Holding the ELT M2 mirror without deforming it: an active support

The secondary mirror (M2) of the Extremely Large Telescope (ELT) is the largest convex precision mirror ever made. Very wide yet very thin, it sags under its own weight. At Safran Reosc, it is measured regularly on an interferometric test bench while it is being polished. To hold it without bending it, the team developed a dedicated active support.

My contribution: I designed the electronic and software control system of the active support, and I led the system validation.

The challenge: applying the right force field

• In the telescope, the mirror is held by its cell at several support points, which compensate for its sag under gravity.
• During measurement, the support must apply exactly the same force field as the telescope cell, with very high accuracy on each force.
• Any force error shows up in the measurement: it bends the mirror and misleads the polishing.

The choice: a controlled pneumatic support

Two solutions were compared: a passive mechanical whiffletree, as in the telescope cell, or a pneumatic support. The pneumatic support won: a proven technique on the VLT, a force field that is easy to change, continuous force measurement and a design that can be reused for the M3 mirror. The price to pay: it needs a control system.

• Pneumatic cylinders, each connected to a mirror support point through a high-accuracy force sensor, calibrated around its working force.
• Gas tanks, acting as capacitors: they filter ground vibrations and limit the effect of a leak on the control loop.
• Flexures at the interface, stiff enough to hold the mirror but soft enough to limit parasitic lateral forces.
• A control system that drives all the cylinders to apply the nominal force field, while setting the mirror height and tilt.

Validating each building block, then the whole system

I started the validation with the subsystems: each cylinder was calibrated, then the cylinder, sensor and electronics assembly was tested on a dedicated bench.

• Drift: the force stays stable over the whole duration of a measurement.
• Linearity: the force-pressure relationship stays linear, with low hysteresis, over the whole working range.

I then tested the complete system with the M2 mirror on it.

• Repeatability: the same command, applied several times, brings the mirror back to the same height and tilt, well within the required tolerances.
• Stability: over several hours, the mirror position and each cylinder's force remain very stable.

Optical performance

The active support is part of the test bench: its defects would show up in some mirror deformation shapes, such as astigmatism and trefoil.

• Measurement repeatability: full measurements repeated over several days stay well below the customer's requirements.
• Lateral forces: after handling, forces may build up in the flexures. A simple procedure releases them by lowering the cylinders, and the measurement regains its accuracy.
• Mirror rotation: we removed the mirror, rotated it and put it back. Both measurements agree, within the set criteria. This test, one of the most demanding of the validation, took several days.

Key takeaways

• The active support applies the telescope force field with comfortable margins against the customer's requirements.
• A more accurate measurement lets the polishing converge faster than expected.
• The same concept was reused for the ELT M3 mirror.

Further reading

H. Monin, R. Mercier-Ythier, N. Lefeuvre et al., Development of a 18 point active supporting tool for the metrology of the ELT M2 secondary mirror, Proc. SPIE 13021-18 (2024), SPIE · DOI