Hi RATs!

OK, I am now happy, for the moment. The coupling cap is ~16nF: 2, 4.7, and 10
nF, paralleled.

I had been under the impression that too big was just right ... this sounds
better than the 0.5uF I started with.

Nothing lasts for long in my brain, nor amp. Ideas and parts come and go. I
like to think I am following the path of best sound, but, who knows?

It is fun.

Happy Ears!
Al


Alan J. Marcy
Phoenix, AZ

PWC/mystic/Earhead

TubeGarden wrote:

Hi RATs!

OK, I am now happy, for the moment. The coupling cap is ~16nF: 2, 4.7, and 10
nF, paralleled.

I had been under the impression that too big was just right ... this sounds
better than the 0.5uF I started with.

Nothing lasts for long in my brain, nor amp. Ideas and parts come and go. I
like to think I am following the path of best sound, but, who knows?

It is fun.

OK, we have heard all about cap value changes,
and it seems you say less is more.
But have you tried changing
grid bias resistors instead, or using chokes as grid bias elements?

A 0.016 uF cap used with 1 megohm gives a LF pole at 10 Hz.
Even if the cap was 0.0047, you'd get a pole at 34 Hz.
This would have little effect on most music.

But if the R was 100k, the poles would be 100 Hz and 340 Hz respectively.
Meanwhile a choke might give "interesting" results,
because it might be say 200H, and with a 0.1 uF cap,
there is a series resonance at 36 Hz.
With 0.016 uF or less, the Q of the resonance gets less;
and I have never tried this form of grid coupling,
so I leave it to others to investigate.

I prefer the sound of systems which have no sharp cut offs
above 10 Hz and below 50 kHz.
If you have 3 tube stages, each with cut offs at 34 Hz, then the system will have
a
cut off at about 68 Hz and it will be an 18 dB/octave roll off,
and maybe this will dry out the music a bit.

If we used 0.47 uF and say 100k grid bias,
then the pole is at 3.4 Hz.
If the amp is encosed with a feedback loop, then the absence of
phase shift with such a LF pole leads to stability at say 20 Hz.
But nevertheless, if several stages with 3.4 Hz poles are used,
and NFB, then expect the amp to oscillate at some LF.
If the coupling caps were all changed to 0.047uF, then the oscillation would just

be moved to a higher F.

When any RC coupled stage begins to draw grid current, the coupling cap
becomes chaged with a DV bias, and this is important with a guitar amp, which
unlike
a hi-fi amp, is driven hard into many dB of over clipping, and the coupling
caps on the output tubes charge up with maybe 50 volts.
Once charged, the caps take time to discharge, and when lower level notes are
played.
the temporary bias voltage in the coupling caps cutds of the tubes,
and the amp seems "all choked up", or paralysed.
Amps with NFB more prone to this effect, because once clipping starts,
the error signal is boosted and the paralysing charge up of coupling caps
is accentuated.

So with guitar amps, they keep NFB to a minimum, and
use coupling caps of low value, which discharge rapidly
after overload, so fast in fact, nobody notices the paralysis, and any
distortion created during the event is all part of the music, since the
amp IS part of the isntrument.

If hi-fi amps, one keeps well away from any grid current,
lest we risk changing the sound of an old italian violin to that of a suzuki
trainer.
But there are reasons why F poles are lower than they really need to be,
and its because of accumulated poles phase shifts in an amp system,
and stability.

But with no NFB, and a 2 stage power amp, one could afford to not be so
strict about LF poles.
At least you may not get any rumble with vinyl!

Patrick Turner.




Happy Ears!
Al

Alan J. Marcy
Phoenix, AZ

PWC/mystic/Earhead

Hi RATs!

Yup, two stage only. NFB is local to each tube via unbassed cathode resistors.

Input stage uses shared AA battery under 663K grid resistors for bias. A shared
8 ohm resistor under the cathodes gets the plate voltage up to 150V and gives a
touch of NFB.

500 ohm stoppers on the 396A grids. Electraprint 3634-2 ITs to the 6F6G grids,
trioded with 5K from screen to plate, ubt-1 output SE iron. Bias is shared 200
ohm resistor under cathodes.

Just keeps getting better

Happy Ears!
Al




Alan J. Marcy
Phoenix, AZ

PWC/mystic/Earhead