DIY all-sky cam?

[yesyes]

It's a bit like looking at a planisphere...

[JamesF]

Yes, but somehow a planisphere isn't real, and this is

James

[iwatkins]

I like that. Bit overkill to devote a DMK41 to though

James

It would be overkill indeed. No, the DMK41 is for solar work but thought I could use it at night for all sky stuff. If I get into it, I'll get a camera to dedicate to it.

Cheers

Ian

[Gina]

I've heard a QHY5 works quite well.

[samtheeagle]

I love the idea of these cameras! I might have to try my lx webcam on this project...

[JamesF]

I'd be interested in seeing how that works.

James

[yesyes]

I've done that with an LX modded SPC900 already. See pages 2 to 4 of this thread...

[JamesF]

Have I missed images taken when it was dark, Chris? I can't recall seeing any from your camera.

James

[yesyes]

There are a few starting from here

http://stargazerslounge.com/diy-astronomer/160629-diy-all-sky-cam-5.html#post2030051

Not exactly as good as I expected, but still usable as a cheap option...

[samtheeagle]

Well I don't have a fish-eye lens or a dome to house it in yet, but my LX SPC900 seems to be doing the business with the standard lens I just popped out with my laptop and the camera and managed to get the following screen grabs from the feed:

~2.5 second exposure

~8 second exposure

I don't know whether the fish-eye lens will reduce the sensitivity? I guess it will have an some kind of impact as the sensor has to try and record a lot more light from a (much) bigger FOV. It's got to be worth a try though, I'm very pleased with this initial test

[samtheeagle]

Is there an optics expert in the house? I've just been looking on ebay at these things and there are soooooo many different options! So, I think from what I have read the smaller sensor in the SPC900 means that I wont get as full a FOV as the lens spec suggests, so I should probably aim for as wide as possible right? One other thing I'm seeing is mention of a lens M.O.D. value, which a bit of googling suggests means multi-order diffractive, so is a bigger or a smaller value for that best?

I've narrowed it down to 3 now I think...

1.9mm Megapixel Board Lens for CCTV Cameras - Security Cameras | eBay

1.9mm Board Lens for CCTV Cameras - Security Cameras | eBay

2.1mm CCTV Board Lens for CCTV Cameras - Security Cameras | eBay

The 3rd one is there due to the lower M.O.D. value, which may or may not be a good thing? If it is a good thing, does it outweigh the narrower 150 degree FOV?

[Gina]

I think the wide angle board lenses I've got are 1.9mm FL and f2 - they give about 90 degrees FOV with 1/4 inch sensors. Nowhere near fisheye. I haven't tried one on my QHY5 yet. I'll probably try that sometime though I think the few 1.9mm lenses I have are all on our CCTV cameras. I got mine from Maplin but they stopped doing the wide angle board lenses some time ago. Thank you for those links - could be useful

As for sensitivity - the shorter the focal length the more light they let in though I believe the standard relationship goes wrong with the ultra-wide fisheye lenses. The f number gives the amount of light they let in - so you go by that for exposure values.

[samtheeagle]

Yes, F2 is super fast, what I'm thinking is that with the wide angle lens you get more FOV, but that means that things like stars will be much smaller in the image, as so much more data is squashed onto the chip... That is hard to explain! I know what I mean, I hope you catch my drift?

Still, 90 degrees is going to give me most of the sky Now if someone could explain what the M.O.D. value is all about? Then again, if your 1.9mm is giving ~90 degrees I'm pretty sure I'll want to get the widest FOV lens possible. The other difference between the two 1.9mm ones is the physical size, presumably the larger one will have superior light gathering?

[Gina]

The only difference I can see is that the more expensive one is described as "megapixel" which presumably means it's better quality - higher resolution. They are both f2 so the same light gathering. Standard board cameras are rather poor resolution generally - hence cheap. OK for just seeing if someone is about or where your livestock is but not much more. I found them quite inadequate for the main surveillance on our main entrance gates and yard. I'm not really sure what quality you'd want for a sky cam though.

[JamesF]

The standard relationships may well go wrong with very short focal lengths because many of the formulae for plate scale etc. are based somewhere along the line on the approximation that "tan(x) = x" for small values of x. In the case of these lenses we may well be talking about values of x (angles) for which that isn't a valid assumption.

However, let's assume for a moment that it is. Given a camera sensor with 5.6um pixels and a focal length of 2mm, i think that would mean that each pixel covers a smidge under 10 arc minutes of sky. The chances of picking up stars on a camera without an exceptionally sensitive sensor don't look too good when you put it that way.

Out of interest, the entire field of view of a 640 pixel wide sensor would be just over 100 degrees. If you can drop down to a 1.8mm focal length then you'd end up with just over ten arcseconds per pixel and a field of view of 114 degrees.

James

[samtheeagle]

Thanks for crunching the numbers Pretty much as I suspected. Still, I think it's got to be worth a tenner, but perhaps I'll be better served by not trying to max out the FOV? Something wider than the default, but not OTT.

Just had another thought... A bit of a deviation from the all-sky cam idea, but maybe a pan & tilt head could be fun? It could be automated to scan the skies with whatever number of steps is required for the FOV available.

[JamesF]

Does anyone have any idea what sort of focal lengths are usually used for webcams? Could be useful to know.

James

[samtheeagle]

The lens of the 840 is a 6mm F/2.0, while the SPC900NC lens is claimed to be a F/2.8 and a measured 4,5mm focal length (I have not found any reference from Philips with respect to the focal length).

Taken from here: Philips SPC900NC uncovered

Not quite sure where that leave me with an SPC880 flashed to be a 900

[samtheeagle]

I love the astrometry webservice! I though it'd be nifty to see what the FOV of the standard webcam lens is, so I fired off a cleaned up version of one of my test captures from last night and this is the result:

The calculated FOV is 47.49 35.67 degrees, and the pixel scale is 267.54 arcsec/pixel.

[JamesF]

Given a 4.5mm focal length lens I'd calculate the figures for an SPC900 as 256.69 arcsec/pixel and 45.63 x 34.22 degrees field of view. I think I can live with that sort of margin of error given we can't guarantee how close the tolerances are on the lens.

The exposure time is clearly what makes this work. At this image scale Mars at its best would only light 1/10th of a pixel and Saturn even less.

James

[iwatkins]

Hi All,

Tried my setup Saturday evening (in anger). Tripod ball head, nailed to a wooden broom handle in turn screwed to the fence for a mount. On top of this was the cheap dome enclosure plus a mono DMK41 with a 2.1mm C mount lens attached.

USB cable running to the shed with my old laptop doing 10 second captures all night. Seems to work well, just need to tweak a little, but looks to have captured a Lyrid OK.

Cheers

Ian

1200x900 (from original 1280x960) image

[Gina]

That's pretty good

[DrRobin]

Hi Ian,

Nice picture, is it a single frame of 10 seconds? If it is the noise is very low, but then it is a DMK. I have been testing something similar with an Sony exView board camera, but integrating 250-500 frames to get one image. I certainly get more noise even after dark frame subtraction, still it costs a lot less than a DMK.

Regards

Robin

[iwatkins]

Thanks Robin,

Yes, 10 sec single frame and noise is pretty low. I did it without noise reduction, must play more with that. DMK is my solar imaging camera, so it can be used at night too now

I'm looking into building a full time all sky camera now, also looking at the Sony based board cameras.

Cheers

Ian

[Gina]

I have some Sony based surveillance cameras but even the best only shows 2 or 3 of the brightest stars. I guess the exposure is just too short at the TV frame rate. I think you'd do better with an LX modded SPC900NC but I haven't got round to trying mine yet. Another problem is the small image sensors and getting a short enough focal length lens.