The first CP40 detector was designed at CERGA for interferometry observations by A. Blazit (Blazit, 1976, 1987). It is an two-stage intensified CCD camera followed by a photon analyzer which computes the coordinates of the photo-events (Foy, 1988b). A duplication of this prototype was financed by INSU to make it available to the French astronomical community and in particular to the instruments of the TBL. This detector was made by the Observatoire de Paris in collaboration with A. Blazit, and was partly designed to be the detector of this speckle camera.
The entrance faceplate is a S20R photo-cathode
followed by a two stage amplification :
electrostatic focusing (Gen I) for the front stage,
and micro-channel plate (Gen II) for the second stage.
The intensifier is cooled to around
-25 
C with a liquid nitrogen cryostat through a
thermal link with a conductive mesh.
The intensified image is split into 4 quadrants
with a fiber optics device whose 4 outputs are
coupled to 4 CCD's (Thomson TH7861), to allow
for quick reading. Each CCD has a format of
square pixels (23 
23 ).
As the two-stage amplifier has a magnification of 2.3x,
this leads to 52 -wide pixels
at the entrance photo-cathode (whose useful sensitive area
is therefore 30 mm 40 mm).
Each CCD is read every 20 msec and the corresponding image is processed in real time by a ``photo centering'' device. Each bright spot on the phosphor caused by the detection of a photon is analyzed and its centroid is computed with an accuracy of one CCD pixel. The coordinates are then sent to a computer to be stored and processed.
The exposure time is a fixed 20 msec, which may be too large for speckle applications when the coherence time is smaller than this value (which is often the case in the visible). To circumvent this difficulty a rotating shutter has been implemented which reduces the exposure time to 5 or 10 msec. This shutter interrupts the light beam in the speckle camera during 1/2 or 3/4 of the 20 msec exposure time with a rotating opaque sector synchronized with the reading clock of the CP40 (phase-locked motor).
The geometrical distortion caused by the two-stage amplifier is rather large, of the order of 20% in the edges (Thiebault, 1994).
Because of a dead zone between the 4 image quadrants, we decided to use only one quadrant and made a special ``off-axis'' mechanical interface between PISCO and the CP40 in order to align the center of the selected quadrant with the optical axis of PISCO.
Another problem is a strong non-uniformity of the sensitivity of the photo-cathode within a single quadrant (down to nearly zero in one edge), which can hardly be corrected by a flat field map and nevertheless causes a big non-uniformity of the signal to noise ratio within the elementary frames. The photometry of the image restoration process was badly affected for intrinsically big objects that spread on non-uniform zones of the whole image.