My take on the F4 - advice needed

[GregH2]

Hi All,

I've just designed a zero feedback current amplifier somewhat like the firstwatt F4, but a little simpler and with a single rail supply (see attached schematic). Just interested to know if there's any glaring issues or omissions or any ideas for improving the circuit?

I hope to run the amp with a tube preamp I'm also working on.

Also, any thoughts on the suitabililty of the J111 as an audio JFET? I chose it because of its high voltage capability (40V), but it looks like it really is a switching device. I was going to run it at 5mA - does that sound ok?

Any advice on the general design would be appreciated.

Thanks,
Greg.

[Zen Mod]

looks good to me , on quick look

[flg]

C3 can't be a .1uF? 10,000uF maybe?

[GregH2]

Thank you both for your replies.

"C3 can't be a .1uF? 10,000uF maybe?"

Output cap is 0.1 Farads (100,000uF). This is an experiment I thought I would try.

In one of his books Doug Self mentions an experiment he did with output capacitors on single rail amplifiers where he was able to remove essentially all output capacitor induced distortion by using an obscenely large cap (0.1F). Since I am not using feedback and have a bunch of 47000uF caps lying around I thought I would give it a go.

I could switch to a dual rail supply but I like the idea of the circuit as it is for three reasons:

1) I can test it with my cheap 0-30v, 5A lab supply.
2) No DC offset issues and a certain degree of passive speaker protection.
3) As another mentioned, current draw from the power supply needs to go through current sources to get to ground and will therefore be constant, ie Iq = Imax. I'm interested to see the effect of this. Should make life easy for the power supply I think.

So, no negative feedback so far. Looking good then.

[GregH2]

Hi All,

I'm interested to know whether there would be any benefit to having a voltage regulator feeding the input buffer? I have modelled it with 1v of 100hz ripple and the increase in distortion in the input buffer is negligible so it would appear superfluous (modelled distortion goes from 0.04% to 0.042%). The jfet ccs seems to do a good job stabilising things. Would this indeed be the case?

Your advice is appreciated, as always!

Greg.

[Zen Mod]

better with reg ; or just cap multiplier

[GregH2]

OK, working on a regulator design at the moment. In the meantime I'll use a pi filter.

Thanks,

Greg.

[GregH2]

Ok,

I wired up the circuit shown in the first post using J202 jfets on the front end, with the ccs self biased at about 3.5mA. The input has a 120R gate stopper and the 240s have 221R stoppers as shown.

Sounds pretty good so far but I'm yet to make a preamp with any real voltage swing so listening levels are low. Distortion at 10kHz with my cheap 40MHz scope and FFT analysis is no higher than my signal generator....apparently.

First figure shows 20kHz sinewave of input and output overlying each other. I have no idea how much distortion you need to be able to see a difference, but things are looking good so far.

Second figure shows 20kHz square wave output (top) and input (bottom)....nasty.

Third figure shows 240 gate waveform, before stoppers (top)....nasty. Problem therefore in JFET stage I guess.

4th figure shows input to JFET after gate stopper (top).....fairly clean...no gate ringing??

So, anyone have any idea why my square waves are so terrible?

[GregH2]

Hi All,

I ran the FFT module on my USB scope software using a 20kHz sinewave at approximately 9v into a 6 ohm load with a 10000uF output cap. The results are shown below in linear and db scales. I have some questions I'm hoping someone can help with.

1) Is using a FFT a valid way to test distortion? If so, is this cheap chinese software likely to do a reasonable job?

2) What can be gleaned from these figures...what do they mean in terms of THD? Can someone explain how I should interpret these figures?

3) Why the apparent lack of 2nd harmonic character?

Any info would be great. Also still keen to hear some feedback on the problem I mentioned in the above post!

Thanks,

Greg.

[Nelson Pass]

Those 20 KHz square waves are not a problem. From the
looks of it, the jittery may be your test equipment.