Ian Thompson-Bell wrote:

I am designing a tube based microphone pre-amplifier. At present I am
trying to sort out the HT supply topology. I would prefer a simple RC
network but this reduces the anode volts available to the first stage. I
want to avoid 'anode starvation' and it attendant colouration so my
question is below what voltage would the anode be considered to be
'starved'? The preamp uses 6AU6 tubes.

Cheers

ian

There is nothing wrong with CRCRCRC type filters for preamps.

You need forst to abandon completely trying to copy
what was done in 1955 where preamps rarely had more than 47uF caps.
For good filtering, R had to be high, and thus you got a dc voltage drop
you didn't
want.
AND you also got a preamp very likely to oscilate at LF, ie, motorboat,
because it is a de-facto phase shift oscillator.

But these problems dissapear if you use LARGE value capacitors
which are now quite cheap, such as 470uF x 450V rated.
These are used now abuntanly in SMPS where they work hard, but in your
preamp
they'll be working real easy, and be very effective.
Modern electros are routinely very stable, of high tolerance, and will
bypass
up to a HF.

If you have 330 ohms plus 470uF, the attenuation factor is the same as
using 3,300 ohms plus 47 uF, or 33k and 4.7 uF.

At 100Hz, 470uF has Z = 3.4 ohms, and with 330R the attenuation of
rectifier hum
is around 1/100, and 3 such cascaded RC filters will reduce say 0.2Vrms
of ripple
at C1 by 1/1,000,000 to less than 1.0uV. If Idc was 10mA, the V drop is
only 10Vdc across
3 x 330 ohm R.

Let the moths out of your purse and buy some decent sized capacitors.
Forget the BS that they sound bad; the opposite is true!

Consider learning about high voltage regulation
for the first stage of the preamp at least.
It need only be a shunt reg type, and one using zeners in a string,
say 5 x 75V rated at 5 watts each and with 10mA of current flow.
Trouble is they make noise as they work; lots of noise in fact across a
wide band frequency,
and so then you need to learn how to arrange the zeners to work so their
noise is filtered
away with a series R and added shunt C.
Yet they are able to give good enough regulation of DC to stop motor
boating of slow
low F undulations due to mains voltage shifts.
Don't use zeners if you don't need to; ie, if the preamp output signal
has less than 0.4mV peak of rise and fall
of the 0V trace when full gain is used.
So once the circuit is built, find out if it is stable with volume
turned up full,
and after bursts of gross overload signals. If there are no peaks in the
measured sine wave
response dow to 0.1Hz, and you SHOULD have gear able to read such LF,
then its stable.
Preamps with bass boost adjustment in a tone control stage were renowned
for
becoming LF oscillators when bass was boosted because of the increase in
gain
made stability poor.



Suitable arrangements of coupling cap values and / or shelving networks
will stop most excessive LF transfer.


All electros used for tube stage supply rails need to be bypassed by
plastic polyprop or polyester
caps placed as close as possible to the circuit it bypasses.


In a mike preamp, bass isn't favoured like in a phono amp. But you don't
want to have a preamp
with its 0V signal wobbling up and down like the crude amps made in 1960
all used to to do.

Tube preamps for microphones are damned noisy where the mic signal is
below say 5mV.

So for good SNR, the old practice was to use a step up microphone
transformer.
The mic was very low Z, and its noise much lower than a tube.
When the mic signal was transformed up 10 times, the noise was still
lower than a tube.
And the tube at the input should be a triode; 6AU6 is fine, but betst
srtapped in triode,
or used with the screen as the anode, and anode taken to 0V.
See RDH4 about such antiquated details.

Of course if you used a j-fet for the input you'd not need a
transformer,
and following stage after the fet can be almost any triode.

A 2SK369 has at least 1/10 of the noise of a well chosen 1/2 12AX7,
with Id = 5mA, Ed = +8V, supply = +20V and RL = 3.3k, and gain = up to
100 times approx.

A cascode circuit is also very good, see http://www.vacuumstate.com and
Allen Wright's
hybrid stages for phono. They can be used for microphone, without the
RIAA filters.

There are some clues on PSU design at my site,

http://www.turneraudio.com.au

Patrick Turner.