its complement. but later versions used C5200/A1493's and even later models used still another variation....

In the final variation i will be using 4 pairs of On Semi MJL21194/3 outputs most likley...

So if i understand you correctly, should the amp use 47ma per pair, or total? I can get it up to 47 but not double that currently.

The amp module was designed as a bass amp, so it probably runs at a lower BIAS anyway to stay cool. They didnt even mount the BIAS transistor on the heatsink! (some versions used a thermistor on the sink).

[sam9]

per pair.

[Steven]

It depends on the value of the emitter resistors of the output emitter followers.
For optimum class AB performance the voltage across each emitter resistor should be 25..35mV. Often this results in a higher bias current than you had expected so make sure to have adequate heatsinking to avoid thermal runaway.
The reason why this is optimum has been described on the forum a number of times.

Steven

[Nelson Pass]

Originally posted by sam9
Depends to some degree on output stage topology. Transistor CFB stages appear to need a lower bias, around 5mA while 47mA is suppossed to be optimal for EF designs.

MOSFET' seem to be a different story all together. Apearently you can just keep increasing the bias and move closer and closer to Class A provided the heatsink is big enough. For more ordinary Class A or AB operation you will find people who swear by everything from 10mA to 100mA. I think you are trading diminishing lower distortion for higher operating temperature.

Pretty much my thinking. Interestingly, it's been my experience
that while you can make a bipolar measure best at a particular
bias figure like 50 - 100 mA, people tend to prefer the sound
when biased higher. This was true of the AB "Stasis" amps
from Threshold, and was also true of a number of other AB
bipolar designs I've done for people over the years. It's a bit
of a mystery.

Mosfets on the other hand, the more bias the merrier.

[ingrast]

Originally posted by Nelson Pass

......
Mosfets on the other hand, the more bias the merrier.

Gate capacitance (big). Unless ensuring not going below Vth, will have crossover, vansihingly small with more bias but still there. (The drive has to kick strong to charge it for a brief moment before device responds).

Rodolfo

[anatech]

Ahhh, but Stassis has no global feedback. I own one and enjoy it. It's wrapped up in a Nakamichi TA-2A in my bedroom.
-Chris