Junkbox F5

[GregH2]

Hi All,

I thought I might jump on the push pull bandwagon and am contemplating building an F5 style amplifier. I don't have any suitable jfets, but I thought I might try using small mosfets for the input stage. I have a bunch of ZVN3310As and ZVP3310As lying around and thought I might try them.

While I know the jfets are superior, there are some nice sounding mosfet only amps out there too A negative tempco in all active devices is a plus as well.

Anyway, I have come up with a schematic and modelled it (see attached). What do you think? Some questions:

1) Could this design actually work and work reasonably well?

2) Will I ever get the bias and DC offset stable? Modelling suggests just an extra ohm on the bias resistance will add many volts of DC offset. If not, can you think of an easy way to stabilise it?

3) What do you think about my choice of input stage gain and amount of feedback?

4) Any problems in general?

As usual your advice is appreciated!

Greg.

[Nelson Pass]

You will want pots for R1 and R4, and the choice
of the 9240 will give you extra 2nd harmonic in the
midrange and above.

Where's the junkbox?

[GregH2]

Thanks Nelson,

A question though, why do R1 and R4 need to be pots when I can set the overall bias using the pots on the input? Is it just to equalize the current in each input fet? If so, what is the importance of this? I figured it wouldn't matter if they were slightly different (which if R1 and R4 are the same will be necessary to correctly bias the output fets).

Many thanks for your advice!

[Nelson Pass]

I didn't see pots, and so my brain refused to register "POT"

[GregH2]

Completely understandable - the diagram is by no means clear - who reads the words anyway? I put single resistors there because it was simpler to model.

So I take it I can leave R1 and R4 as is?

What are your thoughts on the effect of the unequal currents through the two input fets? Does that matter? They need to be different by about 1mA due to the differences between the p and n channel devices.

I guess the input stage gain is set by the D/S resistor ratios and the devices have pretty different characteristics anyway, should it shouldn't matter much, right?

Thanks again,

Greg.

[Nelson Pass]

If you have a distortion analyzer and want to the lowest
distortion possible, you will be trimming R1 and R4 as well.

[godfrey]

Hi Greg

Awhile ago I was playing with an F5-ish type circuit using BJTs and came up with a different biasing arrangement that might also work with mosfet input devices.

The schematic below shows the idea. Moving the bias voltages from the gates to the drains of the input devices makes things a bit simpler and gets rid of the input coupling caps.

btw: LEDs make nice, low noise voltage references and didn't seem to add much (if any) distortion, at least in sims of the BJT version.

Please don't take the values of R3, R4, R5 and R6 seriously. I'm sure they'll need to be changed.

As for adjustments, I suppose the minimum would be to make R9 and R11 adjustable, and add a trimpot between the sources of the input devices. Only problem is if that trimmer goes open circuit, that might blow your output devices. Probably better to have separate adjustments for the two input devices anyway.

Cheers - Godfrey

[GregH2]

I like this a lot. Pure genius. I was trying to think of a way to make it direct coupled and this works nicely.

Going to go home and model it tonight.

So many designs, so little time to build them all...

[GregH2]

Hmm,

Despite my best efforts I have been unable to get this circuit to sim as well as mine (in terms of THD), though I have no idea why. I've tried tweaking all the resistor values to no avail. Will keep fiddling and update in a few days.

Cheers,
Greg.

[eyoung]

Maybe use a TL431 as a bias generator ala burning amp or F4???

regards, Elwood