> what about triode conection with this tube? any predictable behaviour?
Mu at 10mA 250V is 26. We say "Variable Mu" but in fact Mu does not vary much, rarely over 2:1 at useful currents.
At 250V 10mA the Gm (or S) is 6mA/V in pentode mode; as triode it may be 25% higher due to G2 current, but let's ignore that. Call this Gm a cathode internal resistance of 1/6 V/mA, which is 166 ohms. Therefore plate resistance is 4,333 ohms.
In general, I don't like to work the tube below 2mA, because it tends to be too weak for transformer coupling and can have trouble with resistance coupling.
Most tubes go way down in current before Gm drops much. This one is odd.
250V and -35V bias puts it around 1mA, try that. Gm has dropped to 0.12mA/V, or 8K, and triode plate resistance is perhaps near 26*8K or 20*8K or 200K-100K area.
6/0.12 or 50:1 (34dB!) range of Gm with just 10:1 range of plate current is exceptionally good.
When resistance-coupled, change of Rp fights your change of gain. Pentode mode is attractive.
When resistance-coupled, large shift of plate voltage will upset your next stage. Attack times may be faster than the coupling caps can recover. (What IS this new idea to bass-cut limiters? I don't want to lose bass, I don't want bass to clip... music is music.) Say you can tolerate a 36V shift. And current shifts 10mA-1mA. Then the plate resistor can be 4K. Drops 40V at 10mA and 4V at 1mA. Full gain is about 4K/166R or 24. Min gain is about 4K/8K or about 0.5.
4K per side sure sounds like 10KCT, a common transformer.
Gross overload in a limiter usually starts at minimum gain. The gross output overload at min gain is 4V, which suggests 8V at the input. THD will be intolerable long before this point.
The graph on page 8 is measured distortion. We will work between S=6mA/V and S=0.12mA/V. I think "K" is German for distortion. Over most of this range, K is 1% for signal levels near 200mV, rising to 1,000mV at our minimum gain of 0.12mA/V. This graph is surely drawn for single-ended operation, distortion is mostly even-order, we will work push-pull, we might go a bit higher. But it seems we must set input threshold near 1,000mV/50= 20mV to get low THD in the S=0.6mA/V (20dB GR) zone.
Noise: the full-gain Gm is equal to a 166R resistor at 1000 degree heat, or about 500 ohms at room temp, or near 0.3 microvolts over the audio band. That's in triode mode in theory. Pentodes have partition noise which randomly levers the electrons very badly, noise is higher. Page 6 seems to say the equivalent noise resistance is 2K at high gain and rises very fast for lower gains. 2K is about 2uV of noise.
At max gain (no-limiting, a "normal" condition) the S/N is 20mV/2uV or 80dB. At min gain, 500K noise resistance is near 10uV of noise, so S/N is like 1,000mV/10uV or 100dB. It is correct for the calculated S/N to rise as limiter gain drops. Of course what the ear hears is the 80dB max S/N in the below-threshold range, you won't hear 100dB S/N when the music is over-loud.
Maximum output in heavy GR is near 0.5V, you will need to gain-up to nominal 1.23V or 2.45V level. (Huh... it would just-about drive 0.316V in 10K semi-pro load without a booster.)
If the tubes are cheap, try them. The problem with pentodes is you can't get two per bottle, so matching is less likely than a twin-triode. But with a large supply and DIY economics, it will probably work.
-35V is a big grid signal, which means a big sidechain. The decay resistor must not exceed 3M per tube or 1M5 for two or 750K for four; say 1Meg. The ratio of attack to release rates tells the ratio of decay resistor to attack resistor. I like 1mS-2mS attack and 1S-2S decay, so I need 1K attack resistor. If I were careless enough to apply a 34dB overload, and expected perfect attack, I'd need 35V/1K= 35mA drive current. And either polarity. So the sidechain would end in about a 1 Watt amplifier. Not a Fairchild's two-6V6, but not some little 12AT6 either.
> it also seems to have fairly high plate resistance...
Pentodes always do. Good HF IF pentodes must have high Rp to get high Q in the IF tuning circuits.
> it not seems to be made to work with a plate resistor in normal operation
No. Go find an IF amplifier plan. Plate current flows through a ~~1mH coil in parallel with a 150pFd cap. This resonates at 455KHz, a popular IF frequency; it also has very low DC resistance. If fed with 250V B+ (~~125V was more common) it will sit at 249.5V-248V at any current.
But note the plots for R(g2)=60K. As triode, with R(P+g2)= 12K, the same curve may apply. However the 108V plate voltage shift will not cap-couple into the next stage smoothly, and 24K transformer winding is the upper end of good practice.
