RkBlog

Hardware, programming and astronomy tutorials and reviews.

Solar System imaging cameras and hardware for 2019

What's the best or most promising hardware?

In my look at planetary imaging cameras for 2016 I nominated IMX224 as best color sensor and sensors such as IMX178, IMX265, IMX250 and IMX252 as best mono variants. We are almost half a year into 2019 and those picks are still one of the best if not the best. What did changed and what is upcoming for Solar System imaging? Let’s find out.

The cameras

Last few years we didn't saw much changes on the Solar System imaging market. ZWO was quite quick to the market with Sony IMX CMOS based cameras, then QHY followed addressing also some amp glow issues that showed up in the first ASI models. Later ZWO camera revisions also incorporated those fixes. After that we got variants with TEC cooling for DS imaging and few models with bigger sensors, generally for DS imaging. After that we got other vendors like Touptek (also branded as RisingTech and by some astro companies like Altair Astro) to follow.

More recently ZWO as well as other companies were focusing on CMOS DS imaging cameras - we got Panasonic MN34230ALJ based color and mono cameras long time ago but ASI1600MM was a new big mono DS camera since a long time and was quite affordable when compared to other similar cameras. Other companies like Atik, QHY and Touptek released their own variants with this sensor. More recently more APS-C alike in size color Sony based cameras were released for the DS imaging market. During 2019 we got teases from ZWO about a true 16-bit DS cameras while QHY is trying to push their GSense sensor based 16-bit scientific cameras.

As you can tell DS imaging is the the main focus of camera makers. There isn't much new that can enter Solar System imaging market so other markets must be found and Sony CCDs vanishing from DS imaging leaved a void that needs to be filled with new models.

But not all hope is lost, there are some interesting developments going around sensors usable for Solar System imaging or both - DS and Solar System imaging.

Some bigger pixels finally?

If you have a very slow telescope like an f/20 Dall Kirkham or Cassegrain then most of the latest sensors will just have to small pixels. That's one of reasons why you could look for bigger pixels. The other ones can be for example lack of backfocus in an large ultra fast Newtonian or need to make that f/4-5 into f/10 or more for atmospheric dispersion corrector to work more efficiently or correctly. Purely for planetary imaging we want to reach maximal resolving power for given aperture so bigger pixels don't give shorter exposures or more photons as the optimal f/ratio will just be higher for bigger pixels. For lunar/solar imaging sensor area is key but also having less than maximal sampling can help reduce the effect of light diffraction on your images. If the pixels are to small you will have very little wiggle room with that. Aside of the old IMX174 (that is still good but more expensive) most of planetary cameras will have very small pixels, like 2.4-2.8um. That is a big difference to the 5-6um we were used to. It doesn't change anything in terms of planet size at optimal f/ratio but does change the optimal f/ratio value significantly which not every book or guide accounts for. Also it may be hard to get a fractional multiplication of your focal length with current Barlows and small pixels.

As Sony sensors got smaller and smaller pixels due to various markets demands Sony noticed that for example CCTV market might want to upgrade their existing cameras to something new and 2.4um pixels isn't the best pixels size for their applications.

IMX287 has a massive 6.9μm pixel and with low pixel count it can be used for planets, guiding or some EAA uses. Currently it's available as a Touptek as well as a Point Grey (FLIR) Blackfly S camera. It can reach 520 FPS at 720x540 full frame resolution. You can buy Touptek camera via Aliexpress or some astro shops.

Some other sensors like IMX273, IMX252 and IMX250 have a 3.45 pixel. It's just slightly bigger than the current lineup but a set of Moravian Instruments cameras will bring them to the market soon.

Some astrophotographers went even for ON Semi Python 1300 sensors in FLIR USB3 cameras - this sensor offers 4.8um pixels and good but not the best noise and sensitivity levels. You should be able to find some of their images on Cloudynights. Do note that FireCapture supports FLIR/Point Grey cameras via FlyCap SDK where as FLIR is moving their cameras to Spinnaker SDK which isn't supported yet by FireCapture. Some of Spinnaker only latest FLIR cameras will not work with FireCapture.

Universal cameras

Launch of TEC cooled ZWO ASI cameras (ASI174, ASI178 and alike) was the first beacon of universal Solar System and DS imaging cameras. First iterations did have some problems like with the amp glow but that got sorted out (although some amp glow will always be there, plus some noise patterns). 2.4um pixels on IMX178 however is a limitation when picking a telescope to which it can be connected to for usable DS imaging resolutions. Photographic lenses got more attention - modern ones, like for Canon DSLR as well as older ones, M42 threaded. This sensor isn't that big so most good lenses can provide a quite good image quality. With bigger sensors like the Panasonic or IMX294 the field correction requirement increases and with even bigger IMX071 you have to have something designed for superb image correction on a big imaging circle. I wonder if the release of telescopes such as William Optics RedCat 51 or TS-Optics 61EDPH is a response for such cameras being present on the market?

What's worth noting is that mono Panasonic MN34230ALJ sensor isn't mono in terms of how integrated electronics work. The data is still read in RGB-alike separate groups and if there is some difference in electronics performance between each group then a checkerboard pattern may show up. It was/is present in QHY as well as in ZWO cameras but both vendors released driver/software options to tune the performance of each of those channels to even out the signals and remove the pattern. It seems to work for most users, but not for everyone. You could also use a flat frame or binning to limit or remove such artifacts too. This pattern shows only at high gain short exposure scenarios - so planetary imaging, sometimes lunar and solar. If you want to use a Panasonic MN34230ALJ base mono camera as a universal camera be warned that it may pose a problem in Solar System imaging.

QHY GSense cameras could potentially also be used as universal cameras for Solar System imaging. With their excellent performance it's quite tempting, especially for more dim objects. The prices are however much higher than your average normal sensors based camera.

QHY MiniCAM6F is also worth mentioning out. It's a cooled IMX178 camera with integrated filter wheel for 20 12mm filters. That's quite an insane amount. You can use some odd machine vision filters that tend to be made in such form factor (odd narrowband imaging, lunar petrographic imaging, Jupiter system sodium clouds and more). Just do note that small filter not always will be cheaper (due to smaller production runs of some machine vision filters) and usually will perform much worse than an astronomical filter (like an cheaper Edmund Optics filter). You will have to pay a premium for some odd filters, but at least on average - smaller than for a ~25mm filter. Among astro-related filter companies Chroma could be the one to answer your call for high performance 12mm filters. Not sure about Optolong but it could be an option too. Generic machine vision companies that offer such filters are Edmund Optics and Thorlabs. UQG Optics offers some basic ones.

It's all about sampling

As I mentioned before - sampling is something that is often misunderstand or not accounted for when picking those new cameras. Even some books published in recent years still copy paste planetary imaging from Philips SPC900NC era and hardcore f/20 or f/30 as the go-to f/ratio for planetary imaging.

There are some simplified methods of determining optimal-maximal f/ratio for a given pixel size. First one - multiply pixel size by 5. Second, based on Nyquist Criterion - divide pixel size or pixel diagonal size by 0,275. The first method gives you the upper value of the theoretical optimal/maximal f/ratio for a given pixel size. The second one gives slightly lowered value when pixels size is used and upper one when pixel diagonal is used. You can read is as the upper values are good for targets like Mars, Jupiter or Venus that are bright and can have nice distinct features. Slightly lower f/ratio would be better for dimmer targets like Saturn where slightly lower resolution would help bring exposure time down.

Here is a list of optimal f/ratios for popular sensors:

Sensor Optimal f/ratio
IMX178 (2.4) 12
IMX290 (2.9) 14,5
IMX273, IMX252, IMX250 (3.45) 17,25
ASI120/QHY5L-II, IMX185, IMX224, IMX385 (3.75) 18,75
Panasonic MN34230ALJ (3.8) 19
ICX618 (5.6) 28
IMX287 (6.9) 34,5

Maksutov telescopes are around f/14 and sensors like IMX178MM with 2.4um pixels are already to small (and there are sensors with like 2.2um pixels). There is nothing that could prevent you from using higher f/ratio than optimal - just that the image won't showcase more details as that is all on aperture limit which you already crossed. As you move past the optimal f/ratio the image will get dimmer and dimmer giving worse quality stack so it's not really worth to rush with f/ratio, especially when you see on the live preview that conditions are far from perfect.

Do note that most (Sony) CMOS sensors don't have hardware binning so the binning is software based and doesn't give the full SNR benefit of it. Also as it's software based full frame bin1 and bin2 will have the same frame speed limit.

Barlows for fractional multiplication

It's easy to get 2/3/4/5x Barlows or at least ones labeled as such. The actual multiplication of the focal length depends on the distance of the sensor to the Barlow optics. The longer the distance the higher the multiplication. With 4-element Meade TeleXtenders or TeleVue Powermates multiplication stays more constant and can increase or decrease depending on model (check vendor manual for given model).

To get fractional multiplication you can alter the distance between the sensor and the Barlow. This can work if you need more than 2x but - just use some T2 spars or even variable length spacers and set to desired multiplication. FireCapture focal length estimator can help picking the right distance. But with small pixels you may want less than 2x. This gets bit more tricky.

QHY5-III cameras as well as some Touptek/Altair Astro and USB2 ZWO cameras have a slim 1,25" body. It can move into an eyepiece holder of the Barlow and thus moving the sensor closer to the Barlow optics. This can turn 2x Barlow into something less than 2x.

There are also 1,25" filter threaded Barlows. GSO 2x 1,25" Barlow as well as Baader Q-Barlow can be unscrewed from their housing to reveal a 1,25" filter thread. Screw that into a nosepiece to get something around 1,5x. Use a very short nosepiece and get even less. In all cases playing with the distance is the key to reaching ideal multiplication.

The hardware

Introduction of USB3 and fast CMOS sensors put even more stress on the computer storage and RAM caching capabilities. Right now for best framerate you will need a good SSD. Thankfully SSD prices went down and 512GB SSD should be affordable and handle an imaging session with ease. SATA based SSD will be just fine. It doesn’t have to be M.2 NVMe drive. USB2 base computers like Raspberry Pi in ASIAir and QHY equivalent aren't the best choices for fast Solar System imaging as even with expensive microSD card of large enough size the throughput will be limited and shared across every I/O the device has. For quick imaging with a color camera it may suffice but for a longer run with a mono camera and a filter wheel this could quickly start to show it limitations.

With the introduction of AMD Zen architecture we also got CPUs with much more than 4 cores. That can significantly decrease the stacking time of those large AVIs and SER files recorded with all those shiny large frame small pixel CMOS cameras. 8-core mainstream, soon expected to be up to 16 and even more with Threadripper. If you like to stack quickly check what AMD has to offer. Intel also had to respond to this but in the productivity category their price/performance doesn't seems to be there. Also the RAM increase in the new generation to 16-32GB also helps when operating on large data sets. I'm planning at doing some benchmarking as time permits.

In the realm of telescopes we also got Teleskop Express Cassegrain telescopes from 6" to 10" at f/12. Those seems to be based on GSO RC telescopes and the prices aren't the lowest but still it's a pure mirror design in a quite small OTA. If you can't have a Newtonian and SCT spherochromatism and spherical aberration at short wavelengths (or thermal lag of a closed OTA) is not acceptable that's one of possible solutions. If it’s made by GSO then other local vendors may also have it.

Lack of planets and closing thoughts

For the upcoming years most planets will be quite low or extremely low on the horizon for northern hemisphere photographers. That is also one of factors that may bring some stagnation to the planetary cameras market. You are left up with Moon, Soon, some Uranus and Neptune but those are not the most popular planets anyway. To counter low altitude you can use an atmospheric dispersion corrector. Those devices got quite popular over the years. From an odd and rare items to a normal accessory available in most shops and made by various vendors.

With the advent of CMOS sensors into DS imaging we may see more universal cameras although a DS camera will still need a guider so that universal value can be somewhat questionable (unless it's an older guide camera or big pixel small frame sensor). With Sony releasing replacements for low cost CCTV and similar applications we could expect some cheap cameras undercutting existing cameras but those likely won't offer big frame sizes. CCTV doesn't really need 4-8K frame resolution.

Best cameras from 2016 are still pretty much the best in 2019 with some extra very similar sensors to choose from. Those sensors are starting to hit some limits and major gains aside of price are unlikely in short to mid-term. A new technology, new design would be needed to start offering significant improvements over existing sensors and those take years to design and implement and then a bit more for ironed out gen 2 or even gen 3 of it for maximal value.

RkBlog

Astronomy and Astrophotography, 3 May 2019,

Comment article