Controlling a chopping mirror

Moving a mirror shouldn’t be rocket science. But moving a mirror back and forth five times per second at nanometre accuracy is a different matter. This is what SRON instrument scientist Robert Huisman achieved for the HIFI instrument on board the Herschel Space Observatory and the soon-to-be-built METIS instrument of the European Extremely Large telescope (E-ELT).

With a 39-metre main mirror, the E-ELT will be the largest telescope on Earth when it opens for business around 2024. But for one of the three instruments that will analyse the light harvested by this mega telescope, a tiny mirror will be critical to its success or failure. Robert Huisman designed its control system. On 12 February, he will defend his PhD thesis describing how this was done.

The mirror is part of METIS, an instrument designed to analyse infrared light. ‘The problem is that all objects radiate some infrared light’, explains Huisman. ‘So if you want to study a faint infrared source, the light you receive can be many orders of magnitude below the noise from the atmosphere.’ It’s like studying a firefly sitting next to a car headlight.

Astronomers have a way of dealing with such noise. ‘The basic solution is to alternate recordings of the object with a recording of a dark patch of sky near the object. The dark patch gives you the noise, which you subtract from the object.’ Typically, this is done by switching between object and dark patch five times per second. ‘You need to do it frequently, because the noise may not be constant.’

Milliseconds

But you can’t just flip a mega telescope – or even a smaller one – at such a high frequency. That’s where the small mirror comes in. The light collected by the telescope is transferred to the scientific instruments. One of the mirrors inside METIS is moveable. ‘By tilting it slightly, the detectors collect light from a different part of the sky.’ Thus, astronomers can switch between the object and a dark patch without moving the entire telescope.

Huisman was project leader for the development of a ‘chopper mechanism’, as it’s called for the HIFI instrument on board the Herschel Space Observatory. ‘You must switch between positions in milliseconds; otherwise you lose too much valuable observing time. But you also have to switch the mirror at an accuracy of just a few dozen nanometres.’ The mirror must accelerate, decelerate and then freeze in the right position.

The mirror is moved by actuators, little motors that are controlled by a computer system. And this system, basically a box full of mathematics, was designed by Huisman. The most basic system is a feedback loop: the actuator moves the mirror, a sensor checks the movement and compares it to the planned trajectory, and this information feeds back into the actuator. ‘But such a feedback system is a bit slow. So you must add a second system that has more power and can move the mirror quite fast.’

Huisman designed a new control system, in which the feedback system only kicks in when the mirror reaches its required position. This control strategy showed very promising results when applied to the HIFI chopper, but could not meet the specifications for the METIS chopper. ‘The system exhibited hysteresis, a difficult to model non-linearity caused by memory effects in the soft iron of the actuators.’ Huisman shows a graph of mirror movement, which oscillates around the required position for milliseconds. ‘This oscillation means the mirror doesn’t reach the required position to within the specified tolerance for quite a while, because of the very narrow margin that METIS requires.’ In the end, he solved this by applying a repetitive controller which has the ability to learn with every new repetition. After four or five switches, the system got the hang of it and the mirror ‘chopped’ beautifully.

‘So in the end, we were able to show that the system was at or very near to the specifications which the construction team had set us’, Huisman says. This wasn’t just a victory for his control programme: ‘We also closely collaborated with the company that actually built the mirror system, Janssen Precision Engineering (JPE). So at each design step, we could evaluate what this meant for the performance.’ This integrated design approach led to several changes in the design phase. ‘If the company had just given us a finished product, we wouldn’t have achieved the required accuracy.’

The demonstrator mirror works as planned. Huisman and his colleagues at both SRON and the Systems Control group of the ENgineering and TEchnology institute Groningen (ENTEG) at the University of Groningen, led by Huisman’s second supervisor Professor Jacquelien Scherpen, have also made recommendations for improvements to the final mirror. METIS was commissioned by the Dutch astronomy institute NOVA, and NOVA will decide who is going to build the real chopping mirror. Will Huisman work on it? ‘That hasn’t been decided yet. I’m working on different projects at SRON now, but then again I know how to make this work.’