To CMOS or not to CMOS

IC405 Flaming star nebula.

CMOS sensor based cameras are becoming more and more popular among astroimaging amateurs. What differs CMOS from CCD? Is this really revolution? How to get the best from CMOS? I will try to answer these questions basing on my experience with QHY163 mono and color cooled CMOS cameras.

When you take a look into the technical specifications CMOS and CCD cameras are actually very similar. They both offer similar range of quantum efficiency, that is between 40-70% for the selection of amateur devices. So not much to gain here. The pixel well depth normalized to the same micron scale is pretty similar, maybe a little bit better for newer CMOS cameras. So you can expect pixel well depth for about 20000 electrons for 3-4um pixel cameras, about 40000 electrons for 6-7um cameras, and so on. CMOS wins here a bit due to (usually) newer technology. Photons are converted to photoelectrons due to the same photoelectric effect. But here in this stage some differences can already be noted. In CCD sensors photoelectrons captured in the pixels are being read using special clocking signal, that transfers each pixel charge one pixel by one to the output register. Then they are converted to analog voltage, amplified and digitized in analog to digital converter, so they can be transferred to computer or stored. On contrary in CMOS sensors each pixel is surrounded with its own small electrical circuit that is responsible for readout process.

Melotte 15 in IC1805 nebula – narrowband composite made with QHY163M camera.

There is important difference between CCD and CMOS sensors that is due to this fact – CMOS sensor readout noise is very low. And CMOS sensors are very fast. Read process is so fast, that for example with QHY163 camera it is possible to read almost 20 full frames per second, that is 320 million of pixels. That of course requires fast computer with USB3.0 interface. For smaller subframes it is possible to achieve much faster frame rate, over 100fps. And what about readout noise? It modern CMOS cameras it is very low, significantly lower than in CCD cameras. So there is no need to use very long exposures to achieve decent signal to noise ratio in the single frame. However there is common misconception saying that it is possible to produce the same imaging result using CMOS camera in much shorter time, than using CCD camera. That is true only for single very short exposure times in a range of several seconds. For this specific case main noise source in the image is readout noise, so CMOS wins here due to low readout noise. But when the exposure time is getting longer, other noise sources (mainly sky background noise) starts to dominate and differences between images from CMOS and CCD sensor becomes smaller and smaller, and eventually it depends mostly on sensor quantum efficiency. That is why CMOS cameras are much more suitable for short time planetary imaging and for so called deep sky object “lucky imaging” where many short frames (in a range between 0.1 and 10s) are shot and then stacked. That technique enables to use large aperture telescopes without guiding and achieve very decent results that can beat in some extent bad seeing (one can select only best frames and then align and stack them in a similar way planetary imaging is performed). Doing the same with CCD camera would give much worse outcome due to relatively large readout noise that is introduced with every single frame.

Elephant Trunk nebula area pictured with QHY163C color camera and 135mm lens.

So are there any other differences that we should know? Well not really. Usually CCD cameras are equipped with 16 bit analog to digital converter, when CMOS converters are 12 or 14 bit. That is no reason to worry however. 16 bit resolution for most CCD amateur cameras are a little bit overkill. Dynamic range (so the ratio of pixel well capacity to readout noise) for CCD amateur cameras are usually in a range of 10-12 EV, so this ADC resolution will be enough. Plus for deep sky imaging we stack many of single images and that technique increases actual dynamic range of the final image. Normalized pixel well depth is also similar. CMOS cameras have usually lower pixel well depth, but also pixel itself is smaller. So when we compare the same actual area of CCD and CMOS sensor it will turn out, that modern CMOS amateur sensors have actually larger electron capacity. For example, when you replace your Kodak 8300 based camera with QHY163 you will keep the same field of view (because both sensors have the same size), you will gain resolution (if optics allows) and also gain about 40% of pixel capacity, so less number of stars will be oversaturated. If you ask for hot pixels then CMOS sensors behave in very similar way like modern CCD sensors (like for example in the Atik ONE cameras). So much better like old CCD sensors (like Kodak 8300 or 11000 series). 

Center of Auriga constellation. This is composite of Ha made with QHY163M camera and RGB color made with QHY163C camera. Instrument was 135mm telephoto lens.

So cooled CMOS cameras seems to be all-in-one cameras suitable for deep sky imaging, planetary imaging and video astronomy as well. They also offers very friendly size to price ratio. However there must be some drawbacks for sure. And there are. I already explained lower analog to digital converter bit resolution. Another thing is small pixel size in CMOS sensors, but actually you should consider only the physical sensor size – if that fits to your imaging setup and desired field of view. Due to low readout noise of CMOS sensors there is plenty of room for downsizing images, and high resolution should not be considered as disadvantage. It can of course show defects and flaws of our optics, but well, these defects existed before as well, that is not the small pixel camera that made them. It only showed them. Higher number of pixels and fast CMOS operations sometimes may force us to upgrade computer we use for image acquisition – that is something we need to remember, especially for color OSC cameras. For example single frame from 20Mpx QHY183C camera after debayerization will occupy about 120MB of disk space. One more thing worth mentioning is amp glow that is still present in CMOS cameras, although in much less extent than in the past. Amp glow can be noticed in the frames after stretching up as some light gradient in the corners or near frame edges. It is caused by CMOS sensor electronics that heats up a little some sensor areas and that effects with some more electrons collected in the pixels that are a little bit warmer. In modern CMOS cameras amp glow is barely visible in luminance frames and little bit more visible in raw frames made with narrowband filters. But it can be completely removed with dark frames calibration. Another substantial fact is, that at this day CMOS based astro cameras are mainly targeted for small and medium size telescopes. CMOS sensors origins in consumer electronics, and they are mostly small sized. Not so long time ago majority of CMOS astro cameras were planetary cameras without cooling and with small sensors. Nowadays more medium sized sensors become available (like in QHY183 or QHY163 cameras), but still monochromatic CMOS sensors are made up to 4/3 or APS-C format, and no more. There are larger CMOS based astro cameras, but only with color RGB matrix. So if you already have large instrument and you are interested in monochromatic camera that will give you decent field of view, you still need to choose among Kodak CCD sensors. 

Crab nebula narrowband and RGB composite made with QHY163M camera.

So, the fundamental question would be if the CMOS camera for you? There is no clear statement on it, but only some hints and clues. If you have large instrument with good quality mount and are interested in large field of view, then you probably need to look for large CCD sensor camera. But if you have small or medium sized telescope and want to do astropictures you can definitely consider buying CMOS camera. If you would like to have one all rounder for deeps sky and planetary imaging – well, CMOS will do. If you have large instrument without decent mount and guiding (like dobson newtonian with tracking platform) then CMOS will let you achieve decent results with lucky imaging. Short exposed images from CMOS camera will be much better than from CCD camera. However if you do astro pictures using long exposures only and stack many of them, there will be not so much difference in signal to noise ratio between 10 hours of exposures from CCD and CMOS cameras. For this specific case sensor quantum efficiency will be the crucial factor. But on the other hand CMOS cameras usually have smaller pixel that gives you better resolution (if optics quality allows). And due to low readout noise you can do shorter single exposures, so mount and guiding quality can be lower. And last but not least – CMOS cameras offer very good bang for the buck ratio and they are becoming larger and less expensive every year. 

Clear skies with CMOS photodetectors!

And here is selection of astro images I made recently with QHY163 mono and color CMOS cameras and 130mm Photoline refractor:

2 thoughts on “To CMOS or not to CMOS

  1. Jolo, thank you very much for this article. On the our last astroparty in Zatom I have seen a few models of CMOS camera. As I am happy owner of ATIK CCD camera I started to wonder where is the diffrence? Is it a new fashin? Your articel make it clear. Thanks!

    Rafal

    1. It is a little bit a new fashion, but mostly we are pretty happy that someting is going on. Not a revolution, but more camera models in the market is good for us 🙂 Plus prices are quite tempting as well 🙂

      Lukasz

Leave a Reply

Your email address will not be published. Required fields are marked *