Graham North wrote...

>Signals being measured will be in the frequency range 12.5Hz to 20 kHz.
>Could you recommend a possible switching frequency? Or even better say
>how you could work it out? (I'm not scared of learning - honest!)

If it were my task, I'd be looking for a suitable switcher IC
from Maxim, Linear Technology or National Semiconductor; one of
their more recent offerings running at a hundred KHz or more,
with a mind to operating it directly off the 9V battery supply.

My "engineering gut feel" is that this design problem demands an
IC solution to get the speed and efficiency; "rolling your own"
is not going to cut it because, particularly for achieving high
operating frequency, it's simply too difficult to match the
performance of modern switching-regulator chips.

I think in the end you're just going to have to start looking
through the various offerings, culling out the obvious
non-starters and weighing the advantages and disadvantages of the
remainder.  It won't be easy, and there's no avoiding it I'm
afraid.

>My analogue guy says that 10mV pk to pk is a good starting point, but
>obviously I'd like to get less than that!

I think 10 mV peak-to-peak ripple is a good starting point.

>He informs me that he has
>filtered the power supplies to the circuitry where possible (or should
>that be necessary?).

Well, I've no way of knowing that without knowing the details of
what it is you are powering; but I would guess that, yes, local
filtering of the supplies would be absolutely necessary.

>Oh, and now I don't need the +5V, he is going to use the +3.3V
>instead.

That's a relief, I'm sure; one less voltage to generate.

It's also added motivation to power the switcher directly from
the 9V battery supply: otherwise, I suspect you'd find your
greatest challenge to be getting a switcher IC that will do an
adequate job running off the 3.3V supply.  I would NOT want to
take that on, myself.

>So thats: +3.3V, +1.8V, +2.7V and + / - 18V from six AA batteries
>in series (~9V).

Incidentally, (and someone else has already raised this issue in
this thread) why +/- 18 volts?

I ask because it's been a long, LONG time since I've had to use
such high supply voltages on analog stuff, ever since the advent
of low-voltage rail-to-rail I/O opamps.  Most of my designs
operate from +/- 5.0V anymore, with a few going as high as +/-
8V.

Life is so much more pleasant at lower voltage...

Cheers,

Dave

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.