2 or 3 Channel, Single Ended Class "A" Design Thread

[Avro_Arrow]

I like this one even better...
 
Use a dual rail power supply and two channels or
a single ended power supply and three channels.
 
The schematic shows it set up for dual rails/two channels.
One channel is shown, both are identical.
 
Input is at VG1 and Output is at VM1.
Adjust R4 for lowest DC offset at VM1.
 

 

[kvant]

 
Quote:

...R3/R5 set both the gain and the bias. I'm not really sure what R9 is there for...

R9 is there to set the gain = - R3 / R9 as JB197 says. R3 and R5 set the output bias as R3 / R5 times Vgs.
 
Quote:

I like this one even better...

 
Do you realize that input impedance of this circuit is approximately R2, that is some 600 ohm. That looks pretty low to me. The previous circuit had about 5k, which is better but still not great. Why the change to a lower value?

 

[Avro_Arrow]

I think you are right ... I think R2 is supposed to be 5.6K, my bad.
 
Quote:

 

R9 is there to set the gain = - R3 / R9 as JB197 says. R3 and R5 set the output bias as R3 / R5 times Vgs.
 
 
Do you realize that input impedance of this circuit is approximately R2, that is some 600 ohm. That looks pretty low to me. The previous circuit had about 5k, which is better but still not great. Why the change to a lower value?

 

[Avro_Arrow]

OK, here is my last kick at the cat for now...
Maybe someone else can provide some examples
or suggestions.
 
This is Baby Zen #3.
 
This version provides some feedback into the current source
to make it share some of the load too. Increases the range
that the amp stays in class A by about 50%.
I also modified it to have a dual rail supply and no output caps.
This will be the next amp I build.
 

If I'm not totally crazy...R1 now sets the input impedance, R2 and R3 set the gain
and R4 and R5 set the operating point. C1 blocks input DC from upseting the
operating point and C5 keeps DC from going back into your source. The values
of R2,R3,C1 and C5 may still need to be tweaked to find optimum values.
 
The Aleph current source is still covered by U.S. patent # 5,710,522, but
as a matter of policy we do not concern ourselves with DIY efforts. All
others can send checks to:
nelson@passlabs.com

 

[Leny]

Hi, before you go ahead and buy the bits you might need to change the value of R2.
33.2 k feeding the capacitance of the FET is likely to curtail your high frequency response.
I think the internal capacitance is multiplied by 1+Gain in this configuration of amp, so ends up quite high.
 

 

[Avro_Arrow]

Yes, -3dB point is about 18k - 19k. Not too bad for Grado's, but
maybe a bit low for less bright cans.
I'll see if I can lower the impedance a bit and get
it up above 20K. -3dB at 25K is probably enough for most people...
 
Quote:

Hi, before you go ahead and buy the bits you might need to change the value of R2.
33.2 k feeding the capacitance of the FET is likely to curtail your high frequency response.
I think the internal capacitance is multiplied by 1+Gain in this configuration of amp, so ends up quite high.
 

 

[Avro_Arrow]

See updates to the schematic above...
 
Quote:

Hi, before you go ahead and buy the bits you might need to change the value of R2.
33.2 k feeding the capacitance of the FET is likely to curtail your high frequency response.
I think the internal capacitance is multiplied by 1+Gain in this configuration of amp, so ends up quite high.
 

 

[Leny]

I think that should help. All you need now is a low impedance source. Most are these days, so you should be fine.
 
I don't want to patronise, but if running without an output capacitor remember to check the output offset over 30 mins or so before you risk a headphone!
 
I'm going to download some more of the published articles from Pass and FirstWatt and have a read up. Should keep me busy for a few nights. I really should get out more!
 

 

[Avro_Arrow]

I have been using the three channel Szekeres I presented
earlier in this tread for a few weeks now and it keeps within
1mV from cold to hot...
This design might now do as well, I'll have to build it and see.
I could always put a servo on it but I would rather not.
 
I have all the Zen and First Watt articles downloaded...
 
Quote:

I think that should help. All you need now is a low impedance source. Most are these days, so you should be fine.
 
I don't want to patronise, but if running without an output capacitor remember to check the output offset over 30 mins or so before you risk a headphone!
 
I'm going to download some more of the published articles from Pass and FirstWatt and have a read up. Should keep me busy for a few nights. I really should get out more!
 

 

[Avro_Arrow]

No further comments or suggestions?

 

[kvant]

I don't see the point of separating the AC and DC feedback networks as you do in your last schematic if you end up with an input impedance as low as 2k anyway. I would simplify it back to the approach you had there originally. BTW, I find it quite confusing when you change the schematic drawings back in time, the discussion then becomes too much of a moving target for me to follow.

 

[Avro_Arrow]

I am not the original designer of the topology, Nelson Pass is.
I have just adapted the components and values to be more
suitable for headphone use. In his original design, he used
a single ended power supply and an output capacitor. The
outer feedback loop includes the input and output caps to
take their effects into account. Please feel free to muck
around with the values of R1, R2 and R3 to get whatever
gain and input impedance you would like...
The inner loop, R4 and R5 are there to set the operating point.
I used a dual rail power supply to get rid of the output cap.
You could also use th three channel design to achieve the
same effect.
 
Quote:

I don't see the point of separating the AC and DC feedback networks as you do in your last schematic if you end up with an input impedance as low as 2k anyway. I would simplify it back to the approach you had there originally. BTW, I find it quite confusing when you change the schematic drawings back in time, the discussion then becomes too much of a moving target for me to follow.

 

[Avro_Arrow]

OK, now I think I have gone a little over the top...
Here is a three channel version with an input
buffer that raises the input impedance to 47K.
 

 
Comments or suggestions?

 

[Avro_Arrow]

bump

 

[nikongod]

Replace R1, R2 with CCSs.
 
Why not use a more "Supersymmetric" topology if you want an active return path and are taking the very pass approach? It only requires a few more parts and should work a decent chunk better. You would have to reterminate your headphones balanced, but this is a small effort. I have been daydreaming of a headphone-sized F1 for a while (I call it the Head-F1 as a play on the Head-F1 HF2) but can never get off of my lazy bumm to get it built.