results

CCD42-20 NIMO in Bonn

A mechanical sample

Test conditions:

Cooling: LN-dewar.
Operating temperature: -100C (except for dark current studies).
Controller: One of the slow scan Bonn BOCCIA CCU's (CCD Control Unit).
Read out: Through one output with 50 underscan and 50 overscan pixels (2148 pixels total). Read out speed is 32 or 64 Kpix/sec.

Full well (~ 150,000 e-)

The full well was estimated through a photon transfer curve. Strong non-linearity starts around 150,000 e- as expected. The diagram is based on raw images which still contain the PRNU noise. This is the reason for the slight bending of the curve (which would be linear if only poisson noise were present).

Read out noise (RON = 3.5 - 4.0 e-)

The RON was determined at the lower pixel rate (32 Kpix/sec).

Linearity (< +-0.3% up to 100,000e-)

Measurements were made with a stabelized light source and an iris type shutter. A linear fit was calculated for a signal range of up to 90,000e-. The residuals for the fitted signal range show that deviations are well within +-0.3%. Larger deviations at low signal levels are due to the typical inaccuracies of the iris type shutter at short exposure times. Beyond 100,000e- deviations from this fit get larger.

Horizontal Charge transfer efficiency (HCTE = 0.999,9988)

The HCTE was measured with the usual "deferred charge" method: In a well exposed image one does look for the signal level in the first oversan pixel. Two diagrams show this step between column 2098 (50 underscan + 2048 image pixels) and column 2099 with low and high resolution respectively. To improve S/N rows 20 through 200 were averaged. From the image signal level (~ 14000 ADU) and the signal in the first overscan column (~ 35 ADU) one finds (14000 - 35)/14000 ^ 1/2098 = 0.999,9988. For comparison the sharp step between the underscan area and the image area (at columns 50 and 51) is shown again with low and high resolution.

Vertical Charge transfer efficiency (VCTE = TBD)

Dark current

The optical system of DIVA is forseen to be kept at around -20C. The focal plane array is probably cooled down to -30C.
Dark current was measured at -30C and at -20C.

Dark current measurements are usually done by integrating over a certain time, long enough to generate a significant charge floor. For the TDI driven DIVA CCDs we are faced with a completely different situation: We are operating in a regime where (surface) dark current is dynamically suppressed. A detailed discussion is given in a paper by B.E. Burke and St.A. Gajar ("Dynamic suppression of Interface-State Dark Current in Buried-Channel CCDs", IEEE transactions on electron devices, 1991, Vol. 38, No. 2, p285-290). Here only a short summary:
   It is well known that surface dark current is reduced very efficiently by keeping CCD phases in inversion (the basis for MPP/AIMO). Factors of up to 100 for dark current reduction are quoted. After switching back to depletion the surface dark current does NOT raise immediately to its steady state value but the recovery takes some time. This is goverened by a time constant. This time constant depends on temperature and is the time after which the dark current has reached about 90% of its steady state value (e.g. from Fig.2 of the paper mentioned above one finds time constants of 0.1sec and 10000sec at 0C and -80C respectively!). During TDI clocking (i.e. during read out) all phases are continously switched between inversion and depletion. Then the dark current is reduced if the duration of one inversion/depletion cycle (or one row shift) is of the order or shorter than the time constant. For the DIVA CCDs a row shift takes about 1.4msec.
   To quantify the effect of dynamic surface dark current suppression for the CCD42-20 NIMO dark current was measured in steady state and at various row shift cycle times. The following table gives numbers and links to diagrams. The shortest row shift cycle time possible with the available CCD controller is currently 5.6msec.
   The fitted functions which are indicated in the diagrams should not be used to extrapolate the dark current to the 1.4msec row shift cycle time for DIVA. There is no physics behind these functions (I just used was MS-Exel did offer) and at some point we will reach the bulk dark current and/or spurious charge levels which are unknown so far. Bulk dark current is not reduced!

Temperature [C]	Steady state dark current [e-/sec/pixel]	Dark current at 5.6msec cycle time [e-/sec/pixel]	Diagram
-30	100	3	tdi dark current @-30C
-20	350	19	tdi dark current @-20C

Klaus Reif, 24.6.2002