Ryouichi Kashikawa, LIR Engineer, NEC TOSHIBA Space Systems

AKATSUKI has five "eyes," i.e. five cameras, for capturing images of Venus. Each camera captures different wavelengths. AKATSUKI's cameras include the IR1 (for capturing wavelengths of about 1 μm), the IR2 (for wavelengths of about 2 μm), and the LIR (long-wave infrared), which uses the 10 μm wavelength. These three cameras for capturing the infrared region are complemented by the UVI, or ultraviolet imager, which captures the ultraviolet wavelengths, and the LAC, or lightning and airglow camera, which captures lightning and airglows at visible wavelengths. NEC was responsible for developing the LIR and UVI, as well as the mission system computer called DE for global control of the cameras and image processing. Ryouichi Kashikawa, the space camera expert who developed the LIR, shared with us the story of the development of this camera.

From the top: Lighting: Lighting and airglow camera (LAC),
temperature distribution of clouds: Long-wave infrared camera (LIR),
chemical substances at cloud tops: ultraviolet imager (UVI),
ground surface: 1 μm camera (IR1), lower atmosphere: 2 μm camera (IR2)

- What kind of sensor is the bolometer, which was employed for the first time in a space camera in the LIR?

Kashikawa: Most image sensors are CCDs, which are used in digital cameras and video cameras. Such sensors put out signals by converting the electrons excited in the semiconductor by the incident light into a voltage. However, creating an infrared sensor with the same technique (photon detection) would result in greater excitation from the temperature of the sensor itself than the signal unless the sensor is refrigerated.

On the other hand, the bolometer is a resistor that converts the received infrared rays into heat. When the temperature changes, so does the electrical resistance, and voltage fluctuations are read by passing a current through the bolometer. This technique (called thermal detection) has the significant advantage of not requiring refrigeration, allowing the creation of low-weight cameras that use less power, reducing the risk of equipment failure.

Regarding the statement that the LIR was the first case of an uncooled bolometer sensor being used in a space camera, I don't think we checked with all manufacturers, including overseas ones. At any rate, there appear to be no other cases, even overseas, of Venusian observation with the LIR's wavelength range.

- Talking about infrared cameras, did the cameras that were installed at the immigration security gates in airports at the time of the H1N1 flu outbreak in order to measure people's body temperature, showing the faces of those with a fever in red, use that bolometer?

Kashikawa: Yes, I believe so. NEC's Guidance and Electro-Optics Division originally developed bolometer infrared cameras for ground use, and one of the principal investigators who learned that such cameras were being sold for civilian applications since 2004 suggested the use of this technology for the observation of Venus. This launched the development of the LIR camera.

- And so you were put in charge of the LIR camera.

Kashikawa: Yes; and one more thing: Actually, just mounting the various circuits of an IC on a printed circuit board will not get you a working device. Extending circuit wiring for the same signal frequency band, which may for example cause a difference in potential where this is not permissible, ends up degrading performance. Thus, on top of making the circuit configuration as simple as possible, determining which wiring parts on the printed circuit board to shorten, and which to make thicker, is key to optimizing performance.
This is what designing space equipment is about: Just squeaking by while aiming for the lightest weight and highest performance possible under such constraints.
But even though we may mutter to ourselves that the civilian divisions have it good, being able to use all these great parts, making circuits by combining transistors, diodes and so on is actually the most fun part.

Another major difference with civilian cameras is the brightness of the photographic subject. The intensity of the light emitted from a body is proportional to the fourth power of the temperature of that body. The temperature condition of the photographic subject of a civilian camera is presumed to be around 300 K (27°C), while that of the top layer of clouds on Venus is 200 K (-53°C). This ratio is compounded by the fourth power proportionality, so that the intensity of the light that reaches the camera is less than one fifth.

- That's the Stefan-Boltzmann law^*1, isn't it?

Kashikawa: Yes. Moreover, because the peak of the spectral range falls outside the sensitivity range of 8 μm to 12 μm, the signal components are even smaller. Therefore, the sensor's sensitivity must be pushed as high as possible, causing the variations in the bolometer's element resistance between adjacent pixels to exceed the circuit's range.

- Causing the same problem as in the case of the photon detection sensor mentioned earlier, correct?

Kashikawa: That's right. But actually the nature of the variation differs. In the photon detection type, the number of electrons excited by the sensor's own temperature, etc., always has a degree of variation to the order of the square root of that number, whereas the element resistance of the bolometer does not vary over time and thus can be corrected.

Actually, the bolometer sensor is provided with a function to change the bias current passed through each element on a pixel-by-pixel basis. Whether the variation is on the + side or the - side of the circuit range is read for each pixel by a subsequent stage circuit, and these values are then returned to the bolometer sensor as the bias current correction parameter.

- Would accurate calibration of this correction yield clear images of the clouds on Venus?

Kashikawa: Right. The processing to cancel the differences between these two images is done with a piece of equipment called the DE (Digital Electronics image processor; for details, click here). The LIR performs bias current correction for each pixel as described above in real time, and the image data, which still contains more than 1000 levels of variations, must be passed to the DE without distorting the image signal that has just a few levels of variation by burying it in noise.

In order to realize these functions and performance, the LIR was designed using a larger number of development steps than usual. Normally, a working prototype model for ground testing is fabricated and then the flight model for actual launch is created, but in the case of the LIR, an in-house research budget was established prior to the fabrication of the prototype model, and a test model for evaluating the performance of the analog circuits and verifying the basic operation of other major circuits was created. In the case of most circuits, the PCB pattern is automatically generated from the circuit diagram, but this time, in the case of a six-layer PCB for example, circuit diagrams were traced by hand on six separate sheets of double-sized paper, for the purpose of deciding which sections of the wiring to make shorter, and which ones to make thicker, as mentioned above, taking into account interference among the six overlapping layers. Interestingly, when you get absorbed in pattern design, the flows of people on the train platform in the morning and evening start looking like electron flows.
In the end, we somehow managed to achieve our targeted level of performance.

The completed LIR flight model weighed 2.1 kg for the main unit only, and the power block was placed in a different enclosure. The use of a separate enclosure means additional weight, a luxury in space design, but this separation was necessary in order to minimize the effect of the heat emitted by the power supply on observation, and thus we asked the systems people to attach the main unit by itself to a mounting panel with a certain degree of temperature stability. The temperature of the bolometer element proper was controlled to a stability of within 0.1°C by a circuit that took up a whole printed circuit board—this circuit could be realized with just a few ICs for civilian cameras as I mentioned before.

- How do you feel about having developed the first bolometer used on a satellite?

Kashikawa: "I didn't do it on my own" pretty much sums it up. I received a tremendous amount of support from the principal investigators, several NEC divisions, and all those at the Guidance and Electro-Optics Division, which developed the civilian bolometer cameras. Many different manufacturers also participated in the LIR project. Showa Optronics, an NEC Group company, provided high-resolution, low-distortion lenses, and Tamagawa Seiki provided ultra-compact shutter motors. Another company, Axis Corporation, also supported development activities from the initial prototyping stage with its strong technological capabilities. True gratitude to all these parties is how I really feel.

I think that the LIR's development will be marked by three major moments of joy. The first one was when the prototype put out a clear image. The next one was when, after its launch, AKATSUKI photographed the Earth as a camera function test. Although the picture was taken from the nightside of the Earth, the LIR was able to show a round Earth by capturing the infrared light it radiates. The principal investigators too were overcome with joy when we received the image of the round Earth.

I expect the third time to be when AKATSUKI enters Venus's orbit and captures beautiful images of the planet.

Photo of the Earth taken by the LIR of AKATSUKI
during the early-stage function test

As the LIR flies toward Venus, Kashikawa already envisions its future travels to the far reaches of the solar system.

Researched and written by Shinya Matsuura, October 27, 2010

Other articles