Do we really need fancy bias schemes other than a simple voltage divider with two resistors and a bjt mounted on heatsink? Do we even need a capacitor across the voltage divider? The answer from my side for both questions is no, but it depends. Whats yours?
For reference here are three schemes
https://pcbwayfile.s3.us-west-2.amazonaws.com/web/24/04/06/2151267095210.png
https://pcbwayfile.s3.us-west-2.amazonaws.com/web/24/04/11/2340230765654.png
https://pcbwayfile.s3.us-west-2.amazonaws.com/web/24/04/10/0433204760624.png
For reference here are three schemes
https://pcbwayfile.s3.us-west-2.amazonaws.com/web/24/04/06/2151267095210.png
https://pcbwayfile.s3.us-west-2.amazonaws.com/web/24/04/11/2340230765654.png
https://pcbwayfile.s3.us-west-2.amazonaws.com/web/24/04/10/0433204760624.png
Douglas Self made some solid, extensive research work about thermal compensation of biasing in class AB amplifiers:
http://douglas-self.com/ampins/books/apad.htm
Read the chapter thermal compensation and thermal dynamics.
Correct thermal compensation is quite tricky because of it's dynamic nature related to the dynamics of power dissipation within the output devices, which is related to the dynamics of the input signal and which is also related to the thermal capacity of the heatsinks, the thermal conductivity of isolating washers, the thermal capacity of the power transistors housings and the thermal conductivity of the dye isolating substrate. And it depends from wheter it is winter, or summer.
Short answer I: Yes! Adequate thermal AB-class biasing compensation is needed to perform best possible results in terms of overall low distortion figures.
Short answer II: You may not bother about a thermodynamically well performing thermal compensation if you do not bother about best possible distortion figures. Instead, a simple thermal runaway protection will do it's job for you. But then, don't blame solid state for not sounding well. Blame yourself instead.
Short answer III: No! Forget about correct thermal compensated biasing needs and bias your output stages into class A instead. But then, don't complain about the monthly power bill either.
http://douglas-self.com/ampins/books/apad.htm
Read the chapter thermal compensation and thermal dynamics.
Correct thermal compensation is quite tricky because of it's dynamic nature related to the dynamics of power dissipation within the output devices, which is related to the dynamics of the input signal and which is also related to the thermal capacity of the heatsinks, the thermal conductivity of isolating washers, the thermal capacity of the power transistors housings and the thermal conductivity of the dye isolating substrate. And it depends from wheter it is winter, or summer.
Short answer I: Yes! Adequate thermal AB-class biasing compensation is needed to perform best possible results in terms of overall low distortion figures.
Short answer II: You may not bother about a thermodynamically well performing thermal compensation if you do not bother about best possible distortion figures. Instead, a simple thermal runaway protection will do it's job for you. But then, don't blame solid state for not sounding well. Blame yourself instead.
Short answer III: No! Forget about correct thermal compensated biasing needs and bias your output stages into class A instead. But then, don't complain about the monthly power bill either.
Last edited:
Hi
Structures intended to increase the thermal stability of the vas stage can distort the sound, according to my personal listening experience. And cfp driver stages View attachment 1307308
Structures intended to increase the thermal stability of the vas stage can distort the sound, according to my personal listening experience. And cfp driver stages View attachment 1307308
Last edited:
In a very basic amp that is working daily since 2009 i used 4 diodes with two diodes stuck above the case of each darlington transistor, but it s a low power amp with a regulated 32V SMPS, basicaly that s a NE5532 driving two pairs of Sanken darlingtons,
so you can imagine the simplicity, i use it as amp for my PC along with a pair of two way 30W/4R speakers.
so you can imagine the simplicity, i use it as amp for my PC along with a pair of two way 30W/4R speakers.
Addendum: As to my knowledge, there is neither risk of thermal runaway for the VAS, nor does the temperature of the VAS influence on the distortion figures of standard three stages amplifiers. So why then bother for thermal stability of the VAS? And then, despite of this, by which means would you want to thermally stabilize the VAS?
Last edited:
It works the same as the good old peaking current source used on ICs. Choose R3 equal to the differential resistance the rest of the circuit would have had without it and the first-order sensitivity of the voltage between the collector and emitter of Q1 on variations of the current disappears. Whether that's a feature you like to have is up to you; if the current is already quite predictable, there is not much point to it.
Wrong guess of yours. I raise my hand and disagree. Could you please elaborate why R3 might be "totally unnecessary"? What is your understanding of R3 in terms of it's assumed function?
As for my understanding, R3 is indeed not strictly necessary, but it's a cheap, robust and elegant refinement of cicuit functionality. Not strictly necessary means that you also certainly can omit R3. You also can omit a VBE multiplyer altogether and build and run an output stage with two serial diodes as a bias voltage generator if you want. Or take three diodes. Or four diodes. Any of these variants will result in a more or less functional amplifier ...
I would go with a VBE along with a carefully matched R3. R3 matched to IC of the VAS, even in presence of a current source for the VAS. Because no current source is textbook-perfect.
Last edited:
@MarcelvdG, your quite interesting, lets assume a linear transistor is used here, there are even crazier things happening to the output devices
all the transistors are mated on the same heatsink, so should be ok on cold start as well as warm run
???
In Douglas Self's book which I mentionned in a former post there is a Spice simulation showing the effect of a well matched R3 in case of Ic variations. From the shown graph and varying Ic between 5mA and 7mA:
At R3 = 0 Ohm, the delta for the VBE_multiplier Voltage is 40mV
At R3 = 22 Ohm, the delta for the VBE_multiplier Voltage is 2mV
Last edited:
The 40mV number matches my back-of-the-envelope calculation:
Code:
(.02585V)/(.005A)*(4 diode drops)*(.002A delta) = .041V deltaI don't understand what you mean, and not for the first time. Maybe it's a problem of the english language as a foreign language for me and probably for both of us. So it's better for me to leave this babel terrain. Bye. And happy building.
@Daihedz, when we do research and come across some numbers we ask our selves, is this the whole picture, what does it even mean? Once research is published there lies the problem of interpretation couple that with the problem of models having their limitations. When an author publishes a book even if its scientific its interpretation lies in the domain of the arts, it becomes a literary work subject to interpretation. I was summarizing and saying those are fun numbers with no real use for the VBe multiplier