I'm using a 6AU6 pentode as a constant current sink in the tail of an LTP splitter. I can only find one example of such a CCS, namely, the Lowther LL26 (please see attachment below), in which the pentode's control grid is decoupled to its cathode via a 0.1uF capacitor and is also connected to the negative rail via a 1M resistor. That's how I've got mine connected right now and it's working OK.
However, it has been pointed out to me that I would get a better result without the grid decoupled like that. Lowther was using a 6BJ6 remoter cutoff pentode, whereas the 6AU6 I'm using is sharp cutoff, in case that has any bearing on it.
The obvious change to make, to get rid of the decoupling, would be to connect the control grid directly to the negative rail, dispensing with the 1M resistor and 0.1uF capacitor altogether. Please can anyone tell me if there's anything wrong with doing it that way? Do I still need the 1M resistor for some reason?
However, it has been pointed out to me that I would get a better result without the grid decoupled like that. Lowther was using a 6BJ6 remoter cutoff pentode, whereas the 6AU6 I'm using is sharp cutoff, in case that has any bearing on it.
The obvious change to make, to get rid of the decoupling, would be to connect the control grid directly to the negative rail, dispensing with the 1M resistor and 0.1uF capacitor altogether. Please can anyone tell me if there's anything wrong with doing it that way? Do I still need the 1M resistor for some reason?
Attachments
Mystery to me
That arrangement is new to me, and seems somewhat mysterious. Could it be intended to prevent oscillation? (The cap will roll off the HF response of the CCS?) I don't think sharp or remote characteristics would matter, as the current is fixed. Anyway, as you imply, the normal way would be to connect g1 directly to ground, and skip the cap. However, you might like to use a 1k (or so) resistor between grid and ground (mounted close to grid) to prevent oscillation (a grid stopper, that is). Ground can act as an antenna too, so this is not as silly as it might seem. Also, I would try connecting the screen decoupling cap to the cathode rather than ground. This might improve the CCS plate resistance somewhat - maybe not significantly. If it oscillates, a small (100-470 ohm) resistor between screen and screen decoupling cap might solve this.
Good luck,
That arrangement is new to me, and seems somewhat mysterious. Could it be intended to prevent oscillation? (The cap will roll off the HF response of the CCS?) I don't think sharp or remote characteristics would matter, as the current is fixed. Anyway, as you imply, the normal way would be to connect g1 directly to ground, and skip the cap. However, you might like to use a 1k (or so) resistor between grid and ground (mounted close to grid) to prevent oscillation (a grid stopper, that is). Ground can act as an antenna too, so this is not as silly as it might seem. Also, I would try connecting the screen decoupling cap to the cathode rather than ground. This might improve the CCS plate resistance somewhat - maybe not significantly. If it oscillates, a small (100-470 ohm) resistor between screen and screen decoupling cap might solve this.
Good luck,
You want the screen to be firmly coupled to the cathode. In cases where the cathode is at AC ground, you can couple the screen to ground, too, but if its voltage does not follow the cathode's, the tube will not be acting as a pure pentode and CS source resistance will suffer.
Yep.
Also with a resistance under the cathode (and not decoupling it out of circuit as the 1M grid leak and stuff does), degeneration increases impedance, flattening the current characteristic.
Tim
Also with a resistance under the cathode (and not decoupling it out of circuit as the 1M grid leak and stuff does), degeneration increases impedance, flattening the current characteristic.
Tim
- Status
- Not open for further replies.