"Patrick Turner" wrote in message
...
Chris Berry originally asked us all about using a mains tranny.
He has since said in a following post that he wanted to make 700 watts,
not just 50 watts.

Well 700W wound be nice....
Down to specifics though, 3 pairs of kt88's would do me fine.
class AB2, 600V on the annodes that would give about 360- 400W no?
That basically concludes what's going in to the OPT but as to what I'd like
out, that would be a bit more complicated wrt ideal impedances.
Ideally, I'd like 2, 2.66, 4 and 8 ohm windings/combinations... so 3 x 4ohm
taps would do fine - 4 would be better so long as they can handle the power.
The reason being that I have both 4 and 8 ohm speaker cabinets and am likely
to use a second 4 ohm cabinet in future. 1.6 ohms would also be nice to
have - but not essential mind you.



I have asked him for details of his proposed amp, so we could
design a tranny right here on the group should he wish to do so,

Sorry, I didn't get around to seeing that offer. I'll take you up on it
gladly.

for the benefit of all those here interested in building tube amps.
There are many ways to make 700 watts.
One could use a quad of 813 with a 2,500 volt supply,
for mainly class B operation, ie, low% of class A,
with a high voltage swing, low current swing.

Patrick, for some things I want to stay within "convention" for guitar amps.
Since weight's a priority, as is a fairly clean sound, I think I'll take the
ab2 kt88 version as being a good compromise and save some weight with the
transformers.
Working out the figures in terms of weight, using toroidal transformers I
should be able to get the transformers for the amp to weigh in at under 9kg
or 20lbs - making a 20kg amp a distinct possibility.
Just to put things into context a Marshall 50W guitar amp weighs 39lbs - I
should be able to build this under 30lbs at a guess.
That's a jump from 1.2W/lbs to 12W/lbs - a signifficant weight advantage.

Some avenues I've taken a quick look at so far include a few unusual
options.
One quite insane one but probably feasible (safety wise there are doubts but
I think the "killer" would be noise) would be to get the HT without a
transformer at all. Just voltage doublerss would give me 500V round here
(220V) but some kind of soft start using relays to wire the thing up so live
and neutral are the right way round.

Another not so insane but probably too good to be true would be to take a
toroidal core and get someone with laser cutting facilities to cut afine
"crack" into the core itself. so long as the "crack was smaller than the
wire diameter, this wouldn't cause problems winding the transformer
(although a solution to that wouldn't be too difficult either).

I have a good partner in crime here - my father - who has considerable
experience with valves albeit mainly on warships and radio - but enough to
stop any major damage. He's 100% original and verging on vintage (gotta love
parents like that) and I've sent a most of the information you've sent my
way for his processing.

Here's one of his deepest darkest responses:

Just an idea (which, incidentally will add a little weight ...)

Since the problem with fixed bias P-P PA stages with Negative feedback is
what happens when they are overdriven - I forget the link you sent me (put
it in here if you pass this on), how about putting a choke in place of the
grid leak. The problem is the rectified grid current not having a discharge
path from the grid, building up an appreciable charge on the grid, pushing
it into grid current operation - to my mind BAD for business - unless you
sell bottles.

The choke will give a low impedance (instead of 100 to 500k resistor) path
to 0V/ground for the DC generated by the grid current as the Amp passes into
overdrive . A cheap choice would be a single ended OP transformer for (say)
2.5 W, with the secondary removed (quite easy since the Secondary is
normally the outside winding). These TFs have an air gap which means that
they can handle the small DC current without changing the inductance too
much. The effect would be that the bias will not be displaced by any
appreciable amount - DC resistance is about 400 Ohms, so a lot of current
must flow for a single volt to appear. You will nead 2 chokes - one for each
side of the push-pull output sections - no need for one per bottle.

One possible problem might be the impedance (or DC resistance) of the bias
supply if this is a negative voltage sent to the grids (relative to
0V/ground). this probably presents a DC resistance of about 10 to 50k, but a
couple of electrolytics of about 220 to 1000 microFarads should be enough to
stabilise the bias reference points - just allow them to be disconnected
when you are setting up the bias. You want a pretty long time-constant to
allow you to play your bit of overdriven (noise) without upsetting the
situation. In the limit, you could put in a couple of LM317s in a suitable
circuit to provide 30 to 40 V as solid as a rock, but well bypassed against
parasitics with all that inductance around.

If you don't like the idea of old SE OP TFs, you could ask Sowter for a
quote for a couple of 40 Henry chokes at about 250 Ohms (or less) - should
be easy. Normal swinging chokes are quoted at about 40H at 10mA (reducing to
2H at about 150mA), but they are probably too heavy and since we don't need
150mA current carrying capacity, we can get away with thinner copper, which
means a smaller window ... shorter magnetic path ... lighter lump of
metals.

I also like Turner Audio's use of 10 Ohm cathode resistors. This makes
setting the bottle current easier (than 1 Ohm) while limiting the current
drawn before the effect of bias and may save you cooking the OP TF. I would
also mean that an automatic cut-out could be fitted without too much
trouble - a 1N4148 (or a schottky for a higher reliability) from each
cathode to an op-amp set for (say 75mA per bottle through 10 Ohms gives
0.75V. Knock off the 0.6V (0.35V for the Schottkys) you first thought of for
the 1N4148s and you have 0.15 V (0.4V) to set as your theshhold to flip the
op-amp into energizing the relay which cuts off the HT (and locks the relay
on through a high resistance to HT -or the bias supply). There is also a
negligible power loss.

I am not sure whether you want to have separate bias pots for each bottle
if you are going to use 6 of them - one per side, with a common cathode
resistor for each side to measure the current through each triad might well
be easier in the long run. Remember that ultimately, you want equal currents
on each side at equvalent portions of the waveform. It does not matter that
the individual bottles are matched; what you want is a+b+c = c+d+e
(everywhere you can get it). Static biassing only guarantees it at a
quiescent state in any case.

With Turner Audio's 10 ohm resistors, you could also set up an op-amp to
compare a rectified voltage from each side and flash up a "reset bias"
alarm/lamp when things get too far out of kilter. A pot across the bottoms
of the 2 bias pots would enable (with a bi-colour warning LED) you to trim
the bias back into the reasonable area without resorting to measuring
equipment i.e on-stage (fiddle with it while playing, until the little
light goes off).

Well as you can see, I'm taking all this in slowly (probably too slowly for
my tastes) but it's really nice and educational and at the end of it,
there's a possibility of a nice reward. Hats off to this forum though for
being so helpful (ok with one exception) but I'd like to take the oportunity
to say that it's really well appreciated.

Thanks.
cb