Home-Made Plate Curves

[Ardee]

I decided to try drawing my own plate curves using an X-Y plotter and need some help.

The problem is that if I quickly sweep the plate volts from 0 - 200v on a 6BX7 with a fixed voltage on the grid, I get a nice curve initially, but at the end of the sweep the plate current drops significantly over the next few seconds. This means that the curve shape actually varies according to how fast I sweep the HT. Am I right in thinking that this is due to heating of the tube and, if so, how do I get meaningful curves out of this?

Grid voltage has a source resistance of 5k maximum and does not vary. Plate voltage is also stable at any setting.

[ilimzn]

I use a similar setup.
It depends what you define as 'useful'.
First of all, it helps to pre-draw the maximum dissipation curve onto your graph paper. If you expect to run the tube hot, then meaningful curves would be for a heated tube. In that case, draw them 'backwards' from the highest voltage/current.
If you want something universal, draw a separate set for cold and hot. Cold will require a quick sweep from zero volts up and then back down, and a pause before plotting the next curve for the next gate voltage.
You are right about the drop-off, under normal conditions it happens because of thermal dilation of the anode, it's effective distance from the cathode increases slightly. Also, larger variations are possible if you go over the dissipation limit, use positive grid supply etc.
When curves are actually drawn by the manufacturer, it's done in a 'one-shot' manner, with the actual sweep being short, and with a substantial pause inbetween. Tube variation is taken into account by normally having a set of representative tubes in parallel, then scaling current down by the number of tubes in the set.

[Ardee]

Thanks, that is really useful information and advice. I have never seen any reference to this thermal effect but it does seem pretty important.