CFH12K - The CCD focal plane


The MIT Lincoln Laboratories produced the 12 CCDs used in CFH12K.

Figure 1: The MIT/LL 2048x4096 pixel CCD

The CFH12K mosaic focal plane
Figure 2: The 12,288x8,192 pixel CCD mosaic

The 12 CCDs arranged in a 6x2 mosaic create a sensitive area of 12,288x8,192 pixels (100,663,296 pixels per readout!).


Focal plane organisation and orientation
Figure 3: Focal plane organisation and orientation

The output organisation (output A or B used on each CCD) has evolved over the course of the last year and a half to improve noise performance of the camera. We however keep these changes to the strict necessary minimum.


CFH12K CCDs relative performance

There are two types of CCID20 devices: those made out of standard epitaxial silicon (EPI) and those made out of high resistivity bulk silicon (HiRho). The latter have a higher QE (up to 20%) in the red part of the spectrum and produce less fringing than EPI parts due to their larger thickness. The CFH12K contains both types of devices. They are grouped within the mosaic (all the HiRho parts packed together in the lower right corner) in order to cover large areas on the sky with a similar response.

Figure 4: CFH12K CCD type map:
standard EPI & high resistivity (HiRho)

The best CCD in term of sensitivity is CCD03. It should be used by observers interested in objects covering a field of view less than 7 by 14 arcminutes. It has a few cosmetic defaults but overall it is the best CCD in CFH12K.

The average gain over the whole mosaic is 1.6 electrons per ADU.

For some detectors in the mosaic (those with a high gain), the saturation occuring when the pixel well is full reduces the dynamic range allowed by the 16 bits converter: the resulting linearity domain (given in this table for a limit of 1% residual) can be as low as 51,000 ADUs. For the detectors with a lower gain, the dynamic is actually limited by the digital saturation at 65,535 ADUs.

All EPI parts are very similar and CCD09 can be taken as a reference for this type of CCD (the CFH12K exposure time calculator [DIET] numbers are based on that particular CCD).

Figure 5: CCD characteristics (RN = Read Noise, LIN = Linearity Domain)


CCD01, CCD02 and CCD07 (and to a lower level CCD09) have lower serial charge transfer efficiency (CTE) than the other CCDs: 0.99995 versus 0.99999 (at least). With such CTE, tails can be noticed on saturated objects. For non-saturated objects, all the electrons generated in the pixels will be kept within the object profile, i.e. no flux is lost for photometry analysis.

The effect of this lower than normal CTE (for a 2,048 pixel transfer) on the image quality is minor. The CFH12K spatial sampling of 0.2"/pixel (for an typical image quality of 0.7", best 0.5") actually limits the effect of PSF peak displacement to a tenth or two-tenth of a pixel.

The following table gives in percentage the ratio P'/P of the actually read intensity and the initial value in the worst case of a full serial register transfer (2,048 pixels, the read intensity being In=I(n)*CTE^n, with "n" the number of pixels transfered and I(n) the initial value prior transfer). The initial value is determined by a simple model of the FWHM over 3 pixels (for 0.4" FWHM) and 4 pixels (for 0.6" FWHM a very typical case for CFH12K).

A CTE of 0.99995 appears to be fine. It will "shift" the PSF a fraction of a pixel towards the outer edge but for astronomers requiring ultra-precise astrometry, this can be characterized since the calibration fields will be affected the same way. As per image quality degradation, 0.99995 will only affect by 0.1 to 0.2 pixels, which is fine with our current spatial sampling (for 0.6" FWHM the final effect would be then 6%).

         | FWHM    |       0.4"          |              0.6"	       |
     	 | Pixel   |  P1    P2     P3    |   P1      P2     P3     P4  |
	 | Value   |  I/2   I      I/2   |   I/2     I      I      I/2 |
     	 | 1.00000 |  100   100    100   |   100     100    100    100 |
     	 | 0.99999 |  98    99     102   |   98      99     100    102 |
     	 | 0.99995 |  90    95     110   |   90      95     100    110 |
     	 | 0.9999  |  81    90     120   |   81      90     100    120 |

Figure 6: The 0.99995 CTE on CCD01: a negligible effect on non-saturated objects


The cosmetic of CFH12K CCD mosaic is excellent. There is is a total of 200 bad columns, most of them are partial over the whole height of the CCD. And most importantly most of them are concentrated on CCD05.

CCDs CCD06, CCD08, CCD09, CCD10 and CCD11 are free of bad columns.

The largest sets of joined bad columns are all smaller than the gap between CCDs. Hence with the default dithering patterns made available for CFH12K, all the cosmetic defaults can be easily removed.

Figure 7: CFH12K CCD cosmetic: 200 bad columns (most of them on CCD05)


Mosaic geometry
With a short beam of f/4.2, depth of field is critical on such a large surface (21 cm by 14 cm). Emphasis was given on designing a focal plane structure that would achieve a flatness of 60 microns, the depth of field at the CFHT prime focus for a 0.4 arcsecond seeing (pixel size is 15 microns, providing a sampling of 0.2 arcsecond per pixel).

The CCDs individually are flat within 20 microns. With the twelve CCDs mounted together, a flatness of better than 100 microns is measured. The standard distribution of image quality across the field (up to the edges of the 12K) under the best seeing conditions available at CFHT prime focus (0.5") is less than 0.1".

The relative alignment of the CCDs along the X and Y axis was a low priority in the design of the focal plane. Nonetheless, proper manual mounting of the CCDs led to an amazingly good relative alignment of the devices: the gaps between the CCDs (both in X and Y) range from 28 to 43 pixels (about 500 microns) and the relative angles between devices range from -0.3 degree to +0.3 degree.

The average East-West (X) gap between the CCDs from the sensitive edges (last column of one CCD to the first column of the next CCD) is 38 pixels (570 um), or 7.8 arcseconds.

The average North-South (Y) gap between the CCDs from the sensitive edges (last line of one CCD to the first line of the next CCD) is 33 pixels (495 um), or 6.8 arcseconds.

The FITS headers contain the World Coordinates System information for each CCD individually. The astrometry precision is 0.4" based on on-sky data calibration.

Figure 8: CFH12K PSF across the central 8K8K area (courtesy of N. Kaiser, IfA)

Comments on the CFH12K pages to J.-C. Cuillandre: