PNS-Halpha camera
The problem of the "readout bleeding"



In this note we describe a problem detected in the new Halpha arm of the PNS, during run October 2006. This is meant to be an upload of the fault report
17448 (26-Oct-2006).

The problem is a sort of "vertical trail" in the images of the masks, in corrispondence of the mask holes. Its shape is similar to the saturation effect, or to an image obtained with arc lamps on during the ccd read-out (but it is not).

Section 1
The effect is illustrated in the following figure:



The figure refers to the file r903694.fit, in which the R-band filter was on and the Marconi2 CCD was used.

All the images had been taken with the Halpha and PNS shutters open all the time, mask on. The exposure time was constrained by shiching on and off the CuAr+CuNe lamps (the lamp was off ~10 seconds befor ethe CCD readout). As you can see from the radial profile, the image is not saturated, even if the shape reminds a saturation effect.


Section 2
We found that the problem is visible only if we reach a certain threshold, around 6000 -7000ADU. But the threshold is not constant.
In the following picture we show the radial profiles of the spots in several images, some of them with small counts and not showing the problem, some of them with many counts and showing the problem.





As before, the images were taken leaving the shuter open, and swiching on and off the arc lamps to regulate the exposure time.
It is difficoult to say if the effect is or is not linear with the exposure time, because the exposures are usually < 10 sec. and we do not have a proper control of the precise moment in which the lamp is on or off.

The readout bleeding is visible also in other detectors, so it is not directly related to the Marconi 2 chip. The technicias mounted another CCD (together with a new electronics, I guess it was EEV10). After few images in which the problem was not visible, the problem appeared again.

It is not due to some interferences caused by the filter motors in the Halpha camera. Initially we thougt it could be a possibilitybecause the motors are not activated by the controller (but by the cable of the shutter) and so they can be still active during the integration. But it is not the case: in fact, once the motors had parked the filter in the desired position, the cable is taken out and reconnected again to the PNS shutter, so there is no electric power in the motors!


We thought that maybe the effect was produced by the optics, so the first test was to take the filter out of the beam.
We took 2 exposures at different exposure times (the first r903701.fits with 5500 count, the second r903703.fits with 25000 counts, so much higher than the threshold).
There 2 images did not show the problem, as illustrated in the following picture.




Section 3
We took more images, and the problem appeared again, even without the filter. So the problem is not due to the filter.
Here you are an example (r903724.fits).



What happened between exposure 2 (r903703.fits) and exposure 3 (r903724.fits)?

  1. The marconi ccd controller was not attached on the Halpha camera, but standing loose on a table (because they took it off to use the other ccd), after the second exposure and before the 3rd, it was mounted on the Halpha camera.
  2. The PNS left EEV12 CCD crashed... we bring it back to life before the 3rd exposure. During the 2 good exposures, the EEV12 ccd was off.
  3. The technicians refilled the dewars with nitrogen
  4. ~2 hours (according tho the night log)
  5. We took many biases
My only suggestion - driven by the facts that a) we experimented images with no problem at all b) almost during the crash of EEV 12 the
problem was not visible - is that there are some interferences between the CCDs (EEV12 and Halpha).



Section 4


We observed this phonomenon also in the dome flats, again over a certain threshold.
The flats were taken leaving the shutters open, broad R-band filter in. We constrained the exposure time by switching on and off the flat field dome lamps (as in the previous case, lamps were off ~10 seconds before CCD read-out).

In the following picture we show two images without and with the problem respectively. You can see that the bad pixels (black spots) in the first image generates a readout bleeding in the second image. This effect cannot be produced by the optics!!!! Therefore the problem is not in the optics.




Here is a zoom in:






Suggestion: try again with Halpha camera on and EEV12, EEV13 off!!!!!