class A single ended

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have a project in my mind...
it is a single ended pure class A mosfet amp with DC-feedback- no transformer then...
I do fear about output impedance, not very much sure about ocsilation issue.
Is it a good idea to avoid DC at output using an integrator???
What do you think??
 

Attachments

  • a.zip
    40.1 KB · Views: 388
schematic not zipped
 

Attachments

  • a1.jpg
    a1.jpg
    31.4 KB · Views: 1,653
No, you are not being ignored.

As I post this, 223 people have seen this thread.

I know very little of solid state to speak of so I cannot comment, though I do want to learn.

Perhaps you could ask for some more specific critique of your design to elicit a response?

As for a large electrolytic 4.7mF (4.7milli-farad) capacitor in series with the output, it is used to keep DC from the speaker in some designs where the offset is not likely to be stable, or there is a large DC offset at the output as part of the design e.g. original JLH, ESP DOZ etc. As you made that schematic you probably already knew this. In regards to stability, I have absolutely no idea, apart from the fact that they have, and are used in established designs.
 
darkfenriz said:
simulation results seem stange to me - very low distortion in an amp without global feedback ???

by the way , i saw once an amp project with a huge cap (4700u or someyhing) in serial to speaker- can this be stable???

... much confusion...


I have no experience with singleended classa, how low are your
distortions ? Having no global feedback does not necessarily mean
high distortion. There is a reason why singleended classa is a good thing.

About the huge cap in serial to the speaker, as it is serial it does not
change stability. It does not create a capacitive load, and as it is
very big it's only blocking very low freqs. Don't forget, the crossover
in speakers have "big" cap in series with the tweeter (~4.7uf).

Mike
 
darkfenriz said:
simulation results seem stange to me - very low distortion in an amp without global feedback ???

by the way , i saw once an amp project with a huge cap (4700u or someyhing) in serial to speaker- can this be stable???

... much confusion...
I've built single ended amps with caps that big on the output and have had no problems with stability. They were just single stage monsters with giant 300watt resistors in the collector and 50watt in the emitter. Full class-A room heaters!
 
The one and only
Joined 2001
Paid Member
markp said:

I've built single ended amps with caps that big on the output and have had no problems with stability. They were just single stage monsters with giant 300watt resistors in the collector and 50watt in the emitter. Full class-A room heaters!

That's how we like them.

To address the original question, one of the reasons we bother
with single-ended class A is that it allows such good performance
with such simple circuits. In your case I see three gain stages
where you could conceivably get away with one, and lots of
current source regulation where a simple voltage reference
driving the gate of the Mosfet might do.

:cool:
 
Comments on the amp circuit

First stage is statically biased. This may simulate well, but will change quite a bit with operating conditions.
You buffered the input with the follower stage, but it represents 29K impedance - not exactly high input Z.
Same comments about the bias of the voltage-gain stage. Idealy you will DC-couple the first two stages and work out a DC-servo to maintain the bias, or, as Nelson Pass said, drop one stage.
The output stage is again capacitively coupled, when you could DC-couple it (and change the DC-servo to global).
Q1/M0 circuit is again staticaly biased. I suggest that you check Mr. Pass' Zen (DIY project) and Aleph (commercial amp) schematics for a good starting point. You may not like the (harmonic behavior of the) differential input of the Aleph, but check it out anyway for the simplistic DC-coupling and look at the original Zen amp for a more predictable current source. With the Zen output stage you'll need a DC-servo, so that would be a nice contribution, which many people may like to clone!
Have Fun!!!
 
thanks johnyx for C5 suggestion- you're absolutely right

about VAS

one of the reason for using two complementary voltage gain stages was a belief, that it DOES reduce the supply noise.
as far as i know every change of supply voltage (e.g. caused from mains 50 or 60 Hz) is kind of fed back on the latter stage.
i dont know if i am right but it seemed sensible to me
??

about input impedance

any contra for using 10x bigger resistors and have 290k
???

about output impedance

still no idea if this gonna be acceptable
 
Two complementary voltage gain stages:

I don't see how it would reduce supply noise, but the non-linearities of the first stage MAY cancel out by those of the second stage. For that to happen, you need to match the curves at the operating points and the signal swings for each of the stages.
The fact that the DC bias is not adaptable means that the stages would drift, depending on operating conditions (temperature, supply), so you better look closely at this issue.
I would especially worry about the impact of the fixed bias on power consumption of M0. I'd simulate the M0 circuit in the presence of large AC signals. R17 may cause it to be less than ideal DC current source, just by intuition.

Output Z:

It is larger than a feedback amp and would affect sound with some speaker loads. This is not a high power amp, so in general it will be sensitive to speaker load.
You also have a 0.47 ohm (R7) in series with the load. What's the purpose of it?
 
darkfenriz said:
...one of the reason for using two complementary voltage gain stages was a belief, that it DOES reduce the supply noise.
as far as i know every change of supply voltage (e.g. caused from mains 50 or 60 Hz) is kind of fed back on the latter stage.
i dont know if i am right but it seemed sensible to me
??

That could be the case, but not in the way you have designed it. In your amplifier the input signal has a ground reference (V2). The input transistor is refering to the (+) supply via R5, so you are actually listening to your power supply. That works in a simulator with a perfect 24V supply, but not in practice. In practice all the power supply rubbish is added to the input signal. The output of M2 is refered to the (-) supply, which is OK, since the source of second stage M4 is also refering to the (-) supply via R18. Here you have suply rejection. The output of this stage is refering to the (+) rail again (R2). That would have been OK if the output stage was a common source stage, but it isn't. It is a source follower with reference to ground (via the load). So again here you add supply noise and ripple to your output signal.

You have to make sure that both gate and source are refering to the same node, being ground, (-) supply, or (+) supply, or any other node, otherwise you inject unwanted signals to that MOSFET. The same applies to JFETs and also to BJTs, but then the base and emitter should use the same reference.

For your input stage it means that you should have a good AC coupling from source to ground that is not affected by the power supply noise. R5 and C1 is not good enough. Add another resistor (Rx) in series with R5 and connect an elcap (Cx) from the junction to ground. RxCx creates a power supply filter. Of course the ratio R3 to R13 should be changed to create 'voltage space' for Rx.
To isolate the next stage from the supply line, you could replace R2 with a current source and connect R2 to ground instead of the supply line. The current source provides isolation and R2 sets the gain with the ground as reference, just as you need for the source follower M1.

Steven
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.