This makes me wonder... how can this effect
vary from camera to another? It seems to me, it's not a firmware
bug : .
Hi Marko,
since no one else posted about your basic question, thought I'd give it a shot.
I don't know that much about CCD fabrication, but I used to work in an Intel fab for memory products (static RAM) some years ago. I'm assuming the same principles apply, in that I imagine the CCD sensor is fabricated using photolithographic techniques. (I.e., it's made like layers in a PC "paint" software package; you lay down successive layers on top of each other to build the final product). If so, then there are two key parameters that can be unique to
each CCD sensor (and hence, G3) produced: the alignment of the photolithographic layers, and the defects in the sensor.
The alignment of the various layers will be unique to each wafer. So the alignment of the layers in the CCD sensor of your G3 won't be exactly the same as the next guy's. Why does that matter? Well, again I'm speculating, but perhaps as the alignment gets worse, there is a decrease in the insulating ability of the insulating wells between photodetector sites. (See, for example,
http://www.bythom.com/ccds.htm ). Maybe you have a sensor that is on the edge of acceptable alignments (or outside of spec, for all I know). The insulating ability of your wells is lower, and so you get the spillover more easily.
It is also possible that there is simply a defect in your sensor, unrelated to alignment problems, that basically has the same effect.
Either way, it is easy for me to understand how your G3 might not perform as well as some other G3 in conditions where the sensor is really being pushed to the limit -- trying to accumulate huge numbers of photoelectrons by shooting such an intense light source.
It's your camera, not mine, but if it were me: 1) I'd try to find a reproducable way of triggering the effect. I.e., how about a 150 watt tungston lightbulb shot with wideish aperature and slowish time from X feet away? If you get some standard test that always triggers the "bug", then other people can try it with their G3s, and also, you can check any repair to see if it actually helped. 2) Personally I'd take up the offer for a repair if that involved replacing the CCD sensor. Especially if other people try their G3s with the test and yours seems more prone to this problem than others. You might get one with better alignment, if yours is iffy, or you might get one without the defect that's causing the problem, if that's the case. If you have a way of triggering the error it might be helpful to the repair people, and then you can make sure any repair has actually made a difference.
Just as a final thought, with the memory products, the effects of marginal alignment or sub-critical defects wasn't necessarily to make the device fail, it was to decrease the useful operating range (usually in temperature, in that case. For example, it wouldn't run over a military-spec temperature range, but would over a consumer-spec temperature range). You might have an "in-spec" camera, but from your gallery you obviously are a sophisticated user who is pushing the envelope (of light intensity, in this case). Trying to get a better particular sensor might be sensible in your case.