Engineer wrote:

Hi, vacuumlanders on RAT.
I've just about finished my homebrew "Scratch 10" amplifier, so-called
because it's based on the bits I had lying around or could scrounge up
for next to nothing. Power transformer and chassis from John Bartley
(again, many thanks), Hammond "468" OPT from a garage sale, about 50
cents (what a find! This sparked the project.) Tubes from spares bin
(but I had to buy one, the EF86.)

I've posted the schematic on "alt.binaries.schematics.electronics",
plus a photo (under same Subject name.)

Here are some "design notes", well, general comments about what
happened... g

1. I decided to use 2 x 6V6 O/P tubes as I had them, I could then
experiment with some 6K6's and 6F6's also to hand. BTW, OPT is 10,000
ohms CT to 4, 8 & 15 ohms, a Hammond "468", crinkle finish paint. It
is marked "4, 8, 15 ohms" at the terminals but in the 1956 Hammond
catalogue it's shown as "3.2, 6.4, 10 ohms".

2. Phase splitter is the venerable 6SN7 since:
a) I had several of them, and
b) an octal tube base was on the scrap chassis used.

3. The first stage is an EF86. Reasons:
a) lots of gain,
b) low noise and microphony,
c) I wanted to use the "Mullard 5-10" design here as I'd used it
before very successfully, and
d) there was also a B9A tube base on the scrap chassis! g Until
then I had been planning a 6SN7 cascode circuit (maybe try later...)

4. Phase splitter:
I wanted to use the 6SN7 as a long tailed pair but, since it's a low
gain tube, I wanted the highest cathode impedance possible for good
balance. However, I also wanted direct coupling from the EF86 plate to
g1 of the 6SN7 but I was also a bit "short of some B+" headroom" for a
large cathode resistor (no negative bias supply to hand and
disinclined to build one!) I finally used a MJE340 power transistor
constant current circuit in the cathodes, set to 4.2 mA (2.1 mA per
tube.) Seems to work. No lack of PP drive, well balanced and no
clipping up to over full power (in a no NFB test.)

5. I chose a choke input P/S since:
a) I had one, and
b) the power transformer was a bit too high voltage at 425-0-425, so
choke input was a Q&D way to get a well regulated 315 VDC or so B+ for
the 6V6's. This is well within rating since there's 17 volts of
cathode bias. BTW, rectifier is a 5V4 but anything indirectly heated
and big enough will do (forget expensive GZ34's!)

6. Directly coupling the 6SN7 puts the 109 VDC or so EF86 plate
voltage on the 6SN7 g1's giving about 114 VDC on the cathodes at 2 mA,
or so. Recommendations are max 90 VDC heater cathode voltage, didn't
want to violate that. So, I connected the heater 6.3 VAC supply CT to
a divider from B+ to get about +25 VDC (well decoupled.) Then, I used
that as the voltage reference on the constant current power
transistor. I had planned to use a zener diode here but did not have
the right one to hand, so I just tried it "as is" - it works. The B+
droop under continuous maximum power is very small so the 25 volts
does not move much (I may zener it later.)

7. O/P tubes are cathode biased:
a) to drop a few B+ volts,
b) I could not be bothered to build a bias supply and, anyway, I could
not find a 5K balance pot in my spares box! Now I've found one, so I
may do this later to allow a better-balanced O/P stage and push the
6V6's a bit harder.

8. NFB. Amplifier works well with no NFB (on the 'scope that is, not
listened to!) It clips at just under 10 watts RMS continuous (9.6
watts) and has a good square wave response, a small droop on 60 Hz
square wave. I've played with NFB and settled on a 22K f/b resistor
from the 15 ohm tap bypassed by 220 pF.

9. Sounds good. Performance (with NFB as stated):
a) Max power at 1 KHz = 9.6 watts RMS, continuous (just below
clipping) If I up the B+ to 325 (by using 120 VAC supply as opposed
to the rated 115 VAC) I can get 10.4 watts but the heaters are then at
6.57 VAC - not so good.
b) Distortion: can't measure it (no instrument), can't see it on the
scope (below clipping, that is), and, the good news, I can't hear it!
(Cloth ears, you say!)
c) Power bandwidth: from below 20 Hz to 30,000 KHz +/- 1.5 dB at 1.67
watts RMS. Actually, at 20 Hz the output is barely 1 dB down and
around 17 KHz about 2 dB up, but taking it lower, below 16 Hz where it
really rolls off, the waveform distorts badly.
d) Hum and noise about 20 mV across 15 ohms with input shorted.

10. Replacing the 6V6's with 6K6's reduces the maximum power to
around 6 watts RMS, all other figures about the same. I've not tried
the 6F6's yet. I suspect the OPT is the wrong ratio for 6L6's.

So, what do people think? How might the circuit be improved? All
ideas for improvements welcomed.
Disclaimer: don't build the "Scratch 10" as shown until the design is
finally optimized. Or build it and improve on it.
Cheers,
Roger

is there a possibillity that you could post the schematic diagram & picture
on a web page instead of the binaries group. i dont have access to any
binaries groups.

--
"The philosophy behind Free, Open Source Software has parallels in several
cultures which practice ethical cooperaton. Perhaps that's why it threatens
unethical business practices which feed on the unknowing."
-- Bassam A. Hassan. CoFounder of the ILUG,
Vice president of the former IOSO,