Products > Thermal Imaging

Why a disc shape sensor in the "Micro A10 Indigo" thermal core ?



I was wondering why in the thermal core "Micro A10 Indigo" the micro bolometer sensor was under a disc shape and not rectangular shape ?
especially knowing that the resolution is about 160x120 with pixel dimensions around 51 um x 51 um (if i am correct).

From that i would have expected to see a rectangular sensor with about the following (round about) dimensions: around 8.2 mm X 6.1 mm, which is not apparently the case.

But in reality i have a kind of disc shape "sensor" around 25 mm diameter (from my memory), see in the following video, at around 2 min in the timeline.

Thank you in advance for your lights about this question.


Early microbolometers could have either round or rectangular windows. The actual microbolometer die was still rectangular in units that had a round window. It just came down to which casing of the die the manufacturer elected to use. It was not uncommon for the modules that contained the microbolometer die to be a modified version of other casings used in the microwave and optical electronics industry. The casings of microbolometers have become more refined over the years and the vacuum volume smaller. From memory, the early Honeywell microbolometers were housed in a module that had a round window. There may have been a very good reason to choose a round window, such as the modules previous use having had a round metal cap on it , or the manufacturing technology coping better with soldering a round window into the module than a rectangular window. Where a gas tight seal is needed, as in this case, it could well be that a circular solder seal was more reliable at the time. Modern microbolometers can make use of current packaging technology and use either a module with integrated vacuum chamber or they incorporate the vacuum chamber into a silicon multilayer IC with open bond wires. The latter design is quite fragile and may be protected within an outer module casing if desired.

In the past, and likely still true now, companies like ULIS would design and produce a microbolometer die and then incorporate it in more than one casing embodiment to suit the needs of designers. Some casings might be larger, but rugged, whilst others might be the smallest that can be produced to offer product miniaturisation. The same die might be offered with thermoelectric (Peltier) stabilisation, whilst another is offered in a non temperature stabilised version for customers who did not need the greater die temperature stability. This is why it is interesting to read the ULIS microbolometer data sheets.... you can often see more than one casing design yet the actual microbolometer die is clearly the same. That said, some microbolometer dies appeared to be enhanced to give lower NeTD or incorporated thermoelectric temperature stabilisation, as already stated.

When looking at the development of ULIS microbolometers over the years, it provides an insight into hoe microbolometers have developed with each new generation that was released. Some developments are listed below:

1. The casings became smaller and more suited to mass production.
2. The NeTD figure was reduced with each new generation
3. Visible artifacts such as column noise reduced with improved production methods
4. Thermoelectric temperature stabilisation became optional and offered power consumption reduction plus other benefits
5. The microbolometer readout timing electronics became integrated into the ROIC section of the die. Only a master clock is needed
6. The ADC was integrated into the ROIC die to reduce the chip count of the host electronics package
7. "Bare bones" microbolometer dies became a reality. The silicon wafer became the microbolometer module by using the MEMS manufacturing process. It incorporates the vacuum cell and window. Such bare bones microbolometers are often bonded to a PCBA and attached to it via very fine bond wires or BGA solder bumps

It is worthy of note that it was no simple task to incorporate the ADC into the ROIC die as it produces significant heat. Such heat is unwelcome anywhere near a microbolometer and must be carefully managed. This was especially so when electing to not incorporate thermo-electric temperature stabilisation.

Early microbolometers used large pixels, had large, expensive, casings and large windows. Modern microbolometers use far smaller pixels, compact cheaper to produce casings and small windows. All this has helped to reduce the cost of these devices and to drive down the cost to the end user. Progress  :-+ Added to that, the modern noise reduction and image processing power of the host electronics package has greatly improved image quality to offset the lower signal to noise ratio of smaller pixels.

Worthy of note is the fact that Raytheon employed a 'bare bones' microbolometer die in their Thermal Eye 2000AS series cameras. They were well ahead of their time in this development. Very impressive. Picture taken from Bill_W's post to be found here:


A 'basic' round window microbolometer where the window can clearly be seen soldered to the module that houses the microbolometer die.

Bill W:
Early on it was usually down to what detector package was available - they were often from Kyocera designed for visible light CCD's - and then how you get a germanium front window on there in place of the glass.  Germanium windows tended to be round until you get to custom in high quantities. 

A fair number of detectors are simply whole front germanium and far larger than optics would ever require - ease of sealing winning out over reduced germanium content / cost.

Round does also allow a wider choice of lenses to be used.  Consider the light cone from the corner pixel to the lens rear element.  The further the window is way from the detector surface the more rounded this minimum aperture becomes.


Fraser, Bill, again thank you very much for your crystal clear explanations !  :-+

Best regards.



[0] Message Index

There was an error while thanking
Go to full version