6j7 driving WE91 type amplfier
bobluis
Hello, I would like to know your comments regarding 6J7 tube as 300B
driver. I have studied the WE91 classic circuit to build my first SE
amplifier.
As a first try I will stay on a pentode driver like the WE amp. I have
some 6J7 samples (GE) and also 6SJ7, which is a later similar valve.
310A tube is very difficult to obtain for me, and the suggested subs are
6J7, 6SJ7 and even 6C6 which was used in the early WE theater amp. Later
will think in other driver possibilities but for now I like the WE
pentode idea as a first try.
Some points:
1- the image below shows GE plate curves. I have drawn 90K DC load line,
which is the typical driver operating point for some WE91 based
amplifiers including SP Joe Roberts adaptation with 370V B+. Understand
that those resistors are for 310A. 6J7 is a bit different regarding max
current, anode voltage, etc.
So I have drawn another line with lower load (around 50K). My concern is
Miller effect on bandwith, so a lower Ra would help but my limit seems
to be in the max current capability and max dissipation available with
6J7 (red dotted line).
Do you think from a purely theoretical point of view that 50K will work
better?
2- Grid bias remains around -3 volt and on AC I still have more than
+/-70 volts output, enough to make 300B work.
6J7 anode resistance in triode mode is 11K aprox. Using this value I
calculated SG resistance around 55K:
rsg= ra(triode) x (Ia+Isg) / Isg
This value is important to design the SG voltage divider used in the
classic circuit. Correct me If I’m wrong. Classic WE91 circuit uses 75K
upper resistor and 30K lower resistor in voltage divider. To calculate
voltage divider (B+ 420V) for desired SG DC voltage (around 100V) I
should considered lower resistor in parallel with SG resistance.
I have discrepancies here. If SG draws about 0,5 mA at 100V, rsg is
about 200K, not 55K // 30K. What is wrong with my reasoning?
Sorry for so lengthy post But I think could be usefull for other members
too.
+-------------------------------------------------------------------+
|Filename: 6J7 load lineBW.jpg |
|Download: http://www.audiobanter.com/attachment.php?attachmentid=196|
+-------------------------------------------------------------------+
--
bobluis
Andre Jute
want:
http://members.multimania.co.uk/fiultra/JUTE%20ON%20AMPS.htm
including three adaptations of the WE91 by Roberts, Morrison and
Reichert, plus another three or four from me. My T44 is closest to
what you're trying to do, but uses triodes. The T39 is an expensive
ultra-quiet version. The KISS Amp articles contain a lot of
information about building 300B amps
http://members.multimania.co.uk/fiultra/The%20KISS%20Amp%20INDEX.htm
Basically, if you're worried about Miller, you want hefty current, at
least 8mA and preferably more, on the plate of the driver to the 300B.
I demonstrated on this forum many years ago that high voltage and high
current together make for a silent stage and amp. The article from my
site you want for Miller is
http://members.multimania.co.uk/fiultra/KISS%20126%20by%20Andre%20Jute.htm
Good luck.
Andre Jute
Visit Jute on Amps at
http://www.audio-talk.co.uk/fiultra/
"wonderfully well written and reasoned information for the tube audio
constructor"
John Broskie TubeCAD & GlassWare
"an unbelievably comprehensive web site containing vital gems of
wisdom"
Stuart Perry Hi-Fi News & Record Review
WardSmall
I have drawn 90K DC load line, which is the typical driver operating
point for some WE91 based amplifiers including SP Joe Roberts adaptation
with 370V B+. Understand that those resistors are for 310A. 6J7 is a bit
different regarding max current, anode voltage, etc.
+-------------------------------------------------------------------+
+-------------------------------------------------------------------+
--
WardSmall
Patrick Turner
To drive a 300B to clipping with its grid bias up to = -90Vdc, you
need up to 63vrms of fairly undistorted drive signal, ie, +/- 90Vpk
swing.
But its good practice to always have a driver stage able to produce
50% more than what is required, ie, 91Vrms or +/- 135Vpk swing.
Let us suppose we have 300B bias Rg = 100k, so then we *should* have
the dcRL = 1/3 of the ac RL, or 33k, because RDH4 recommends 1 : 3
ratio of RLdc to RLac.
This would give a total parallel load of 25k only.
To get a swing of 135Vpk into 25k, we have a 5.4mA swing, and the idle
current of the driver tube would need to be more to avoid cut off and
grid current so say 7mAdc, and to get a clean peak swing of 135Vpk, Ea
would need to be at least +200Vdc.
Using a B+ rail of +400Vdc, and RLdc = 33k, one could get a hot idle
point of Ea = +200Vdc at 6mAdc.
But one can see the 6J7 curves would not accomodate the required idle
current or give the swing without severe grid current.
So what would work with 6J7, if anything? Something should work
because I have worked on 300B amp samples which have indeed used a
6SJ7 which is like a 6J7 but without the top cap grid connection.
I came to work on the 300B amps because the guy who bought the amps
cheap from another guy who didn't like the sound thought maybe
somethig was wrong and maybe I could fix it. It turned out the amps
had very wrong screen voltage and anode voltages and THD of the amps
was over 10% at 3 watts, so no wonder these amps didn't sound well.
But the seller didn't have the knowledge to see his errors when he
wired the amps. He was also the "technical officer" to his audio club
but obviously he mssed seeing his own mistakes.
Anyway, 30 minutes work had the amps running fine, and giving 8W at
less than 2% THD, and less than 0.5% at a watt or two.
So, how to use a 6J7/6SJ7?
We should start with Rg for 300B = 220k, and then cheat a little with
RLdc = 100k for the 6SJ7. Then we might say have set up the 300B with
bias = -70Vdc for 7W
and say +/-80Vpk swing is enough drive.
So RLac = 220k//100k = 68k, wanted Ia swing = 1.2mApk.
This looks like we might accomodate it on your 6J7 curves, but we
would like Iadc at idle = 2mA.
But can we get sufficient Ia with RLdc and with a B+ rail of say
400Vdc max?
100kRLdc gives 2mA at Ea = 200V.
Swing = 90 - 200 - 335 = 245p-p = 120Vrms = 84vrms at 5.1% 2H which is
more than what is probably generated in 300B, so a lot of the THD/IMD
you hear will be created by the pentode driver tube and its second
order products created with the 300B, (not so wonderful).
Gain = 122, so Vin for clipping = 0.4Vrms, not bad.
The 6J7 Ra > 1M, and Miller pole is determined by Miller C of about
80pF and 60k, ( the ac RL//Ra ), so pole might be at 33kHz which is
just OK.
You could try RLdc = 66k, and Rg = 150k giving RLac = 46k, and with Ia
3mAdc, and Ea at the same +200Vdc, get about the same swing with lower
RLac and thus slightly higher HF pole.
With higher Ia, the pentode gm is higher so gain is higher if RLac
stays constant. But having a high Rg to bias any output tube in
undesirable because you risk grid current as the tube ages and a
rising +bias voltage appearing at the OP tube grid which tends to turn
the tube on more, heat it more, which causes more grid current at
idle. So the lower Rg can be made, the better, which conflicts with
the need to keep RLac high and for the ratio of RLdc and Rg to be
favourable to avoid distortions.
I don't like pentode driver tubes at all. Maybe they are OK at input
tubes operating at low levels, or in a PP amp such as Quad-II because
their high even order THD is cancelled in the input pair connection.
The best driver is another triode, and I'd suggest the 45, although an
EL84 strapped as a triode with Ia at 15mA is also fine.
The use of EL84 with CCS RLdc allows a very low Rg for 300B and will
just give enough drive if the rail is +400V.
The EL84 triode Ra < 3k, so Miller cut off is so high you never need
worry about it. THD of the driver triode won't have any odd numbered H
compared to any pentode. A triode driver THD is mostly 2H and will be
less than 2H of the 300B and will be cancelling, with low second order
effects.
But with EL84 driving a 300B, the gain is about 17, so you'd need
3Vrms drive to the EL84.
if you were to use a 6EJ7 in triode you'd get gain of about 40,
needing only 1.3Vrms input.
But the 6J7 you have can be used to make a nice preamp when triode
strapped.
I prefer 3 stage amps with all low µ triodes and then some 12dB global
NFB. This may have you feeling nauseous, but its given most ppl I know
the best possible sound with low Noise and Distortion of < 0.5% at
clipping and the best bass tightness because Rout of the amp is
reduced to <0.5 ohms. The other alternative to an EL84 in triode is a
6SN7 or 6CG7 with both triodes paralleled. With B+ at 400V, you can
just get enough drive with RLdc, or CCS anode supply.
One could always experiment with some other pentode than 6J7. The more
powerful ( higher gm ) octal based 6SH7 and 6AC7 and even 6V6 come to
mind, but maybe you'd find the distortion is horrendous.
All the old WE amps designed for early movie sound and using 300B
outputs gave poor low bass content and restricted HF and are samples
of old fashioned bad sounding junk, and are samples of what not to do
IMHO.
They served their purpose in front of crowds of non-critical listeners
who wouldn't have known hi-fi even if it bit them on their bums. But
hey, they had sound, and that was a lot better than the silent movies
with someone playing a piano, even though with a piano the N&D was <
0.001%, Other artifacts were crowd generated, sometimes by dudes in
the rear seats occasionally hurling Jaffa lollies to hit people
sitting closer to the screen. Then there were the popcorn rustlers.
And in 1935, smoking was still allowed......and ppl drank......and
stank....and adda good niteout, maybe getta root afta.
Just my 2.35 c worth.
Patrick Turner.
>
> --
> bobluis