invitro-FETs

[Eric Barbour]

>the bias arrangements for some Glow-FETs. In particular 12AX7s and 12AU7s. I
>want to bias them up to about 200 volts to overcome the clouding problem but
>I'm not sure how to calculate the bias resistors. Actually, I have no clue
>at all to be exact. Whilst I like John Simonton's idea of using a starved
>tube scheme, I want a crystal clear sound rather than a distorted one. I
>just want those little non-linearities that valves bring to the table.

Good for you! A pleasure to see somebody realize that starved tubes are
not happy tubes.


  To use 12AX7s at 200 volts, the plate resistor would be in the range of
120k to 220k ohms. I recommend 180k, 2 watts to be conservative.
The cathode bias resistor would be 1k ohm bypassed by a large electrolytic
capacitor, say 100 uF at 15 volts or so. The grid must return to ground
thru a 1-meg resistor or a volume pot.

  12AU7s use the same scheme, except for the plate resistor. Try 47k at 2 watts.

  If you want REALLY clean sound, neither of these types is optimum.



  One way is to use the Svetlana EF86 pentode, connected as a triode. This will
give a VERY linear triode, very clean in sound WITHOUT ANY FEEDBACK!
Distortion with 10vRMS output signal will be on the order of 0.02%. You read
that
right......
The EF86 can use the 12AU7 circuit above. Pinout is hooked up as follows:
pin 1 - screen grid, connect to plate (pin 6)
2 - internal shield - ground it
3 - cathode - to 1k resistor
4,5 - heater - takes 6.3v at 0.2 amps
6 - plate - connect to pin 1 and to 47k resistor
7 - shield - ground it
8 - suppressor grid - connect to cathode (pin 3)
9 - control grid - input signal (connect to 1-meg grounded resistor or pot)

Advantages of EF86:
1) STILL BEING MADE!
2) very low distortion in triode connection, beats 12AU7s by a mile
3) you can drive the screen grid with a CV and CONTROL THE TUBE'S GAIN.
disadvantages:
1) it's a single device--you need 2 tubes to equal a 12AU7.




Another good suggestion is the 5687 dual triode. Although no longer made, it is
very easy to find in surplus. The American military and computer manufacturers
of the 1950s loved it. It has VERY low distortion compared to even the best
12AU7s,
on the order of the trioded EF86 above.

The 5687 can be used with the same circuit as the 12AU7. Note that its pinout is
DIFFERENT FROM THE 12AU7:
pin 1 = plate of triode #1
2 = grid of triode 1
3 = cathode of triode 1
4,5 = heater (12.6 volts at 0.45 amps)
6 = cathode of triode #2
7 = grid of triode 2
8 = heater center tap, allows use with 6.3v power just like pin 9 on 12AU/AX7
9 = plate of triode 2

There are super-premium 5687 types, such as the 6900 and the 7044.


If you want gain more like the 12AX7, try the 5751. Same pinout, only slightly
lower gain (70 vs. 100) than 12AX7, very low distortion.


By the way, when the rating for heater voltage says 12.6 volts, that is NOT
an absolute
requirement! It can be as much as 10% low. (10% HIGH will shorten the life
of the
tube and give slightly more gain, but 10% low will have little effect on the
tube's gain and will extend its life considerably.)

I routinely run 6.3-volt heaters on 5.5 volts--no problem! And easier to get
from
computer-type switching power supplies.


There are many other tubes that are far better than the 12AU7. E-mail me
for more information.



>I would assume that the bias for a 12AX7 would be identical to that for a
>12AU7. Whilst I'm not really up on invitro technology, I'm led to believe
>that the AU7 is just a lower noise, lower gain version of the AX7. So it
>will be those little suckers of which I'll more than likely be using.

That is not really true. "Low noise" is a complex business in tubes. Some
things, like the 417A triode, have VERY low noise, because they have high
transconductance. (A relation that doesn't really apply to transistors or
FETs, which are even MORE complex in noise behavior.)

The 12AU7 (mu=voltage gain=20) is lower in gain than 12AX7 (mu=100).
A trioded EF86 has a mu of 25, while a 5687 has a mu of 17.

  Any well-made 12AU7 or 12AX7 is capable of low noise, with the right circuit
connection. Heater hum is a different problem, easily solved by running the
heater on DC power.

Eric Barbour
Svetlana Electron Devices
Portola Valley CA USA