I have been messing around with a few designs on the simulator, and I have a simple design using a class A p/p output stage I'd like to build. Is it a prerequisite to use global negative feedback with this type of output stage, or is it acceptable to run it open-loop? What are your opinions on this? The front end is basically a single stage design with an input buffer and a voltage amplifier, btw.
If you use a normal push/pull output without some kind of errorcorrection,
you need the global negative feedback to keep outputresistance low,
hence dampingfactor high.
Also, without a DC-servo you need this feedback to keep DC-offset low.
Feedback is not only for keeping distortions low.
But on the other hand, to get feedback working, the openloop gain
must be much higher than the intended overallgain.
Mike
you need the global negative feedback to keep outputresistance low,
hence dampingfactor high.
Also, without a DC-servo you need this feedback to keep DC-offset low.
Feedback is not only for keeping distortions low.
But on the other hand, to get feedback working, the openloop gain
must be much higher than the intended overallgain.
Mike
A mosfet output has a much higher output resistance than a bipolar transistor output has.
Because you use class A the variation in output resistance as a function of the output current wil be less than with class (A)B. And class A has no/less crossover and switching distortion than class (A)B, which must be reduced by feedback.
I think a mosfet class A p/p amp is possible without feedback but I rather use bipolar transistors.
Marc.
Because you use class A the variation in output resistance as a function of the output current wil be less than with class (A)B. And class A has no/less crossover and switching distortion than class (A)B, which must be reduced by feedback.
I think a mosfet class A p/p amp is possible without feedback but I rather use bipolar transistors.
Marc.
Not using feedback around the output stage is quite possible with BJT's. Of course no transistor is actually a linear device, and without feedback, the symetrical distortion will be greater, and the "linear" operating region will be cut shorter.
Can you post a schematic(s) of the circuit(s) you have been simulating?
I think the you can do away without global negative feedback using mosfet output if you are willing to accept distortion figures around 0,5 to 1 % THD.
Argo
I think the you can do away without global negative feedback using mosfet output if you are willing to accept distortion figures around 0,5 to 1 % THD.
Argo
My experience with mosfetoutput is, that even with some feedback,
dampingfactor is ugly low. (not any numbers from sim, audible low)
As i use 4ohm vented boxes, this sounds catastrophal for me.
dampingfactor is ugly low. (not any numbers from sim, audible low)
As i use 4ohm vented boxes, this sounds catastrophal for me.
So, a little feedback would probably be a good thing. One reason I don't want to use it is to keep the voltage amp stage simple, with as few gain stages as possible, which makes for low open loop gain. I can apply up to about 10db feedback to this amp. Would that make much of a difference? Thanks for the replies.
What damping factor did you want? If you're running Class A
(which is the only way to do this decently), then you take the
cumulative (1/Siemens + Source resistance) and divide it into
1, getting an output impedance. For example, for 3 parallel
complementary pairs with .1 ohm Source resistance and a
transconductance value of 3 (at about an amp each or so), then
the apparent resistance of each device is .43 ohms and dividing
this by 6 we get .07 ohms, for a damping factor of about 100.
This is without feedback, and is plenty good enough in my book.
(which is the only way to do this decently), then you take the
cumulative (1/Siemens + Source resistance) and divide it into
1, getting an output impedance. For example, for 3 parallel
complementary pairs with .1 ohm Source resistance and a
transconductance value of 3 (at about an amp each or so), then
the apparent resistance of each device is .43 ohms and dividing
this by 6 we get .07 ohms, for a damping factor of about 100.
This is without feedback, and is plenty good enough in my book.
Thanks for the formula! That looks good enought to me too. Is that one reason why more pairs/higher bias sounds better? With 4 ohm speakers, would it be beneficial to add more output pairs / bias above and beyond the class A bias point? I guess this would have to be balanced with the input capacitance of the output stage, and whether or not the front end could drive that many pairs with good frequency response.
Here is my class A output stage with damping factor about 100 and no need of global NFB:
http://www.pha.inecnet.cz/macura/next.htm
http://www.pha.inecnet.cz/macura/next.htm
Thanks for the link... looks good. I would prefer to stay with mosfets. I know that the bipolars are more linear, I just already have the mosfets.
I do have a mosfet version of the circuit, with more or less same parameters. Only efficiency is a bit lower (Vgs voltage of mosfets).
Higher bias always seems to sound better with Mosfets, but
there is such a thing as too many devices in parallel.
- Status
- Not open for further replies.