F5 power amplifier

[Blues]

http://6moons.com/audioreviews/firstwatt7/f5_2.html

[fab]

Originally posted by jackinnj


I have been able to run the lateral version at ~20W, but can't get the distortion below 0.3% -- it is pretty much "untweaked", however.....

What power supply voltage and how many output pairs?
I got about same THD even after tried to tweak it but I had only one output pair...but the max voltage swing was limited under 8 ohms load

[Nelson Pass]

One thing you can do is to reduce the gain of the amp. 15 dB of
gain is plenty, and compared to 20 dB, that will reduce your
distortion by about half.

[Skorpio]

I have found at suiteable heatsink for each channel and the distance between each output MOS-FET will be 20cm. This gives some questions about the physical layout:

1. Do I mount each JFET close to the MOSFET and run wires to the other half?

2. Do I mount the input JFETs close to each other and run wires to the MOSFETs?

a) Do I let each half have its own feedback loop before running the output to output terminal, where the two half meet?

b) Do I connect the drains of the output MOSFET and run this point to the output terminal. And from here take feedback back to input?

My own idea is to use 1b, but is that the best way?
The principle question is where to have short wires and where to run longer...

[Goffe]

This is my second design, if you find any error please advise me or to make the design better. I will release etch layout later if the design is ok. Size is 7,5x5cm
I don't think the tracks can handle 10A current...etc.

Thanks Nelson
Regards

[Goffe]

...now you got 3D

[jackinnj]

Originally posted by jacco vermeulen
Carol-Ex,

care to give a recap ? (titter revisited)

The gate resistors value was ?
And the Ap-plot is for both N- and P-channel Fairchilds, right ?

I got tired of Carol, but am thinking to use one of the pictures from Janneman's visit to the lovely women of SEAS. The current av is a solar flare.

The plot which I published depicted the P-channel replaced. I then replaced the N-channel as well and the result is the same -- still peaks at around 500kHz (but I can't do a screenshot with my HP3577.)

I am using 47R gatestoppers.

This is the product of my other labors:

[jameshillj]

Hi Goffe,
Perhaps you could seperate R9 / R10 and move the Q1 and Q2 together (with a "clip")
It wouldn't be difficult to seperate R9 & R10 to fit a bipass cap from V+ to V-, as I have often found this to be beneficial to the sound.
I'd perhaps move the V+ (& V-) terminals(?) to the edge of the pcb, ie, between TH!- Q3 (TH2- Q4) where the round pads are now, and reroute the tracks from R13,21,19 (14, 22, 20) under R11 (12).
Suggest a bit more room around the IN & GND pads, and possibly add a second centre pad for Q5,6 to aid different "pin-outs".
As is often done on NP's designs, adding Cmultipliers to the power rails, near to the main o/p Fets - good results, generally.

[P. Lambda]

I would suggest that people doing F5 pcbs look at Nelson's layout in the pictures of the most recent 6moons article. The location of the "current" feedback resistors are close to the jfet input stage. I believe he is minimizing the trace length on the voltage sensing side of the resistors which are more sensitive to noise than the output stage end of the resistors.

[Goffe]

Originally posted by jameshillj
Hi Goffe,
Perhaps you could seperate R9 / R10 and move the Q1 and Q2 together (with a "clip")
It wouldn't be difficult to seperate R9 & R10 to fit a bipass cap from V+ to V-, as I have often found this to be beneficial to the sound.
I'd perhaps move the V+ (& V-) terminals(?) to the edge of the pcb, ie, between TH!- Q3 (TH2- Q4) where the round pads are now, and reroute the tracks from R13,21,19 (14, 22, 20) under R11 (12).
Suggest a bit more room around the IN & GND pads, and possibly add a second centre pad for Q5,6 to aid different "pin-outs".
As is often done on NP's designs, adding Cmultipliers to the power rails, near to the main o/p Fets - good results, generally.

I still learning Thanks
I need to see the 6moons article and get back to work