Engineers working on a ground-based infrared telescope search for a a missing ampere of drive current
By Bill Rappoport, Contributing Writer
About 20 years ago I was on a team working on a ground based infrared telescope. The telescope was designed to track unresolved points of infrared radiation seen against a sky background.
In the infrared, the sky looks bright, even at night, due to thermal emissions from the atmosphere. An oscillating mirror in the optical path was used to rapidly move the point image on and off an infrared detector. When the detector was pointed away from the target, the signal on the detector was due only to the sky; during the next half cycle, the combined sky and target were observed. The resulting AC signal amplitude was the difference between the radiance of the bright sky and sky plus target. The demodulated signal provided the amplitude of the target alone.
The oscillating mirror was an interesting piece of hardware. The oscillation frequency was 3000 Hz to put the signal frequency above the 1/f noise knee of the detector. To achieve this high frequency, and a high mechanical Q for efficiency, the mirror and its torsion bar were machined out of a single piece of aluminum. A magnet bonded to the mirror was attracted by an electromagnet coil mounted on the baseplate; a second magnet was sensed by a feedback sensor.
The whole assembly, including both the mechanical components and drive electronics, was designed and built by an inventive one-man local shop. When it was delivered, we tested it on the lab bench and it worked great, although the unit made an extremely annoying high-pitched squeal. This was not unexpected; the assembly was, after all, a powered tuning fork, and the acoustic noise would not be a problem in the intended application. So everything seemed good until we installed it onto the telescope.
Once installed, the unit refused to operate; it sat there as silent as a door knob. A couple of young engineers were assigned to troubleshoot. They made sure there were no physical obstructions, checked the power leads, and traced the signals through the sense amplifier, the compensation network, and to the drive amplifier. Eventually they reported back that the drive amplifier was pumping out plenty of current, but that didn’t seem to be making the mirror budge. They had no idea where the ampere or so of drive current was going.
I thought for a few moments on the differences between the lab setup and the telescope installation. During the lab testing, I recalled, the unit was sitting on a wooden top bench, and the mirror assembly was electrically floating. With that realization, I was pretty sure of the cause of the problem.
I suggested to the engineers that they check for a short circuit from the drive coil to the chassis. My suspicions were soon confirmed when a short was indeed found. In the lab, without a ground return to the power supply, the short didn’t matter. On the telescope, the mirror was bolted into an aluminum structure. The power supply returns were tied to the chassis, thus providing a low resistance path from the amplifier back to the power supply. The amplifier could pump out all of the current it was capable of, but most of it was being shunted through the chassis.
We removed the unit from the telescope, and disassembled the coil. Upon inspection, we found a defect in the coil wire insulation that allowed contact between the wire and the metal bobbin. Once the coil was rewound, and the unit reinstalled, it squealed as expected.