Click here to go to our main heat page
Click here to go to our page on thermal analysis
Why do you need to measure temperature? In the microwave world, just about everything varies over temperature. That phased array you want to operate may need a whole phone book of calibration settings once you consider that its temperature will change with transmitter duty cycle, and you will need to measure the temperature of each T/R module on the fly to apply the proper calibration. Further, increasing temperature means lower reliability. For this reason you need to know what the channel temperature of your FETs are. There is no convenient way to directly measure the temperature of something that small; you must measure the baseplate temperature using one of the temperature sensors described below, then do some math to determine channel temperature.
When we talk about temperature sensors, we are talking analog only at this point. Hey, we're microwave guys and live in an analog world, we hate digital stuff! There are a whole lot of ways to measure temperature, we'll discuss some of them below.
Thermocouples
Thermistors
RTDs
Current source transducers
Infrared transducers
Change-of-state devices
Current source transducers
Since transistors were developed way back in the middle ages, it has been known that most of their parameters vary over temperature. Sometimes you can use these effects to your advantage (but usually not!) A two-terminal current source, constructed from a bipolar transistor, exhibits a current that changes with temperature.
The current source temperature transducer in its most basic form is the AD590, made by Analog Devices. Click here to download the AD590 data sheet from Analog's web site. This part provides a current in microamps that is nearly exactly equal to its temperature in Kelvin. You read that right, you can throw away your thermistor calibration tables and read the temperature directly with your DVM! The AD590 can be used to measure from -55C to 150C. Any DC voltage from 4 volts to 44 volts will operate the device. This type of sensor is far less prone to errors due to lead resistance that a voltage sensor, thermistor or RTD might be. It's a beautiful thing!
On the down side, as pointed out by astute reader who works on payloads at an aerospace company best known for sending Gene Roddenberry's ashes into space, if you're only interested in a narrow range of temperatures on the Kelvin scale, and converting the AD590 output to digital using an analog-to-digital convertor, your higher-order bits are being thrown away. For this reason he prefers thermocouples for temperature measurements. Here is his reasoning:
"I don't like AD590s. They use them on some satellites, but the resolution we get is not that great. We are only interested in temperatures between about -20C to +80C, but the AD590 zero-point is at 0 Kelvin. Our A/D converter likes voltages between 0 and 5 volts, so we put a suitable resistor across the AD590 to convert current to voltage, and we end up with a voltage maybe between 3V and 4V (a fraction of our A/D range). A thermistor can be set up (with a parallel normalizing resistor) to give you a resistance between ~0 and 5K over the range we are interested in. Pump in a 1mA current source, and you'll get a nice 0-5V signal for your A/D converter to digitize for optimum resolution."
Infrared transducers
Here is a handy "non-contact" type of temperature transducer. However, the accuracy is limited. The temperature you sense depends on the color and shade of what you are trying to measure. Infrared transducers behave best on stuff that is painted flat black. Also, the "spot size" that is measured must be considered. You will never be able to measure FET channel temperatures using infrared. You will be able to measure the surface temperature of the ICs on your circuit card design within a few degrees. This will give you a fast answer to the question "which one of these bugs is gonna blow first?"
Change of state devices
An example of a state-change device is a liquid crystal. At the transition temperature, the crystal changes from clear to opaque. The problem with this type of measurement is that you need a whole lot of different phase change materials if you want to measure temperature over a large range. On top of that, "painting" your device with phase change material can be destructive.