JFET input, MOSFET VAS, LATERAL output = Perfect!!

[mikelm]

Andrew,

I'm beginning to suspect that increasing the size of the interstage filter caps, or rather adding another much larger one, would yeald better results that just adding more capacitance to the o/p stage PSU.

If we keep the feedback device current really clean that in turn will improve PSRR and looking at it another way . . . any noise in the i/p device is amplified about 20 x so reducing noise their pays a premium

it may be that your small series chokes will be enough but I would still recommend common mode chokes.

[AndrewT]

I read here that the RC time constants of the passive input filter and the NFB DC blocking filter and the PSU filter have to be in a crictical ratio to work effectively and avoid LF instability.

1.) PSU filter C in Farads * Speaker impedance in ohms >= sqrt(2) * NFB filter C * Lower leg R.
2.) NFB filter R*C >= sqrt(2) * Rin * input DC blocking filter.

Your increasing of the NFB CR is using the second of those "rules".

I have been doing this for quite a few years now and for me it clearly improves LF performance. The really nice bit is once these CR ratios are setup/built in, it is very easy to reduce the passive input CR to hear the effect of early bass roll-off without contravening those rules.

[mikelm]

I should remind you Andrew unless I say otherwise my comments refer to my DC linked version. My PSU cap is 33mF and at present my filter caps are 470uF.

[AndrewT]

which post shows the 470uF filter cap location/s?

[mikelm]

This one:

[AndrewT]

Supply rail filtering is different from amplifier gain filtering.
However, the four uF caps shown will all have an effect on bass performance.

Each must filter the modulation on the supply rails to a low enough value to be inaudible at the output for whatever range of signals you pass through.

I would expect variations in audio performance when any of those caps is made too low in value that the rail modulation leaks through to the output.

[mikelm]

yes, I agree and I will try a much bigger filter cap soon and assess it's subjective effect.

But spice indicates that no practical amount of capacitance can ever give as good a result as a separate supply so that will be the next thing I do.

[mikelm]

After some more research this diagram shows what I will be working towards over the next few weeks.

Both the i/p buffer and the o/p transistor "pairs" give better stability which means the amps should sound clean & smooth with less compensation.

But I am hoping for the biggest improvement from the separate supplies which give enormous gains in PSRR in the 100 - 1000 hz range.

Very soon I will be listening on my new IPL transmission line speakers which I'm just constructing

http://www.iplacoustics.co.uk/S3tl%20Rib%20CP.htm

[keantoken]

Nice! I've been thinking amplifier pulse behavior depends way too much on the input impedance, so I think an input buffer is a really good idea, if you want to fine tune stability, and possibly the sonics as well.

- keantoken

[mikelm]

yeah, as I mentioned already, when I reduced the i/p series R to 20 the sonics improved significantly but an i/p filter was not possible. With the buffer I hope to get the advantages of both. The supply to it will probably be a bit better than shown.