Modified Dual Differential F5

[Melon Head]

I built this a while ago also with cascodes and 40V rails.
I have left the cascodes out of the drawing to keep it simple.
I really like this amp but I am wondering if I can squeeze more performance out of it with a few more mods to the circuit.

What do you guys think?

[Melon Head]

The only thing I am considering at the moment is putting in 2sj28/2sk82 to replace irfp9240/240. I don't have a data sheet for those, so I don't know how they will perform.

Any ideas about further modifying the circuit would be welcomed (I like to experiment).

[jam]

You must have measured a reduction in speed and bandwidth, which some conditions can help but have you compared it to the original. My bets are the original might sound better, but then again I could be wrong.

Jam

[Melon Head]

Yes I have compared it to the original, and I like this version better. However that could be just due to the extra headroom (40V rails and 2.2A bias).

[Zen Mod]

you obviously slipped when Papa sez something about effects of heavy current feedback net , and benefits of that ;

I think he mention some sort of bootstrapping of input Jfets with that .....

I can't see that you still have that virtue in your iteration

everything else in F5 , besides specific feedback , is beaten to death .

but - that eeeny weeeny tiny detail is work of ....... no one else but this sort of Genius :

[Melon Head]

I wonder if parralleling the inputs of the original would be better than dual differential.

I really don't understand enough about electronics (I have never read one text book on the subject).

Every thing I have learnt has come through trial and error based on ideas from Nelson Pass, Erno Borbley, Joe Curl, and everyone here.

[jam]

thanh,

The cascode should help. I have found how that you bias the cascode can affect the sound and I like using a bipolar as the cascode device.
One thing you might try is using current mirrors on the differentials which would lower distortion but I have found might not improve the sound, you will have to try and see.

Jam

P.S. Paralleing the inputs should help as long as the input capacitance does not get too high but in this case the cascode woud help reduce non-linear capacitance.

[Melon Head]

Originally Posted by Zen Mod

you obviously slipped when Papa sez something about effects of heavy current feedback net , and benefits of that ;

I am using the exact same feedback as the original.

[Melon Head]

Originally Posted by jam

thanh,

The cascode should help. I have found how that you bias the cascode can affect the sound and I like using a bipolar as the cascode device.
One thing you might try is using current mirrors on the differentials which would lower distortion but I have found might not improve the sound, you will have to try and see.

Jam

P.S. Paralleing the inputs should help as long as the input capacitance does not get too high but in this case the cascode woud help reduce non-linear capacitance.

When you say biasing cascodes. Do you mean votage dividing?
I have set it so the Jfets see around 22V.

[juma]

Originally Posted by thanh1973

I am using the exact same feedback as the original.

No you don't. In original F5, the feedback goes to Low Z point (Source) and in your simulator-pipe-dream schematic it's going to JFETs Gate (high Z).