Temperature-Controlled Test Bed

Image-heavy thread warning!

As part of my ongoing project to make a transistor tester to my own satisfaction, I solicited ideas. One was a temperature-controlled test bed to hold transistors at a desired temperature for testing.

This is it (so far):


The working schematic:
This circuit is a derivative of one in the AD592 data sheet, modified to suit the need and the parts in my spares at home.


Control is *very* tight, with high (enough for me) rates of heating and cooling. With R3 at something like 25k, max. temperature is 150ºF, and it will try to control down to sub-ambient. The whole works is at a stopping point until the rest of the tester is built, at which time it can be properly mounted to the chassis. Transistors under test will be held in place with adjustable clamps and kelvin clips for smaller devices. TO-3s will get a socket mounted . . . somehow . . . .

Current cost for this: $0. Thank God for a job where I can snag "junk" like this free!

The heart of the system is the AD592 temperature sensor. You give it a few volts and it gives a current output directly proportional to temperature in ºK. Run that through a resistor and you have a temperature-dependent voltage. Compare that to the voltage on a set-point resistive divider (in this case, a 3/4 Bourns pot) and you can switch an output based on temperature.

My outputs are power to a pair of Minco ribbon heaters in series and a pair of Intel processor fans in series. The switches are 2N6043 darlington NPNs, driven by an LM324 that both limits their base drive voltage and pretends to be a comparator. The temperature sensor circuit is running off a 10V output from a 7805, and the whole shebang is eventually going to run from an Omron industrial 24V power supply. I considered using Peltier coolers vs. fans but dropped the idea after burning out a cooler by accident :(

The outputs of this thermostat circuit are mounted on a heatsink from a long-defunct home theater system-in-a-box. The AD592 is under the zip tie, and is a TO-92 device next to a life-size TO-3 drawing (to give you an idea how big this heat sink really is).


The circuit is on three Radio Shack project boards bodged together. I managed to keep the components mostly on the component side. Trust me, this IS the cleaned-up version! The output transistors run pretty hot, without the heat sinks. I burned a finger before realizing the need for the transistors to be mounted off-board on the heat sinks. Fortunately, the parts bin had a couple of sinks with mounting holes that lined up perfectly with the mounting holes at the corners of the PCBs . . . .


Solder side, note the baby copper heat sink soldered to unused pads on the PCBs to hold the boards together:


This is version 2.5. The v1.x circuit was based on a resistor with a strong thermal coefficient, and was finicky. The sockets were installed after the first set of opamps blew up (literally). This was from a Vacuum General/Tylan/Millipore capacitance manometer's temperature regulator and worked for what it was intended to do . . . which was not this project.