I am using this camera mostly for photometry. I purchased it specifically for its non-anti-blooming chip, and am curious about its dynamic range. The attached "Star Profile" plot shows a peak value that clips at roughly 47,500 ADU. Should it not be 2^16, or 65,500 ADU? Tom
65,535 is the maximum value the Analog-to-Digital converter will produce; that's separate from the electron saturation per pixel of the KAF-6303 sensor itself, specified as 85,000-120,000 e-. http://www.onsemi.com/pub_link/Collateral/KAF-6303-D.PDF The camera design tries to map the range of the sensor to the ADC, so its always going to be adjusted so that the upper end of the sensor is lower than the top end of the ADC. It will vary from camera to camera. It's never going to be a perfect match. Factors like chip temperature / thermal noise impact the saturation to some extent as well, so from one night to the next, the range will be a bit different. In an ideal world, that 120,000 would correspond to 65k ADC out, but if your chip was saturating at 85,000 e-, then it will never have a full 65k out, more like 46,400. (i.e. 65536/120,000 * 85,000). It might be tweakable at the factory, but this seems pretty normal to me.
Thanks, Colin. So my understanding is that the electron saturation has a wide range, and my particular reference image is saturated near 85,000 e-. Would you expect that there would be some conditions under which my images would saturate near 120,000 e-? If not, might I gain some dynamic range by having SBIG tweak the ADC upward? Tom
Hi Tom, if I understand it correctly, your chip will have a certain saturation at a given set of conditions (thermal, electrical). I don't know enough of the semiconductor physics to know the thermal/electrical conditions to predict that. You might ping Bill Lynch at SBIG service to get his opinion.
The cameras are adjusted at the factory. I'll give you a rough idea of how this is done. The goal is to fully sample the dynamic range of the sensor. This is actually not fixed - well depth depends on the substrate voltage setting, so first we have to optimize that. Substrate voltage is a setting that can increase well depth... until the chip starts messing up, and then you back it off a little. (On ABG chips anti-blooming effectiveness is also a trade-off versus well depth... but not relevant here.) Our procedure in production is to reduce the converter gain so it never saturates, and then adjust the substrate to optimize it. (For some CCD models the sensor manufacturer provides a recommended substrate level - it's stamped on their shipping box.) Once the substrate is adjusted, we readjust the A/D converter gain and offset to target 1,000 bias (to prevent clipping) and sensor saturation slightly above 65,535. Why slightly above? Because people complain - they paid for 16 bits dammit, and they want all their bits! This actually very slightly reduces their real dynamic range, but in practice it's not significant and avoids complaints so we go with the flow! (Also some sensors have double the dynamic range when binned (higher well depth in readout register), so in some cases we drop the converter gain when binning to preserve dynamic range.) Sensors such as the 6303 are actually oversampled by a 16-bit converter. In fact most sensors are; only really large pixel cameras can push the limits of a 16-bit converter. Most cameras on the market have even smaller pixels and are ludicrously oversampled. Market pressures push the manufacturers to use 16-bit even if it's not really appropriate. So, to summarize, there's nothing seriously wrong with your camera. Your not losing anything by running it like that. But if you want to get it tweaked you can send it in for adjustment.
Thanks for the explanations. As it stands, my photometry has very good precision, so I am not disappointed with the camera. I should have mentioned in my original post that I have been running 2x2 binned. I am most curious about what I have gained in terms of linearity with this camera versus similar SBIG cameras that have anti-blooming. If my linearity ends at around 45,000 ADU, what would be the expected non-linear level for, say, a 16803 camera? This information would come in handy next time I am asked that question by somebody who already owns one of these cameras. Tom
Linearity has little to do with the A/D converter - they are spec'd to 1/2 ADU level. Really these cameras are extremely linear until you approach saturation, and then it's the sensor that is going nonlinear.
I suppose the more direct question is: did I gain anything for photometry by buying an ST-6303 rather than a model with anti-blooming?
Both sensors have 100,000 e- full well. However ABG sonsors may be less linear near the top than NABG. Microlenses are also a potential consideration. Some people don't like to use them for photometry. KAF-16801 is better than 16803 for these applications but costs much more.
