Building a crude bolometer, or the world's worst thermal camera.

A bolometer detects the heat from absorbing light. This comes with limits sensitivity and speed, but works at wavelengths which are very difficult to detect, like thermal infrared.¹

At its simplest, a bolometer is simply a black object attached to a heat sensor, both somewhat isolated from the surroundings. Typically a thermistor is used as the heat sensor, but I did not have any on hand, so I used a 1N4148 silicon diode. A diode’s forward voltage falls with temperature, generally around -2 mV/K.

I covered the diode with lampblack (carbon) becuase it is a fairly good wideband light absorber, and built up a quick circuit test the principle:

After the diode cooled to ambient temperature, placing my finger on it dropped the output voltage by a good 30 mV. But, this simple circuit won’t work as a bolometer, becuase it is effected by ambient temperature, even a fraction of a degree of error will overpower the tiny effect from thermal radiation. To compensate for this, I added a second diode as a baseline temperatred reference, and measured the difference between the two voltage drops:

There will be some difference even at the same temperature becuase diodes will not be perfectly matched, but it can easily be zeroed out. We are getting somewhere, but this diode bridge circuit still can’t measure the actual amount of heat flowing into the sensor.

To do this, I added small heater, a 1k resistor to the sensor. To maintain a constant temperature, the total power flowing into the sensor has to be constant, therefore the needed heater power will fall by exactly the input power. With an opamp adjusting the heater power to keep a constant sensor temperature, the input power can be measured as the drop in applied heater power:

Just about any opamp will do, as long as it can suppy 5 mA of current, which nearly all can. If yours has an offset null adjustment, you could use it for the bias adjustment in place a potentiometer on the diode bridge.

I glued to the sensor diode (D1) to the heater, but taping them together works fine. The assembly should not be touching the circuit board to keep heat from escaping before it can be measured. Watch out for the photoelectric effect, the diodes need to be shielded from light if they are in a glass package (like the 1N4148). The thermal mass of the absorber/sensor/heater assembly should be as small as possible to keep the thermal time constant reasonable.

The detector is quite sensitive to drafts, make sure to shield the diodes from any air currents. Commercial bolometers are often placed in a vacuum, but most common vacuum chamber materials are opaque to thermal infrared, making them useless.

To make a measurement, first measure the baseline heater voltage with the sensor covered, then expose sensor to light or thermal radiation. After it reaches equilibrium (the output stops changing), measure the heater voltage again. Convert the 2 voltages to powers, Power = Voltage^2 / 1000 Ohms, and subtract for the measured power.²

Covering the sensor except for a small aperture it turns it into a terrible 1 pixel thermal camera. Mine can detect the thermal radiation from a person from a few meters away, but it takes a good 20 seconds to respond.

It also works as a half decent laser power meter, but with a 1k heater resistor running off 5 volts, it will max out at just 25 mW. This could be improved using a smaller value resistor for the heater, using 100 ohms would bring the max up to 250 mW.

With a 50 ohm resistor attached to the sensor, it works as an RF power meter, but I can’t get more then around 20 dB of dynamic range. On the plus side, it works from DC to daylight (literally).