Your Apache can go in AB1 mode, which simplifies tube matching.
Set up the Apache for linear operation, no drive.
Have a DVM attached to the wiper of the bias control.

With one 6146 installed (remove the second tube completely,
don't just lift the plate cap), record the voltage reading of the DVM
required for 20 mA resting plate current with HV on, PTT on.
(If you don't have a Heath SB-10 adapter, you may have to jumper
a connector on the back of the Apache so you have operate bias,
and not only cutoff bias.)

Repeat for the second 6146. The DVM readings should be within
a few volts for tube balance. By varying the bias around
the 20 mA point, you can judge how much the plate current will
vary with grid voltage.

I did this with about thirty 6146's, using a HT-32B as the test bed
(with plate current meter inserted). Even new tubes with the same
date code sometimes varied significantly. The average bias
needed in my HT-32B was -55V for 20 mA plate current per tube.

Note: this matching was done after go/no-go testing with
a TV-7 tube tester.

A few tubes, which tested good in the TV-7, required greater
negative voltage than the HT-32B has available in order
to operate at 20 mA. (E.g., greater than -65V bias,
and the tube still drew 28 mA).

I concur with the warning against mixing 6146A and 6146B.
No problem mixing plain 6146 and 6146A.

73,
Ed K4PF


Space Charge wrote:
Need some informed opinion, as I've heard various things about finals
that use two 6146's, like my TX-1 Apache:

1. Make sure the transconductance on both is very close.

2. Make sure the plate current on both is very close.

3. Make sure BOTH are close, especially the plates, which should be
within 2ma, and _both_ should be either 6146, 6146A, or 6146B. Don't
mix types.

4. None of the above matter, so long as both test well on a good
checker. (I use a Hickock/'Hiccup' 539B.)

Can someone shed some light on this? What's correct?

Thanks to the group in advance,

Lin/KJ6EF
For email responses, remove "nospam_".