On Aug 24, 2019, at 1:49 PM, Mike Dodd firstname.lastname@example.org [ap-gto] <email@example.com> wrote:I run a 1600MM and have settled on cooling at either -15 or -10C, depending on the season. The amount of noise that is removed from the sensor by going colder than -15 is truly negligible. You can use PI's Statistics process to investigate the difference using your darks. Open a master dark of the same exposure time, gain, and offset for each temperature you have. Turn nomalization (and, optionally, scientific notation) off via the checkboxes. Hit the wrench icon on the right side and make sure that Variance and Std. Deviation display are on. Then compare the Variance and StdDev values for your -10C dark against the equivalent -15C dark, and -20C, and so-on. You'll find that after -15C, there really isn't much of an appreciable difference, and hence reduction, in noise on these sensors. This tends to hold true for CMOS in general.
For example, my master dark for an exposure time of 180 seconds @ Gain=139, Offset=21:
I've done test -25C and -30C darks and the differences between those are similarly negligible, but I unfortunately neglected to save the results after deciding I wouldn't bother going that cold.
Given that, I'd much rather have more cooler power headroom instead, preferring to run it at no more than 75% power. I'm also a mobile imager, so there's also some consideration for the amount of juice the TEC is slurping up when running at high duty cycles.