I'm afraid I'm completely out of my depth here.

I would like to use a tungsten filament to radiatively heat something near to it (in vacuum). My issue is I cannot figure out the easiest and/or cheapest way to control the power to the filament. The answer I keep circling around is something involving an SCR and maybe a transformer to step down the line voltage. But some things I've read (Chromalox's product materials, mostly) indicate that I don't need a step-down transformer, or if I did I just need a phase-angle fired SCR. Other places I've looked (product literature from other people, such as Omega) indicate that an SCR, even phase-angle fired, can't drive an inductive load reliably. I've also considered putting the transformer first, but most SCRs I've seen require line voltage input (either for auxiliary electronics or for their own operation).

Another point of confusion is that I am not sure how, if at all, I am controlling the current in the filament (high temperature resistance=0.1 ohms ~= 15x low temperature resistance). The SCR just turns the output on and off, after all, to produce some sort of average power, right?

Does anyone have experience with this sort of system or goal? The filament in question is designed to be driven at 7-ish volts and up to 62 amps, although I assume I could always drive it with higher voltage. I would like to plug into a 120V single phase wall socket. I am not wedded to the SCR, or really any element of the design, as long as it can be computer-controlled eventually (and get real hot).

Thank you!


We have a test rig that works something like this, to heat up small objects (typically cubes of material with sides a few mm long) to about 1000C fairly quickly.

The heat source is a "tube" of ordinary (and fairly cheap) tungsten-halogen car headlight bulbs surrounding the test piece, rated 12V 55W each, with home made parabolic reflectors covered in cooking foil behind them. The bulbs have their own internal reflectors to provide a directional beam, as required for a car headlight. The foil reflectors are just to mop up any stray light and heat output - mainly from the radiation that missed the test piece and needs reflecting back from the opposite side of the "tube."

So long as the power supply doesn't exceed 12V DC the bulbs will survive OK. So far we have only fried one - and that was when somebody knocked the thing of the bench onto the floor while it was running!

We used a PID controller that "just happened to be lying around" to control the voltage of a high current DC power supply, with a thermocouple providing the feedback loop into the controller. We never bothered to tune the PID much - the light output is continually flickering "randomly" at a few Hz, but the thermal inertia of the rig and the test piece keeps their temperatures pretty stable, and the bulbs don't seem to mind that level of abuse.

So don't make your design more complicated than it needs to be - throw something together, try it, and fix any problems you hit. We never really had to do any hard thinking about the "fix the problems" step - it worked "well enough" almost from the first test run.

A phase-angle supply switching at 50 or 60 Hz might give the bulbs a harder time than this "very noisy DC voltage" approach - but we haven't tried it.

Note: everything in this rig this running in air, not in a vacuum, so there is a lot of heat lost through convection. Even so, four bulbs (220W) can get it up to temperature in 10 or 20 minutes.

  • 1
    $\begingroup$ "Don't over engineer it" good ! $\endgroup$
    – Solar Mike
    Oct 26 '17 at 4:11

I recommend a variac (variable ac transformer) rated at 500 watts, This will downconvert the wall outlet voltage 120VAC to anything in the range of 0 to 120. Alternatively, to get easier control, you could put a 10 to 1 stepdown transformer after the variac; as the variac voltage is turned up or down over the range of 0 to 120VAC, the output of the stepdown tranformer will vary from 0 to 12VAC. Note that the stepdown transformer has to be rated at 500 watts too.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.