Forgot to add... Source follower output has another good feature, it isolates our triode and the load (capacitive cable + low input resistance of the amp). If we have rel. high input amp (say, >100k), we can try anode output, too. Then our triode has load Rccs//Ramp - not so high anymore.
Our DN2540 facing upward (in TO-220 case) has leads (from left to right) - G,D,S. Note that D is connected at the tab. We must use a little heatsink (see the picture, hosted by "ImageShack" :-)), and thermally conductive insulator. Better not to mount heatsink on the chasis - safety and negative influence of the stray capacitance.
We`d use two series connected cheap LED, cathode (short lead) goes to ground.

I have seen measurements made by a trustworthy and skilled experimenter indicating that the mu output, at the currents we are looking at( 10 mA ) is granting output z in the few hundreds of Ohms.

So the low freq performance is dictated by the cap, and at mid-band won't be worse than a few hundred Ohms. Makes silly-expensive patch cords less of a requirment, no?
regards,
Douglas

You might ask OrCAD to send you their demo CD. Click here to go to the request form. It will analyze circuits having up to 50 parts (I think), which is plenty for many tube designs. I'll be happy to make a model of any of these circuits we're talking about, but I think you would enjoy having the demo version of OrCAD/PSpice.

The Kenwood L-07 pre-amp had optional 10m cables to connect to the L-07m monoblock amps. They claimed that the super-low output z of the pre made the 5000 pF of cable capacitance ( 39 and change feet worth...) a non-issue. It was capable of driving a capacitive load, or so the claims went.

So the few hundred ohms will do just fine with a few feet of moderatly capacitive patch cords...
regards,
Douglas