Fri, 12/29/2023 - 01:03
I have just purchased a Player One IMX533 mono camera to replace my CCD and would appreciate a recommendation on optimal mode and gain settings for BVRI photometry. This is the 'baby' version of the IMX571.
All sensor characteristic charts can be found here: https://player-one-astronomy.com/product/ares-m-pro-usb3-0-mono-camera-imx533/
Also should I bother binning 2x2 with CMOS, and if so should I let the camera driver bin or bin in post-processing? At 1x1 I'm at 0.38"/pixel (C11 at 2060mm focal length).
Thanks very much,
Keith
Keith:
A key question is, "What will limit your photometry performance: light pollution or raw camera performance?" I'm going to assume that the answer is "light pollution" (this is my situation at my location on an island in Rhode Island) since this is what I'm most familiar with.
When you're fighting light pollution, Enemy Number 1 is the high background image noise due to Poisson variations in the number of light pollution photons entering the camera during each exposure. Only two things can help with this: keeping your photometry aperture as small as possible (reducing the amount of light pollution entering your measurement aperture) and gathering as many total photons as possible from the target (to drive up your signal/noise ratio). Read noise doesn't matter under these conditions. This is where a great mount will pay off, if it enables you to reduce your FWHM a bit -- this is necessary if you're going to achieve any real gain by reducing aperture radius.
The camera has a 14-bit A/D converter, so it can generate pixel ADU levels from 0 to 16,383. There appears to be no penalty whatsoever for using the low-read-noise (LRN) mode, so let's start with a gain setting of 0 in LRN mode. From the graphs, we see:
We do a quick check for A/D converter saturation: 72,889/4.449 = 16,383. This tells us that A/D converter saturation (rather than true e- well depth) is what is causing nonlinearity at saturation, which is a great condition to be in, because it means the camera is probably really linear right up to 16,383.
Then adjust exposure time until your brightest star of interest is not saturating. This will give you best SNR in a light-polluted sky.
You also asked about binning. I run my QHY268M (IMX571 -- same size pixels) binned 3x3, giving me a binned pixel scale of 1.0 arcsecond. I gain exactly one benefit from binning like this: my image file sizes are smaller (by a factor of 9) than an unbinned image. Photometry performance is identical, whether binned or unbinned, as long as you scale your photometry apertures according to your binning. I do the binning during post-processing, but before the image is written to disk; my main reason for doing this is to help know exactly how close I am to (unbinned) pixel saturation, since having any single unbinned pixel saturated invalidates the entire binned pixel, and it isn't always obvious by looking only at the binned pixels whether or not one of the contributing unbinned pixels was saturated. (This is mentioned in the ASAVSO Photometry Guide, which gives some advice on how to deal with this if you're at the mercy of the vendor's software.)
Good luck & keep us posted!
- Mark M
Thanks very much Mark! Yes I'm at a suburban site so not particularly dark skies. So LRN mode at gain 125 with the read noise reduction is then not much benefit given sky noise dominates (and full well only 16666 e-)?
Of note there appears to be a bit of a dip in read noise at gain 5 without much loss of full well (so dynamic range peaks), but I'll assume this is of negligible benefit over gain 0 so I'll go with the latter.
Keith
Keith:
Yes, I noticed that behavior around a gain setting of 5. I didn't understand why it's there, so I didn't say anything about it. But if you look at the dynamic range graph, it actually shows that the dynamic range is better at a gain setting of 5. So, yes, it probably makes sense to use the gain setting of 5 instead of 0.
I encourage you to measure gain and read noise yourself. A number of observers have discovered that their actual camera behavior deviated in some significant ways from the vendor curves. Knowledge is power.
- Mark M
I was able to reproduce Player One's curves very well including the dynamic range 'bump' at gain 5. However I'm really struggling to get a reproducible linearity curve, not sure if its partly light source stability.
One thing that is clear is the the count rate (ADU/s) really takes off below about 2000 ADU, for example about 25% higher ADU/s at 500 ADU compared to 30,000 ADU (note the ADU are scaled to a 16 bit range). Does anyone have data for a comparable sensor like the IMX571 at low ADU values? Most plots I see are pretty scarce there.
Keith
I'm wondering if someone could take a look at some sample images and see if they look "normal" for a CMOS, specifically the low-level banding (most apparent in I and R filters) that changes every frame and therefore cannot be entirely calibrated out. As the peak-to-peak amplitude of the banding appears to be about 1.5ADU I'm assuming this is not impactful to photometry with reasonable star signal. I'm coming from a SBIG ST8XME CCD that while is much more noisy, does not exhibit this behaviour, so just would appreciate a CMOS 'sanity check'.
Player One says this banding is normal, and even show this in a sample dark on their website. This is the Ares-M Pro IMX533 camera.
I've put some sample images (a few raw lights, calibrated lights, a few raw darks and the master dark (20-frame average) and the master flats (16-frame average)).
https://drive.google.com/drive/folders/1qoyczj8A1FYflPYiCrP5UNACov4Pq9bV?usp=sharing
Much appreciated!
Keith