Modelling the Effects of Tube Degradation with Time

[sbench]

Hi All,
Some Spices (including LTSpice) provide a .STEP command that allows you to alter variables, models or parameters during the simulation. Normally this function allows you to gain some insight into the sensitivity of circuit performance to parts variation. This can also apply to vacuum tubes, to provide you with a guide to how your circuit varies with device to device variation, or what happens to your circuit as the tube degrades over life.

As we all know, for triodes, as they age, the MU remains essentially constant (since that parameter is primarily related to physical dimensions) but gm slowly degrades towards zero (and thus plate resistance rises).

In the DMTRIODEP.INC file, I added a parameter called EMISSIONRANGE that allows this variation. This added one additional global line to the file:
.param EMISSIONRANGE=1
and altered the transfer equation in TRIODE and TRIODENH subcircuits. The default makes sure that if you don't use this parameter, everything remains unaltered.

However, you can use the step command as...
.STEP param EMISSIONRANGE List 0.5 0.9 1.0 1.1
for instance to give you performance plots of 4 tubes... a "bogey" (1.0), a slightly worse than average (0.9), a slightly "hotter" than average (1.1) and a tube with lots of hours on it (0.5) to see how your circuit changes. You can, of course, put in any values you desire into that list.

Attached are two .GIF files illustrating this use... The first shows a set of D3a/7721 tube curves and the source "circuit" that produces these curves. In this instance, I have provided a "step" commant with 0.5 and 1.0 steps, to show you how the tube curves are altered. The second one shows a second generation of my LR phono preamp frequency response with the tubes at 0.4 nominal, 0.5 nominal (tubes at replacement stage), 0.9 nominal, 1.0, and 1.1 nominal (5 curves).

Steve