PARALLEL WIRING A TWIN TRIODE 12AV7

[PIANOLYDIA]

Hi Everyone,
Was wondering if anyone has paralleled wired a 12AV7,12AX7,ETC. Making it a single triode instead of
a twin triode only using half the tube per channel, what would be the benefit, and is it better to just run
half the tube? I have heard of using 1/2 as the gain, and using the other half as the driver to the output tubes.
What components would have to be added and how would you wire that up?

[DeathRex]

JL labs uses both sections of a 6SL7 to drive a EL34. For low current tubes, using both sides gives you double drive the current.

[kheper]

Hi Everyone,
Was wondering if anyone has paralleled wired a 12AV7,12AX7,ETC. Making it a single triode instead of
a twin triode only using half the tube per channel, what would be the benefit, and is it better to just run
half the tube? I have heard of using 1/2 as the gain, and using the other half as the driver to the output tubes.
What components would have to be added and how would you wire that up?

By strapping the grids and the plates of a twin triode together, the transconductance doubles, the (internal) plate resistance is halved (relative to a single triode section), but the mu (gain) remains the same.

http://tubes.nekhbet.com/math.shtml#para

[Geek]

By strapping the grids and the plates of a twin triode together, the transconductance doubles, the (internal) plate resistance is halved (relative to a single triode section), but the mu (gain) remains the same.

http://tubes.nekhbet.com/math.shtml#para

And the noise is halved...

[EWBrown]

Adjust the cathode and plate resistors accordingly (use 1/2 the RA and RK value for a single tube).

If using LED cathode biasing, then use the same as for a single triode section, as the cathode voltage should not change significantly.

/ed B

[PIANOLYDIA]

I have been having trouble figuring out the 12AV7 biasing mA/Voltage and tried using Ohm's law as well as looking at the data load lines/data, and looks like the 22k resistor from the 262 volt 131mA
source into that resistor is not eactly as what is specified, at 22k. I wonder because looking at the
12AV7 Data, seems Vp (Max) is 205.71V and the voltage drop would be about 57.61VDC and by calculating the resistor value I have come up with a 2k Plate resistor and not a 22k. What do you think of that? Looking at the Schematic of the 12AU7, it would be easy If I use that model, because all of the figures are done for you, But I am just not good in math, and I cannot get the values I need! I would like to actually LED bias like two red LED'S in series, but I do not know right off hand what the cathode voltage is going to be. I used Ohm's law and in Parallel, I added the two sides and came up with a Cathode voltage of 12.69Vdc that would be a string of Red LED's probably 6 total, But my big concern here, is I need help, in calculating and implementing this Paralleled wired 12AV7 to be installed into my Amplifier The En Vogue Darling as is in the near post on this forum. So, If anyone can calculate that for me, I would be so Happy and this project will then go forward and we can start building this up. I feel that probably the gm = 3500; Plate current about 13mA at 131VDC Per side of the 12AV7 that will dissipate the tube at 4.7watts of total dissipation. Biasing at around 13mA with a grid voltage of -1.75 volts What have you come up with? Anyone? Am I close here? P.s. I feel I totally messed this one up! Need help....
PianoLydia

[Geek]

You're over thinking it

For a paralleled 12AV7, use a 47K plate resistor and then choose a cathode resistor that gives you a plate voltage of B+/2

Make sure you have a DC grid return to ground - aka hang a 100K to 1Meg resistor from the grid to ground.

Cheers!

[PIANOLYDIA]

Okay, Made some changes to be able to Parallel the 12AV7 using LED Red's at 1.6V both sides
of the Cathodes. Will this do the trick? I have concluded That I will use LED'S OR Battery Biasing
may try both options to see which is going to sound better! Ed, What do you think of this implementation also I changed the Plate B++ Resistor to 47k As per Geeks Recommendations. That
will divide the 47/2 = 23.5K to each Plate So; this is good! Any comments? Not sure what the Bias is going to be set at yet...May need to place 2 LED's in series on both sides yielding 3.2V. Each
P.s. There will be 262V to the 12AV7's and 272V to The 1626's
Ed, I need your Calculations here. Thanks so much you guys, Geek, DeathRex, etc....Pianolydia

[EWBrown]

A 47K resistor to paralleled plates would be equivalent to having an individual 94K resistor on each plate. For the cathodes,
just connect them together and use a single LED. No need to use two separate ones.

Plate voltage will be 262VDC - (47K x plate current in mA), ideally you want to have the plate at 131 VDC, and 2.78 mA plate current. It doesn't have to be
exactly that value, anything within 10% should be good. The actual voltage is determined by the cathode-to-grid voltage differential.

/ed B

[PIANOLYDIA]

A 47K resistor to paralleled plates would be equivalent to having an individual 94K resistor on each plate. For the cathodes,
just connect them together and use a single LED. No need to use two separate ones.

Plate voltage will be 262VDC - (47K x plate current in mA), ideally you want to have the plate at 131 VDC, and 2.78 mA plate current. It doesn't have to be
exactly that value, anything within 10% should be good. The actual voltage is determined by the cathode-to-grid voltage differential.

/ed B

Hi Ed,
New Changes, I changed the value to 11K Plate resistor that will be used in paralleling, and the current of the Plate to 2.78mA , and I also Halved the original values of Cathode bias resistor and by-pass cap for biasing. I put the 100K resistor on the hot side of the 11K resistor near the 262V. A straight line is at -3.45V grid so, I believe this is a true Parallel of what I was trying to reach! If it's wrong then possibly the Cathodes cannot be tied together... So I would have two sets of cathode values of 235R and .11uF for each one. any comments? I really don't want to do the SRPP just seems so complicated! If you can draw out on paper as I have said with my amplifier with ALL of the values in place that would really fantastic, but once again, that would be time consuming. Thanks Ed. PianoLydia

[EWBrown]

The 100K LED pull-up resistor should be connected the top of the 22K resistor, right to the 262 VDC B++, having it on the plate end of the resistor will make it do all sorts of weird, unwanted things...
If you need 3 v on the cathode, then two red LEDs in series, would be 3.2 to 3.6V, you may have to tweek the plate resistor value slightly upwards, to 24 or 27K, in order to hit the good part of the curve.

I should load the 12AV7 parameters into Tube-CAD and see what it says about this... Getting late now, and my eyes are bothering me (allergies).

/ed B

[Gingertube]

By strapping the grids and the plates of a twin triode together, the transconductance doubles, the (internal) plate resistance is halved (relative to a single triode section), but the mu (gain) remains the same.

http://tubes.nekhbet.com/math.shtml#para

And the noise is halved...

NO - And the noise is 0.7 (divided by root 2).

Cheers,
Ian

[Geek]

By strapping the grids and the plates of a twin triode together, the transconductance doubles, the (internal) plate resistance is halved (relative to a single triode section), but the mu (gain) remains the same.

http://tubes.nekhbet.com/math.shtml#para

And the noise is halved...

NO - And the noise is 0.7 (divided by root 2).

Cheers,
Ian

Paralelling is -3dB... 0.707 voltage as you say, but 0.5 power.

We were speaking different scales.
http://www.daycounter.com/Calculators/D ... ator.phtml

[EWBrown]

I ran some sims on tube-cad, and found that an SRPP driver stage should be better than paralleled triodes and LED biasing.
The benefits are that the lower RK can be un-bypassed, which will preserve the inherent NFB of the stage, and that the output
driving impedance will be much lower, which is good when driving paralleled 1626s.

Typical SRPP driver circuit, as used in SorenJ's 6AV5GA project, everything to the left side of the 0.47 uF coupling cap:



For 12AV7, at your originally specified VBB+ 262VDC, and 6.36 mA, RA and RK should be 400 ohms, 390 or 430 ohms is good.
Stage voltage gain will be 20.
If you need higher gain, bypass RK with 1-2 uF / 10V (or greater) cap, for operating it with a 2 KHz LF roll-off. 100 uF or greater for full range response.


I also ran SRPP sims on 12AT7, 262V, 2.1 mA, RA & RK will be 680 ohms, stage voltage gain is 28, unbypassed.

For 5965 (one of my favorite SRPP driver tubes) 262V, 2.5 mA, RA & RK = 1K, stage gain will be 22, unbypassed.

My own "standard SRPP" stage, I run at 30 to 325 V B+, and RA & RK =1K, this allows using 12AX7, 12AT7, 12AU7, 5965, 5751, 6CG7, 12BH7,
just about any 9 pin twin triode tube (other than 6DJ8) with the 9A or 9AJ base pinout.
6DJ8 is a good tube, just needs to observe the lower maximum plate voltage of 130VDC, this would be good with a 200VDC supply.

6/13/2013 Update/Addition:

One important thing not shown in the SRPP circuit, is that the filaments should be "biased" (floated) at approximately 1/4 of the stage's B+ supply, in this case it is 310VDC, so approx. 78 V would be correct, this isn't really critical, anything between 75 and 80V is more than "good enough". The most easy way to do this is to make up a 1:4 voltage divider, I'd use a 300K 1W resistor in series with a 100K 1W resistor, connect a small (1 or 2 uF, 200V) cap across the 100K resistor, and take the resultant 78VDC and connect that to the" filament winding CT, or if there is no CT, connect two 100 ohm resistors in series across the filament, and then use the junction between the two resistors as a "virtual center tap. This balances the absolute potential difference between the cathodes and the heater - the delta can be either plus or minus voltage, and it also serves to reduce hum and noise.

The 1626 filaments could also be "floated" to the same voltage, if only a single filament winding is available. noise and other stage instability issues. This filament "biasing" draws virtually zero current, and the voltage divider resistors consume only about a milliamp, or less.

HTH

/ed B