In early June 2008 I bought an old Vigor Burnout Furnace (model CA 1065)
for $95 from a seller on Craigslist. The furnace appears to have been
built in 1979, but does work.

The furnace came with a so-called "infinite" control, which is an
electromechanical contraption with ten intensity settings (from mostly
off to always on). What varies is the average duty cycle of power on
versus power off. There is no feedback control. This is far too crude
and clumsy to heat-treat metals, so I decided to add an electronic
furnace controller.

So I bought a 1/32 DIN PID controller and solid-state relay (SSR) from
Omega Engineering. I'm still using the thermocouple that came with the
furnace, but not for long, as the original thermocouple is old and
oxidized and a bit off. But it works well enough for now, and will hold
to within a degree or so of a temperature, far better than is required
in heat treating of say O1 oil hardening steel or A2 air hardening steel.

The big problem was that the original design didn't have a big enough
heatsink on the SSR, and there wasn't enough thermal isolation between
the control box and the furnace box, so the controller and SSR both got
too hot when the furnace went to max temperature (about 990 degrees C).
The controller would get to 50 degrees C, which is right at its upper
temperature limit, which is asking for reliability problems. The SSR
would get almost to 60 degrees C, which is also in its range, but asking
for trouble.

So, for the SSR I got a large heatsink from Omega, which dropped the
temperature to 36 degrees C.

The controller was a bit more difficult. I ended up making a set of
spacers from 0.5" diameter gummy aluminum rod on the lathe, and cutting
two aluminum plates with clearance holes, all to space the control box
away from the body of the furnace, with two parallel plates between,
thermally isolating control box from furnace body. Now, the PID
controller temperature is 35 degrees C max.

At this point, I have declared victory, as 35 or 36 degrees C is low
enough that reliability won't be much reduced.

Joe Gwinn