[Don[_4_]]

Thanks Patrick for your help. Please, everyone that's interested see
the latest version and comment.

http://mysite.verizon.net/vze11z1rm/index.html

[Patrick Turner]

On Sep 10, 2:33*pm, Don wrote:
Thanks Patrick for your help. Please, everyone that's interested see
the latest version and comment.

http://mysite.verizon.net/vze11z1rm/index.html

That's looking a lot better. The 6SN7 LTP still worries me because
with CCS taken to ground and Ek at say +8Vdc, there's only 8 volts of
headroom for the CCS.
I've just re-wired the first of a pair of Quad-II-Forties which were
almost identical to the original Quad-II amps except that 6SH7 were
used instead of EF86. To improve on the pentode performance I retained
one 6SH7 but its used in triode with Ia = 5mA, and replaced the other
6SH7 with a 6SN7 LTP. This LTP has to make a maximum Vo of about
50Vrms to each KT88 output grid during worst case voltage levels with
a low load. The available B+ is +400V, but I was able to make an R
divider with 150k + 120k + 8k2 + 5k6 to 0V to give two low Vdc rails
of +19Vdc and + 7.5Vdc. Both are bypassed to 0V via 2uf caps. The +19V
provides the bias for both grids, and the +7.5V feeds the base of an
MJE340, with its emitter having a 1k0 to 0V, so emitter current and
collector current stays at a consant 6.8Ma. The Ek = +25V, so there's
plenty of headroom on the CCS, and I have not reduced the total
headroom of the tube which is 375Vdc. The 3.4mA per a/2 6SN7 is
enough. I've put Rg at 120k, bootstraped off the CFB windings, so
effective loading by Rg is about 240k. RLdc to each 1/2 6SN7 = 50k, so
total AC+DC load = aprpox 40k, and Vo max at LTP clipping would be
approx 80Vrms at each anode, quite enough, and better than in the
original Quad pentode circuit. At low levels of 2 watts, -12dB below
max Vo, and without any GNFB or gain/phase shelving or Zobels, open
loop bandwidth is from 10Hz to 100kHz with no HF resonances below
200kHz. Not bad, the Chinese didn't manage to totally ruin Andy Groves
OPT ideas. These amps will have better performance than Andy or the
Great Peter who went before him were able to design. ((( No bean
counters here at my workshop; if there were, they'd be handed a broom
to sweep up, then be slowly given all the very worst tasks to do, and
only allowed to slowly work upwards if they excelled at each step. If
they mentioned money, before they learnt how to make a good amp which
can be well assembled AND easily serviced, they'd be sacked
immediately. I don't like unhelpful ppl. )))

Don't forget the trials of your amp with 5khz square wave and pure C
loads between say 0.05uF and 2 uF at low levels. Maybe you need the
output zobel, maybe you don't, maybe you need the zobels across the
1/2 primaries of the OPT instead, or as well as.

All zobel R&C values need optimising for least ring overshoot and
maximum bandwidth with pure R loads. I aim for 65kHz.

I'll be doing a page for the website on my latest Quad-II-Forty mods.
In the fullness of time of course. The clumsiness of the Chinese and
there lousy tube sockets and their ommision of electolytics needs to
be exposed and explained. These amps cost a huge sum of money new, yet
have terrible defects. Paint job is nice, but not as rugged as the
1950s Quad finish.

Patrick Turner.

[Don[_4_]]

In article 8141c87a-d779-43ab-b974-
,
says...
On Sep 10, 2:33*pm, Don wrote:
Thanks Patrick for your help. Please, everyone that's interested see
the latest version and comment.

http://mysite.verizon.net/vze11z1rm/index.html

That's looking a lot better. The 6SN7 LTP still worries me because
with CCS taken to ground and Ek at say +8Vdc, there's only 8 volts of
headroom for the CCS.
Patrick Turner.


I plan to use this chip, which only needs 1 volt to operate, set to 6
mA. Hell if I know if it will work. My power transformer has a 70 volt
bias tap, sharing the CT with the B+. I don't have the specs for this
tap, but the transformer is (360-70-0-360) at 500mA center tapped.
Would this mean I'd get 20% of the current available for the bias tap?
-Don

National - LM334Z

1uA - 10mA ; 1V - 40V
3-Terminal Adjustable Current Source
Package TO-92
IC, ADJUSTABLE CURRENT SOURCE, 334; Sensor Case Style:TO-92; No. of
Pins:3; Supply Voltage Range:1V to 40V; Operating Temperature Range:0°C
to +70°C; SVHC:No SVHC (15-Dec-2010); Accuracy:12%; Base Number:334; IC
Generic Number:334; IC Temperature Range:Commercial; Input Voltage
Max:40V; Input Voltage Min:1V; Logic Function Number:334; Operating
Temperature Max:70°C; Operating Temperature Min:0°C; Output Current
Max:10mA; Package / Case:TO-92; Sensor / Transducer Type:Current;
Supply Voltage Max:40V; Supply Voltage Min:1V; Termination Type:Through
Hole

[Patrick Turner]

On Sep 12, 1:54*pm, Don wrote:
In article 8141c87a-d779-43ab-b974-
,
says...

On Sep 10, 2:33*pm, Don wrote:
Thanks Patrick for your help. Please, everyone that's interested see
the latest version and comment.

http://mysite.verizon.net/vze11z1rm/index.html

That's looking a lot better. The 6SN7 LTP still worries me because
with CCS taken to ground and Ek at say +8Vdc, there's only 8 volts of
headroom for the CCS.
Patrick Turner.

I plan to use this chip, which only needs 1 volt to operate, set to 6
mA. Hell if I know if it will work. My power transformer has a 70 volt
bias tap, sharing the CT with the B+. I don't have the specs for this
tap, but the transformer is *(360-70-0-360) at 500mA center tapped.
Would this mean I'd get 20% of the current available for the bias tap?
-Don

You are lucky you have that 70V tap because you can make a 1/2 wave
rectifier produce -100Vdc, then RC filter that to about -80V to apply
it to your bias network.
The would be enough ability to give an extra 100V at 6mA for the CCS
feed.

BTW, the voltage at the existing entry point of -95Vdc at your bias
network should be well filtered but not regulated. The absense of
regulation means that if mains voltage rises and falls, so will bias
voltage, so increased Pda with rise of mains isn't as bad as with
regulated bias. The wipers of the pots can have 10uF to 0V to stop
bias adjustments being "jerky" or too sudden; better than having your
33uF caps where they are. Bias pots should be wire wound types, or
good large cermet types unlikely to ever malfunction. I usually use
10k linear wire wounds, rated for 3 Watts.

The CCS chip may indeed give CCS within a 1V headroom, and maybe
that's enough. I use a common variety single bjt, MJE340 because they
are so easy to source, and so simple and good for the purpose, and
300V rated. There isn't enough voltage headroon with a bjt and emitter
resistance to set the Ik.

Say the largest expected signal voltage output from 6SN7 LTP = 50Vrms
at each anode. Then that's 100Vrms Va-a, and with one grid grounded
and voltage gain of roughly 15, then you'd need about 6.6Vrms applied
to the live input grid, and you should see about 3.3Vrms at the
commoned cathodes. 3.3Vrms = about 4.8pkV, and if Ek = 8Vdc, then the
Ek min = 8 - 4.8 = 3.2 V pk, rather small. Maybe you chip will work
just fine, but I prefer what I think is a more rugged device if ever
the 6SN7 decides to go beserko and Ek were to rise to a voltage which
might cause your iddy biddy chip to **** its nappy.


National - LM334Z

1uA - 10mA ; 1V - 40V
3-Terminal Adjustable Current Source
Package TO-92
IC, ADJUSTABLE CURRENT SOURCE, 334; Sensor Case Style:TO-92; No. of
Pins:3; Supply Voltage Range:1V to 40V; Operating Temperature Range:0°C
to +70°C; SVHC:No SVHC (15-Dec-2010); Accuracy:12%; Base Number:334; IC
Generic Number:334; IC Temperature Range:Commercial; Input Voltage
Max:40V; Input Voltage Min:1V; Logic Function Number:334; Operating
Temperature Max:70°C; Operating Temperature Min:0°C; Output Current
Max:10mA; Package / Case:TO-92; Sensor / Transducer Type:Current;
Supply Voltage Max:40V; Supply Voltage Min:1V; Termination Type:Through
Hole

The specs say V range 1V to 40V, not real high. My method does not
need a pot to set Ik; and pots to set Idc are plain dodgy, not my
style at all, and another reason I wouldn't use a chip CCS unless I
was building an SS amp where the chip really would be happier. Max
Iout = 10mA, and you have a tube which can easily conduct much more
than that. My MJE340 has a much higher Ic rating. Another good device
for CCS is mosfet 2SK1924 and I've used these in series to make a CCS
for anode loads, so that even if the EL84 shorts the B+ voltage is
divided 250V + 250V and one device do3es not cop the whole B+.

I'm called Mr Negative about so many things on this news group. Lemme
tellya, I am very positively cautious, and I have seen so much smoke
come from amps over the years and have worn out my knees chasing hum,
noise, and distortion around the paddock all because designers were so
ridiculously optimistic about their favourite pet design where they
never ever figured out how things could possibly malfunction.

Yesterday must have brought sadness to all Americans remembering 9-11.
Had the designers of those twin towers made their designs resistant to
heat, the tragedy would never have been so bad. When I learnt to be a
builder in a former career as a young man, it was drummed into
everyone at my Building Certificate course that concrete encasement of
all steel work in a tall steel building was mandatory, and there must
be plenty of columns, and even if 50% of the supports are removed, the
building stays up, and does not fall as a house of cards when arsoles
attack in aircraft. And just why a city building needs to be 110
stories high escapes me. Not many catch fire, but hey, the ladders
from the fire trucks don't reach very high. Last night I watched the
film made about the rookie fireman who'd joined a branch of the NY
Fire Dept, and how difficult their task was because the lifts all
failed and ppl had to use stairs, and how those above the fires were
trapped, and all because of design shortcomings. Hundreds of firemen
died trying to help ppl get out of the buildings before they
collapsed. Many are dying early from the toxicity of the dust which
filled their lungs, or from the psychological traumas.

When I saw the story about the servicing job which needed to be done
on the Hubble Telescope I was boyed by America at its best. The NASA
guys work out ALL POSSIBLE failure modes, and make sure there's always
an answer. The last repair job on Hubble has given all people of the
world a truly majestic view of Space, and the Wonderment of the
Firmanent.

Patrick Turner.*