F4 Type amp but with gain

[fab]

Having almost quit the F5 topology using lateral mosfet outputs in common source mode (see F5 thread), I came back to common drain mode with Laterals. I noticed the PAss F4 design but the gain less than 1 was not meeting my requirements. So I came up with this idea: F4 buffer output type but with a VAS to get a gain of about 20 to match my other amp gain in a bi-amp system.
I realised afterwards that it was similar to a Buzquito from J-M Plantefeve (same designer as the Profet amp). The main difference is no overall feddback on the output stage and the feedback resistors arrangement like the F5.

I have not built it but only simulated so far.

See attached for comments.

[Nelson Pass]

Very Cool. Let us know how it goes.

[jacco vermeulen]

Would be nice if someone also tried the Fab-Five with Vertical MOSFETs in common source.

[fab]

Originally posted by jacco vermeulen
Would be nice if someone also tried the Fab-Five with Vertical MOSFETs in common source.

Sorry I can not try it with verticals since I do not have the complementary pair. Have in mind that the resistors value shown are for laterals with their low Vgs compared to verticals. Also, I would replace the resistors at base of input cascode transistors with Leds in series or else which produce a more constant voltage under power supply voltage variation at high power demand. The "Fab-Five" with verticals is like the F5 with cascode at input stage (if you use the feedback arrangement of the F5 instead of the Profet)..

[fab]

I have a proto board (breadboard) but not finished tweaking it. Sa far the gain stage is making several volts loss for maximum output swing and maybe a secondary power supply with about 7v or more would be suitable for the driver stage to increase efficiency... At least common drain output stage allows for that.

THD at about 1W/8 ohms is about 0.05% but increases gradually to about 1% at 17Vac with 23Vdc power supply. But the clipping is very nice rounded and I suppose it is the low feedback gain stage and the no overall feedback effects...

Square wave is fast thanks to "current feedback" like design...

And I used a cascode input stage like the "Fab-Five"...

I may compare with the same basic design but taking the feedback at output stage to be more like the Buzquito to compare electrical performances...

[jam]

Nice design Fab, would there be any advantage cascoding Q11 and Q12? I would use lower values of R3 and R4 though.

Jam

[fab]

Originally posted by jam
Nice design Fab, would there be any advantage cascoding Q11 and Q12? I would use lower values of R3 and R4 though.

Jam

Thanks jam

Cascode the VAS, yes I think so.

For the lower value or R3 and R4 it increases the jfet input current but implies reducing the load resistors of the jfet accordingly to stay with same VGS at output mosfet. From the test I have done on my breadboard, lowering R3 and R4 increases THD.

By the way, the THD (1KHz) goes from 0.05% at about 1W to 0.3% at max class A current ( 9W). My first measurement at 1% was done in class AB region...I only have one output mosfet pair biased at 0.6A.

I have tested the design with overall feedback and the THD goes from 0.05% (1W) to 0.06% (max class A - 9W)...

I wonder if I could put a switch to easily try "non-overall" and "overall" feedback with real musical signal for comparison purpose....

[fab]

The article from Nelson here http://www.diyaudio.com/forums/showt...62#post1553062
explains a bit the reasoning behind the so called "Class XD" (tm) from Doug Self for Cambrigbe as disclosed in the AZUR 840 amp documentation.

I experimented yesterday with the concept on my F4-Fab type amp (but it has changed since then, cascode input, cascode VAS, driver stage).

I only have one mosfet pair in my prototype and tested at about 330 ma bias (I have increased my power supply voltage from 24 to 37 Vdc). The test done is for 1Wrms into 8 ohms with NO overall feedback:

I am posting the THD scope traces without and with a current source "displacer" between output and negative power supply as show here: http://www.tubecad.com/2006/12/blog0088.htm

First picture attached is without the displacer:
sorry for the quality but 100Kb max for an attachment...

THD is 0.08% (I know slightly higher than preceeding post but the bias is now reduced and it is a different circuit)

[fab]

With the added current displacer, the THD is reduced by half to 0.04% but probably more important is that now it is mostly second harmonic ...

I used 175 ma CCS so it is about 6.5W more dissipation.

I did not follow the 1/5 of the bias current since my Bias is relatively low (330 ma)...

Another drawback is that to obtain only second harmonic THD, I need to re-adjust values trying to minimize DC offset generated...

[fab]

Has somebody already experimented similar results on this forum?
Am I the only one with Mr Pass?
Should I post on the solid state forum instead?
If someone would share their experience I would appreciate it...
Thanks