What exactly is meant by a differential temperature thermal imager?

[Fraser]

 

[Max Planck]

https://www.hamamatsu.com/resources/pdf/ssd/e07_handbook_Thermal_detectors.pdf

See Fig. 2.4. A microbolometer redout circuit. There are two resistors, a reference one and a measurement one (microbolometer). The scene temperature is measured based on these two elements resistance (temperature) difference. Thus, a differential thermal imager.

In quantum detectors, simplifying, there is direct photon counting. Thus, absolute measurement.

Max

[Lambda]

Hello.

Thank you Max Planck for this document!

I was always asking myself what concretely was this "reference resistor" and how it was electronically implemented for obtaining the final differential output signal.

Best regards.

Stéphane

[Bill W]

Probably worth adding that there can be 'reference pixels'.  The resistor noted above is setting up the DC conditions of the integrating amplifier, removing or supplying the 'non signal' current in the active pixels.

ULIS (now Lynred) have a similar explanation in their early published papers, search for 'ULIS readout circuit':
https://www.researchgate.net/figure/Pixel-Readout-Architecture_fig6_242268489
and similar

Some detector designs had 'blind' or reference pixels, which were otherwise normal pixels but thermally connected to the substrate and therefore did not respond to the scene.

Bill

[Fraser]

It is interesting that the reference resistance is shown as a fixed value in that document. It was my understanding that in a thermal imaging camera microbolometer, there are active and blind pixels that are effectively identical but the blind pixels are not exposed to the incoming energy through the microbolomter window. The blind pixels are exposed to the same internal temperature changes in the microbolometer die as the active pixels so there is effectively a built in correction for die temperature change  The differential output is effectively that between two pixels of the same characteristics.... one seeing only the inside of the microbolometer and the other seeing the incoming energy from the optical block. It is a clever self compensating system to address the influence of changing microbolometer die temperature.

So in the diagram in that reference document, I would have shown both the ‘reference’ and active pixel resistance as variable and with no external excitation they should have resistance that closely tracks each other.

In the real World there are not blind pixels for each active pixel. There are a number of blind pixels on the microbolometer die that act as the blind pixel references for many active pixels.

It is some time since I studied microbolometer design and my memory is not what it used to be. I think I have captured the key points in this post however. It is important to understand why such a differential pixel reading scheme was needed.... we are dealing with what are effectively high sensitivity thermistors and changes in the temperature around them or induced within them due to the readout current will effectively skew measurement accuracy. The readout current is usually short duration pulses to manage self heating but the die temperature may be influenced by the ROIC beneath it or external temperature change. Even a microbolometer that is stabilised by a TEC will suffer some die temperature variation. Small thermistors are used to measure the die temperature for the TEC but the use of blind pixels and differential pixel output compensates for much of the die temperature changes that are unavoidable. In non stabilised microbolometer systems this countermeasure to die temperature change could be seen as essential. It is all about taming the undesirable thermal behaviour of the microbolometer.

Fraser

[Fraser]

Bill, you beat me to it   

Fraser

[Bill W]

It is interesting that the reference resistance is shown as a fixed value in that document.
Fraser

The so-called 'reference' is a fixed resistor, and is part of the column amplifier circuit, so there are only 160/320/384 etc of them.
Another ULIS paper figure attached from researchgate.net.
The 'blind' bolometer Rb is to provide the constant current ib to counteract the current ip for the 'typical' pixel at ambient.
Therefore only deviations from ip get integrated in the amplifier and appear as signal.  Even then those deviations in ip are mostly due to differences in pixel fixed resistance, not so much is actual thermal signal.

Bill