I'm building a P3a at the moment and would like to have parallel outputs with MJ15003/4 just to be safe if I need to be able to drive 4R loads. Supply voltage will be about +-40v & emitter resistors changed to 0R1 as per P68. My question for the experts is regarding the bias for parallel outputs. With 1 output pair bias is normally set to 100ma, but would this not need to be increased for 2 pairs of outputs?
Thanks in advance
Paul
Thanks in advance
Paul
Itt will work with 50mA/devices. But of course You can try with double. If it sounds better use that way!
Sajti
Sajti
sam9
I don't think so. I'm more familiar with the requirements for another topology but I think the same logic will apply. What you are adjusting, a little indirectly if you use the produdure specified, is the voltage applied to the base of the drivers. Adding parrallel output devices won't change the voltage required or seen. However, if you add too many parrallel output devices the might be a problem with the current on the base (I'm not sure of this), but cranking up the voltage won't be the cure for that.
You could e-mail Rod Elliot and just ask him. He is usually responsive to polite inquiries.
I don't think so. I'm more familiar with the requirements for another topology but I think the same logic will apply. What you are adjusting, a little indirectly if you use the produdure specified, is the voltage applied to the base of the drivers. Adding parrallel output devices won't change the voltage required or seen. However, if you add too many parrallel output devices the might be a problem with the current on the base (I'm not sure of this), but cranking up the voltage won't be the cure for that.
You could e-mail Rod Elliot and just ask him. He is usually responsive to polite inquiries.
Since they are parallelled, set it to 100mA, and all outputs are running at 100mA. With, say, three parallelled stages, there would be 300mA of current total going through these stages. However, the adjustment procedure is the same...set to 100mA.
One another point of view:
If I set the bias to 100mA, that means the amplifier works in class A untill the output current reach 200mA (double the bias). This is necessary to avoid the crossover distortion.
If I doubling the ouput devices I don't need more class A output power. But it POSSIBLE, due the increased dissipation capacity.
If I use more, and more output transistors, and keep the 100mA bias, there will be less problems with the thermal runaway possibility, due the lower dissipation to each of them.
Sajti
If I set the bias to 100mA, that means the amplifier works in class A untill the output current reach 200mA (double the bias). This is necessary to avoid the crossover distortion.
If I doubling the ouput devices I don't need more class A output power. But it POSSIBLE, due the increased dissipation capacity.
If I use more, and more output transistors, and keep the 100mA bias, there will be less problems with the thermal runaway possibility, due the lower dissipation to each of them.
Sajti
Sajti:
Why should it sound better to increase bias from 100mA to 200mA? It's not much of a class A amp anyway.
You'll just move the points of turn-on/off distortion.
Seen in a voltage perspective, the critical signal (to trigger a turn-on/off in the output device) is increased compared to the voltage error generated. This calls for a high bias.
Looking at the speed required to turn on/off during a higher signal (and current
) that follows from the increased bias, it increases. This puts a higher stress on the driver stage and eventually the Vas stage.
In other words, one could argue that the best bias (if not pure class A) would be close to 0 (but slightly above that to avoid off periods at changing temperatures).
Paulr:
The changed number of output devices will not but a greater stress on your Vas and driver stage, if you adjust the total idle current (100mA shared as 2 x 50mA) to remain as in your original schematic.
Increasing the supply (if that was the second issue?) is not the cure for everything. An increased output voltage vill cause a greater current in the load (as will a lower load resistance such as a 4ohm speaker in stead of 8ohm), which in turn requires more from the driver stage, and indirectly the Vas stage. More output devices will not change this issue.
Jens
Why should it sound better to increase bias from 100mA to 200mA? It's not much of a class A amp anyway.
You'll just move the points of turn-on/off distortion.
Seen in a voltage perspective, the critical signal (to trigger a turn-on/off in the output device) is increased compared to the voltage error generated. This calls for a high bias.
Looking at the speed required to turn on/off during a higher signal (and current
) that follows from the increased bias, it increases. This puts a higher stress on the driver stage and eventually the Vas stage.In other words, one could argue that the best bias (if not pure class A) would be close to 0 (but slightly above that to avoid off periods at changing temperatures).
Paulr:
The changed number of output devices will not but a greater stress on your Vas and driver stage, if you adjust the total idle current (100mA shared as 2 x 50mA) to remain as in your original schematic.
Increasing the supply (if that was the second issue?) is not the cure for everything. An increased output voltage vill cause a greater current in the load (as will a lower load resistance such as a 4ohm speaker in stead of 8ohm), which in turn requires more from the driver stage, and indirectly the Vas stage. More output devices will not change this issue.
Jens
I believe in higher bias. The switch on/off distortion is just one thing. Higher bias increase the gm in the output stage, and this results lower output impedance.
Of course high bias can cause the "gm doubling effect". This means that the input impedance of the output stage is higher when both of the output devices are working (in class A mode). The higher input impedance increase the gain of the driver stage (VAS). When one half of the output stage switch off, the gain falls, and results distortion. (This effect allways appears if the output stage is not class A biased)
It' easy to avoid. Just use some loading resitors on the VAS. In my amplifier I use 3stage output emitter followers. This gives very high impedance load for the VAS stage, and it's possible to use resistors to fix the gain on this stage. And I use high biased class A driver stages.
There are some mesurements on the web, made by Mr. Self. He compared the crossover distortion with some different bias levels. As I saw the transfer characteristics, the lower biased output stage gives very narrow spikes at the crossover level.
Sajti
Of course high bias can cause the "gm doubling effect". This means that the input impedance of the output stage is higher when both of the output devices are working (in class A mode). The higher input impedance increase the gain of the driver stage (VAS). When one half of the output stage switch off, the gain falls, and results distortion. (This effect allways appears if the output stage is not class A biased)
It' easy to avoid. Just use some loading resitors on the VAS. In my amplifier I use 3stage output emitter followers. This gives very high impedance load for the VAS stage, and it's possible to use resistors to fix the gain on this stage. And I use high biased class A driver stages.
There are some mesurements on the web, made by Mr. Self. He compared the crossover distortion with some different bias levels. As I saw the transfer characteristics, the lower biased output stage gives very narrow spikes at the crossover level.
Sajti
Paulr,
I have a P3A running at +/-52V, with two pairs of 2SA1216/2SC2922. I run 200mA bias (100mA per transistor). I've experimented with higher bias current around 500mA, but I didn't think the sound got any better, so I lowered it to 200mA again🙂
Try different bias settings.
/Freddie
I have a P3A running at +/-52V, with two pairs of 2SA1216/2SC2922. I run 200mA bias (100mA per transistor). I've experimented with higher bias current around 500mA, but I didn't think the sound got any better, so I lowered it to 200mA again🙂
Try different bias settings.
/Freddie
Sajti,
I agree with you part of the way, but then again... we're talking abouit the same dilemma...
I know quite a bit of Self's work, but this one (if we talk about the same) is looking at this particular stage, without thwe changed loading of the previous stage; be it driver, follower or Vas stage.
The double gm effect will to some extend happen at any "not pure class A" amp., but at varying levels, depending on the bias.
In other words, we both talk about the same problem, and moving it around. The only true answer is class A, but I believe this was not the aim of the amp in question...
Jens
I agree with you part of the way, but then again... we're talking abouit the same dilemma...
I know quite a bit of Self's work, but this one (if we talk about the same) is looking at this particular stage, without thwe changed loading of the previous stage; be it driver, follower or Vas stage.
The double gm effect will to some extend happen at any "not pure class A" amp., but at varying levels, depending on the bias.
In other words, we both talk about the same problem, and moving it around. The only true answer is class A, but I believe this was not the aim of the amp in question...
Jens
Thanks for all the info guys, as you guessed, the idea was not to get more power output, just to avoid running into problems on 4R loads. Rod Elliott advises not to use 4R on 42v rails & 40v isn't that much lower, but he's using a A1302/C3281 pair and it has been said that he is very conservative with some of his ratings.
Cheers
Paul
Cheers
Paul
- Status
- Not open for further replies.