I have read Your interventions, don't understand of thing it you is speaking, but me amuses your heat.
You teach me that there are about tens of configurations of the output stage, and diverge devices by use, and that every device has a his exact characteristic of transfer.
D.Self has analysed a part of this (MOSFET and BJT) in any EF and CFP configurations ( Complementary feedback pair ) and has noticed the polarizations with minor crossover THD ( 114mA EF and 11mA @Re=0.22 CFP class B stage).
They exist a lot of techniques sophisticated to improve the polarization, as the " non linear common mode loop", "square law rules" etc.
How to be has to polarize?: simple, with the near current studied for that circuit, from it more from it less...
Ciao
Mauro
You teach me that there are about tens of configurations of the output stage, and diverge devices by use, and that every device has a his exact characteristic of transfer.
D.Self has analysed a part of this (MOSFET and BJT) in any EF and CFP configurations ( Complementary feedback pair ) and has noticed the polarizations with minor crossover THD ( 114mA EF and 11mA @Re=0.22 CFP class B stage).
They exist a lot of techniques sophisticated to improve the polarization, as the " non linear common mode loop", "square law rules" etc.
How to be has to polarize?: simple, with the near current studied for that circuit, from it more from it less...
Ciao
Mauro
My comment regarding more is better in the case of L-MOSFETs was intended to convey a certain wry humor regarding some practices and recomendations. Just because you *can* do something dopesn't mean it is a good idea.
Regardinf BJT's I think Self is valid. However, I don't think the specific optimal bias voltages cited are applicable to all topologies. For instance when dealing with a mirror image VAS, a EF biased at Self's 2.89V ran unreasonably hot and wanted to run away even with a very large heatsink. He alludes to other topologies requiring a different bias in his comments regarding output tripples.
Regardinf BJT's I think Self is valid. However, I don't think the specific optimal bias voltages cited are applicable to all topologies. For instance when dealing with a mirror image VAS, a EF biased at Self's 2.89V ran unreasonably hot and wanted to run away even with a very large heatsink. He alludes to other topologies requiring a different bias in his comments regarding output tripples.
Hi, Sam9.
Be saying the same thing. I have done reference to Self to show that in every circuit is has to analyse the point of work "ideal" ( and he has done it ).
My irony is tied up one's own to the tendency to generalize the concept. Sense of humour to depart, I concern me to scientific and demonstrable concepts when intervene in this forum.
Creed that my affirmation:"-How to be has to polarize?: simple, with the near current studied for that circuit, from it more from it less..." non-being from it extremist from it questionable!
Ciao
Mauro
Be saying the same thing. I have done reference to Self to show that in every circuit is has to analyse the point of work "ideal" ( and he has done it ).
My irony is tied up one's own to the tendency to generalize the concept. Sense of humour to depart, I concern me to scientific and demonstrable concepts when intervene in this forum.
Creed that my affirmation:"-How to be has to polarize?: simple, with the near current studied for that circuit, from it more from it less..." non-being from it extremist from it questionable!
Ciao
Mauro
janneman said:
Don't get me wrong here Jan, i do not mean that you should selekt your bias settings for this, not at all.
I just mean that if you got less difference in the amount off power that needs to be dissipated than there will be less thermical stress, and will also result in less mechanical stress, witch is a good thing for livespam.
Rudy
If you mean avoiding runaway, I agree. If you mean maintaining optimal bias under all conditions, I think EF configurations still offer "opportunities for improvement". Mitigating it, if my experience is valid, is that there is a useful window where sub-optimal bias has little discernable effect. The preliminary announcement by Onsemi of BJTs with and integral diode for thermal tracking is very interesting.
Sorry about yesterday's comments - I had some local stress issues that spilled over.
On the AB bias issue, in 1974 I produced a range of commercial PA amplifiers using the then new MJ15003/4. The output stage of all models used CFT's - complementary feedback triples! Each output BJT was biassed at 20-25mA. They also used twin slope SOA protection and supply fuses. THD was typically 0.005%.
My point is it's really interesting reading of the virtues of CFP's etc 25-30 years later by someone who cannot even acknowledge the importance of output stage/power supply commutation when even his own experiments showed a reduction in THD from 0.04% to 0.006% just by regulating the supply eliminating it - but dwells on the nuances of crossover optimisation. Please.
The other point was that one should not need to assiduously tweak for optimum bias for distortion to achieve performance goals - as variations occur with spreads,ageing,etc.. My experience is that once there is not a discontinuity causing loss of gain in the forward path then the global NFB does it's job of defining the THD performance.
On the AB bias issue, in 1974 I produced a range of commercial PA amplifiers using the then new MJ15003/4. The output stage of all models used CFT's - complementary feedback triples! Each output BJT was biassed at 20-25mA. They also used twin slope SOA protection and supply fuses. THD was typically 0.005%.
My point is it's really interesting reading of the virtues of CFP's etc 25-30 years later by someone who cannot even acknowledge the importance of output stage/power supply commutation when even his own experiments showed a reduction in THD from 0.04% to 0.006% just by regulating the supply eliminating it - but dwells on the nuances of crossover optimisation. Please.
The other point was that one should not need to assiduously tweak for optimum bias for distortion to achieve performance goals - as variations occur with spreads,ageing,etc.. My experience is that once there is not a discontinuity causing loss of gain in the forward path then the global NFB does it's job of defining the THD performance.
"I must remember the more published guff the more cred..."
Gee, my IM analyzer can't read, and my tests were done around 1985 (Self had not published this by then).
The emitter resistors on most of the amps I checked were about 0R47, so the bias currents were more on the order of 20mA~40mA.
I would drive at full power for a few minutes to heat things up, and then re-check the bias to make sure it wasn't overcompensated.
One amp (which will remain nameless) drew about 8A at full power. When I turned the generator down it still drew 8A! The bias seemed to be a little undercompensated on this model
Gee, my IM analyzer can't read, and my tests were done around 1985 (Self had not published this by then).
The emitter resistors on most of the amps I checked were about 0R47, so the bias currents were more on the order of 20mA~40mA.
I would drive at full power for a few minutes to heat things up, and then re-check the bias to make sure it wasn't overcompensated.
One amp (which will remain nameless) drew about 8A at full power. When I turned the generator down it still drew 8A! The bias seemed to be a little undercompensated on this model
sam9 said:
Interesting, yes. But I see no info on the relative temp drift of the diodes versus the B-E junctions. My gut feeling is that a lot of work needs to be done to come up with a circuit that balances the two, to avoid either over- or undercompensation. Integrating the diodes on the power chip is NOT enough by a long shot. Again, Doug Self has studied this in depth and the result was that is it very hard to get it just right.
BTW, I can't find that Self article on the biasing of power stages mentioned before; anybody else has that info?
Jan Didden
janneman said:
I think you may be after information from Douglas Self's Audio Power Amplifier Design Handbook - Second Edition
Topics: (page 133)
Output stage quiescent conditions
An experiment on crossover distortion
Vq as the critical quiesent parameter
Summary:
EF type I - 8 ohm
Re=0.1 Vq=42.6 mV
Re=0.22 Vq=46.2 mV
Re=0.33 Vq=47.6 mV
Re=0.47 Vq=54.8 mV
CFP - 8 ohm
Re=0.1 Vq=3.06 mV
Re=0.22 Vq=4.62 mV
Re=0.33 Vq=5.64 mV
Re=0.47 Vq=7.18 mV
Note: quiescent current doesn't matter. Vq is the vital quantity.
Quote from Randy Slone, "All bipolar transistors have an integral current-dependent emitter impedance referred to as Re prime."
grege said:
Yes! Thanks, Actually, I have the 3rd edition and there it is on page 150. The numbers are the same. And indeed, Iq doesn't matter, its the voltage across the Re. Meaning that if one changes Re one must keep the same voltage across the Re which means the Iq will/must change.
It also means that for lowest non-linearity in class AB, the EF stage dissipates about 10 times as much as the CFP stage, all other things being equal.
Thanks also to Al for the Re prime. I understand the intrinsic Re but had never heard the term Re prime before. One always learns...
Jan Didden
Excuse if insert me in Your discourses of ends cognizant of the physics, but I continue to read you comments on basic EF and CFP.
If and if I said you that an excellent technique of polarization of BJTs and MOSFETs ( output stage ) is a AB " Square law rules ", what do it be applied of fact on a change of CFP?
This is an example of strong polarizations that give advantages and justify the lowering of the efficiency.
Do think?
References:
This technique foresees to polarize the device in an exact zone of the curve of transfer, (what usually gets with current with hundreds mA) to reduce the problem of crossover THD . It iis described by a lot of authors on EW+WW and an interesting applications example the tricks on EW+WW Sept.1995.
If and if I said you that an excellent technique of polarization of BJTs and MOSFETs ( output stage ) is a AB " Square law rules ", what do it be applied of fact on a change of CFP?
This is an example of strong polarizations that give advantages and justify the lowering of the efficiency.
Do think?
References:
This technique foresees to polarize the device in an exact zone of the curve of transfer, (what usually gets with current with hundreds mA) to reduce the problem of crossover THD . It iis described by a lot of authors on EW+WW and an interesting applications example the tricks on EW+WW Sept.1995.
"All mosfets have an integral Voltage-dependent Source impedance referred to as Rs prime." could we say it..........
Hi grege,
So you are very informative person..... thanks for your informaton on Re prime....
Everyone,
The topic circles around low-bias or high bias... the biasing simply effects the elimination of cross-over... the overbiasing yields poor soundstage , as some guys says... but Class-A amps are highly biased into several amperes...i think...
How about nice its as been done on optimising the biasing to a state so that we cannot hear what is referred to cross-over distortion....
regards,
Kanwar
Hi grege,
So you are very informative person..... thanks for your informaton on Re prime....
Everyone,
The topic circles around low-bias or high bias... the biasing simply effects the elimination of cross-over... the overbiasing yields poor soundstage , as some guys says... but Class-A amps are highly biased into several amperes...i think...
Since Vq inherently depends upon Iq therefore its independent effective role cannot justifies the statement.......
How about nice its as been done on optimising the biasing to a state so that we cannot hear what is referred to cross-over distortion....
regards,
Kanwar
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.