what's a good average bias setting

jaybell

Member
2010-09-16 7:46 pm
I have a cheap, 4000w 4 channel amp that was having severe crossover distortion. I turned up the bias from its setting of .888v to 1.200v and the sound cleaned right up. but now i am wondering if i need to turn it down a bit. It's a Technical Pro lz4001, and there is no service manual anywhere that i can find. so i am wondering what is a typical setting i might try.
 
0.888 to 1.200 Volt across emitter resistor sounds very high.
How much current does it take to get away with crossover distortion?
I would say for bipolar power transistors, anything from 30 to 60 mA.
As we do not know what value your big emitter resistors have
we can not say how much voltage it takes to get 30-60 mA idle current.
 
The optimal bias is a voltage.

This may be true.
But within certain limits of current to bias into Class AB.
So the current level plays a role.

It is not like you can take 10 Ohm and put a very small current.
Or take 0.001 Ohm and put an enormous current across resistor.

Most people advocate a resistance of 0.100 Ohm to 0.470 Ohm
Depending on the specific amplifier this of course can be different.
Above and below the 0.100-0.470 Ohm.
A special case is when we have parallell transistors.
 
The optimal bias is a voltage.

It is not like you can take 10 Ohm and put a very small current.

Hi,

Yes you can if optimally biasing the output stage of a low power amplifier.
(Though here you may prefer AB biasing to aB, poosibly whacking up the
A biasing to be full Class A for normal loads, only going into B for high loads.)
The important parameter is the voltage, not the current. Practicalities
rule out very low emittter resistors, 0.1R being the lowest I've seen.
But consider a tripled parallel output stage with 0.1R per transistor,
Re effective is 0.033R, bias current is triple, bias voltage the same.

For a given topology there is one optimum bias voltage. Details
of the topology determine the current related to this voltage.
Its not true each topology has an optimum current.

rgds, sreten.
 
Last edited:
It's important to do it this way only when the amplifier is very cold. If it's warm, the bias may not be enough when it's cold or on a colder day. This depends on how good the temperature compensation is of course but even great amps may lapse into crossover distortion if the bias is visually set when its too warm.
 
For a given topology there is one optimum bias voltage. Details
of the topology determine the current related to this voltage.
Its not true each topology has an optimum current.
This (one optimum bias voltage) is almost true. The aim is to get the transconductance of one output half to be roughly equal to the emitter resistor Re, so that both together (plus the resistors) give a transconductance of 1/Re at zero signal. Then for large signals, when the active half has very high transconductance Re dominates. So the output gain is related to Re for both conditions. It can't be done perfectly, but NFB will improve things.

For a straight emitter follower (or Darlington) output it turns out that you need about 25mV across Re. This is because the transconductance of a BJT is directly related to the current.

For a complementary pair output it is not so simple. The driver, in effect, boosts the transconductance of the output BJT but the amount of the boost depends on the resistor Rb bypassing the base of the output. This means that the required voltage across Re depends on the ratio of Re to base resistor, but is generally much smaller than 25mV. I wrote up the theory of this in an article in Wireless World ("Better Buffers pt. 3", November 2000). The voltage should be about 1 + 600 Re/Rb mV, so for Re=0.25 and Rb=100 this gives 2.5mV or quiescent current 10mA.
 
I wrote up the theory of this in an article in Wireless World ("Better Buffers pt. 3", November 2000). The voltage should be about 1 + 600 Re/Rb mV, so for Re=0.25 and Rb=100 this gives 2.5mV or quiescent current 10mA.

This goes well with what i found in practice by slowing raising bias until cross over distortion goes.
 
I don't know what Y'all do for these AB's , but as well as seeing that little hump , it can definitely be heard. :D

I just built a 4 OP pair AB and for testing I'll do 13mv across a .27R emitter resistor. I check all the other 8 and they either are 13 or 14mv (slight beta differences). This is 48-53ma , enough for testing and for a (4)0302/(4)0281 pair, just about enough to make the "glitch" disappear.

For full power testing on startup I trim to 16mv (.27R).. @60ma per device , this is where these (above devices) sound nice. As the day heats up and the music cranks , I'll accept no more than than 20mv and am happy if it stabilizes at 18mv after some abuse. So after all this, my average is 18 ... perfect for the multiple pairs of toshiba clones.

On my single pair creations , the clones like a little more , even as the glitches go away at 10-11mv , single pairs sound best at 20-22mv/.27R (75-82ma). This was confirmed long ago on my" super A" JVC with REAL toshiba's ...15mv/.22R - 68mA OEM.
OS
 

buzz1167

Member
2010-06-27 5:24 am
Has anyone considered that since he is not one of you guys that he is talking about the voltage drop between the two transistor bases and not the voltage drop across the emitter resistor?

1.2V on the emitter resistor is like 4-12Amps with a typical emitter resistor. I doubt he has this and hasn't blown something up... Just my opinion.

In the case I am talking about, 1.2V is about right, although I think mine is slightly under biased for safety, or at least the imitation of safety... But at .88 volts i know that the distortion is audible because I've accidentally been there before.

I think I am currently sitting around 1.05V and with my output transistors that gets me around 10ma of idle which I think is at least reasonable if not ideal. Each transistor pair will be a little different due to construction.

Jaybell:
So I would suggest to the author that you try to measure the voltage drop across the emitter resistor (it should be the big one between the transistors and the output) then when you divide that voltage drop by the resistance of the thing you will get a current.

My further suggestion is to aim to make that current more than 5ma and less than 100mA by fiddling with the bias, I have a feeling that 1.2V will give a bit too much but its just a guess, that's quite the power amp so its possible you wouldn't notice an amp of idle :D If you have multiple outputs try to get them all in this range.

My 2Cents,
Hope this helps Jaybell!
 
Last edited:
My further suggestion is to aim to make that current more than 5ma and less than 100mA by fiddling with the bias, I have a feeling that 1.2V will give a bit too much but its just a guess, that's quite the power amp so its possible you wouldn't notice an amp of idle :D If you have multiple outputs try to get them all in this range.

My 2Cents,
Hope this helps Jaybell!

1.2v is nearly perfect for both my .22R amp and my big 4 pair .27R amp (just measured them) .22r read 1.070v ,.27R read 1.187V - both are reading 70ma across their resistors. I turned the first amp up to 100ma (22mv/.22 re) and read 1.190v. This would be totally dependent on the output device being biased as when I had a MJ15003/4 setup (TO-3 cans) , it took less than a 1 volt to get my 70+ma. So the only valid way to determine bias would be at the Resistor.
P and N channel will also require different voltages , on both of mine it is like .597v / .621v or close to it between pairs.
OS
 

buzz1167

Member
2010-06-27 5:24 am
I agree Ostripper, see I use .1ohm Emitter resistors right now (which I'm thinking of increasing), so My voltages are a tad lower than yours.

Since we don't know what hes got, I figured suggesting a lower number is "safer" until he measures it. I wouldn't want him to blow it up accidentally. What if he's got To-3 Cans and 0.1ohm emitters? 1.2 V might be dangerous...

Regards,