Optique théorique et appliquée
The Homer test bench
The elements of Homer
The real time computer (RTC)
The development of the Real Time Computer has benefited from a high degree of synergy with the INCA project. It is supplied by the Shaktiware company. The hardware solution as well as the installed algorithms are common to the two projects although some of the latter will have greater importance for HOMER. Specific developments have been carried out in order to interface the RTC with the ANDOR WFS camera and the two ALPAO mirrors.
The RTC developed by Shaktiware is a PC running under a Linux 64-bit operating system, fitted with an AMD K8 Athlon 64 processor, running at 1.8 GHz. The RTC is based on C code giving it great flexibility of use and upgrading.
Figure 1: RTC developed by Shaktiware
Algorithms of the RTC
The RTC is based on 4 modules:
- the equalization module that performs the pre-processing of the acquired images (handling of dead pixels, application of a background and a gain card)
- the wave front computation module that provides wave front measurements in terms of gradient
- the control module that deduces from these measurements the command to be sent to the mirrors to correct the wave front
- the module applying the control voltages that apply several different processes (saturation management, addition of turbulent voltages) to the voltages calculated by the control module.
We also have the following functionalities at our disposal:
- multi-zone wave front analysis: a digital windowing of the sub-pupil in several sub-zones of interest corresponding to a multi-object analysis (see figure 3 in the WFS module)
- a wave front calculation by conventional CoG, by weighted CoG and by correlation for each sub-zone
- the possibility of applying a thresholding function that is static (at a pre-determined fixed value) or dynamic (recalculated for each image in accordance with a defined procedure)
- the application of a classic control (least squares reconstructor and integrator) or optimal (LQG) control [1]
- the systematic addition to the voltages vector, calculated by the control, of a supplementary voltages vector, variable over time, representing a known and controlled perturbation. This is used to simulate an atmospheric turbulence on the deformable mirrors and to firstly avoid the use of phase screens.

[1] LQG: Linear Quadratic Gaussian: Control law of a reconstructed state feedback form based on a Kalman filtering. In AO without mirrors dynamics, the predicted value of the turbulent phase is obtained with a Kalman filtering then simply projected onto the deformable mirror(s).