Seeking advice for building fireball station.

[Anton Norup Sorensen]

Dear fireball hunters,

I am member of a small group that will try to set up a network of meteor
cameras in Denmark. The cameras are intended to be distributed to
high-schools, where they will serve as educational tools as well as provide
data on fireballs.
So far, funding for a prototype station seems certain, and when this is
working well, we are confident that funding for about five more stations can be obtained.

At this time, I have four meteor webcams running, located at two sites with 50km between.
However, there are several problems with my type of station that makes it unsuitable for distributing to high-schools.

I would like to hear your recommendations for cameras, optics, processing etc.

I hope this project will be of benefit to German observers also by sharing
observation data - already there are some examples of fireballs observed
from both countries.

Thanks,
Anton Norup Sørensen

[Mark Vornhusen]

Dear Anton,
recently I made some improvements of my fireball detection system.

1. I use a small Linutop computer ( http://www.linutop.com ) with no moving parts (no harddisk, no fan) and only 6W power consumption instead of a tower PC.

2. Rewritten my meteor detection software to run under Linux operating system (was Windows before).

3. The meteor detection software triggers the video quad splitter via RS232 interface to switch to the camera with the meteor detection to full screen. So I have a high resolution live video of the meteor (720x576 pixel instead of 360x288 pixel). The software can handle color or b/w cameras with the same performance.

Four b/w video cameras are connected to the quad splitter. The video signal from the quad goes into a USB capture box ( Pinnacle Dazzle DVC 100) and the capture box is connected to the Linutop computer. Currently I have placed the cameras at windows in all four directions of my house and long video cables connect them with the quad splitter.

I want to build a compact prototype where 4 cameras, the quad splitter and the Linutop computer fit into a single outdoor housing. Then one can place the camera system on top of a roof and the only connections needed are power and network cable. At the moment I'm looking for a more compact replacement for the quad splitter ( AVC704, http://www.rfconcepts.co.uk/quads.htm ), which is too large. There are some USB capture devices with 4 inputs ( http://www.security-cameras-cctv.com/ch ... -1566.html ) but I don't think they will work under Linux. If I find one with a Linux driver available then I could switch the channel easily with the meteor software without the need of a RS232 cable.

I run the meteor software day and night. I noticed I get a lot of false detections from raindrops on the window at day. Also I get false detections from cars with headlights. For this reason I have to make some more improvements of the meteor software, especially a feature to edit a mask area with no sensitivity for meteor detections ( mask the street). False detections from raindrops and lightning can be reduced by counting the number of detections in a defined time interval. When there are more than 5 in 10 minutes I can stop the detection temporary.

Because the computer has no harddisk, all captured videos must be transfered to an external server via ftp soon after the detection. It would be nice to have a webserver where everybody can see the meteor videos minutes after a detection, especially when there is a network of many of this kind of meteor cameras. Of course it would also be possible to connect an external harddisk to the Linutop with hundreds of GB space and store all videos on it for a long period of time.

I currently use low light 1/3" b/w cameras:

CS-Mont: http://www.rfconcepts.co.uk/2046xai.htm

and

SK-1004X (32 x 32mm, Sony EXview 0,003 Lux board camera with 3.6mm lens)

For fireballs the cheap board mount type is sufficient. However, there are not many stars visible in live video with this kind of camera.


Mark

[Anton Norup Sorensen]

Thanks for your quick and detailed recommendation, Mark.

I am eager to get away from webcams, and will try to get the USB capture box and use it with a video camera currently collecting dust. I suppose that your B/W video camera performs better in the dark than a typical webcam, although you say it is not so good?

A few stars in the field is nice to have, and a high-school teacher asked for better performance here, so the camera could be used more generally for astronomy, rather than just fireballs. I would be interested to hear about reasonably priced videos cameras that perform well for stars.

It is encouraging that you can use long (how long?) video cables. The limitation of short USB cables is frustrating.

I would prefer to avoid the four cameras and quad splitter - it reminds me too much of the hassle I currently have with running two webcams on one computer.
In stead, I am considering a single downward-looking camera with an upward facing convex mirror. I often have trouble translating the image position to altitude/azimuth, and expect this to be simpler with a all-sky mirror camera.

I also get a lot of false detections, perhaps ten for every meteor. A mask as you describe is implemented, and I have different detection thresholds for day and night. I try to make an adaptive threshold by monitoring the noise level and having a threshold that is a constant multiplied by this level. This adjusts with some succes to rain and drifting clouds, but is somewhat CPU intensive.

When your more compact station is complete, do you consider making more copies for other stations, or provide detailed building instructions?

Kind regards
Anton

[Mark Vornhusen]

I suppose that your B/W video camera performs better in the dark than a typical webcam, although you say it is not so good?

Yes, it performs much better than a webcam. I have never used USB webcams like Philips ToUcam, but I guess there are no stars at all visible in live video (25fps). With the Sony Exview board mount camera and 3.8mm lens you have about 2-3 stars visible in live video. With the CS-Mount version and f0.8/3.8mm lens there are more than twice as many stars in the live video. For example I see the "W" of Cassiopeia in the live video. But much better cameras exist. If you want maximum performance use cameras with 1/2" CCD, for example Watec 902H2.

A few stars in the field is nice to have, and a high-school teacher asked for better performance here, so the camera could be used more generally for astronomy, rather than just fireballs. I would be interested to hear about reasonably priced videos cameras that perform well for stars.

One can dramatically increase the number of stars with frame integration video or long time exposure photo. But in this case you can't capture live video of the fireball and you loose the speed information of the fireball. I have a Watec 120N camera which shows hundreds of stars in a two second exposure. But I found that this camera is not usable to monitor the sky automatically. You have to adjust the exposure time depending on the moon phase and also the camera does not work in daylight. I also thought about digital SLR. But it takes about 3 seconds to download the image after each photo. So you can't monitor the sky nonstop. Also the maximum exposure time is limited to 30s with the gphoto software. I guess the lifetime of the shutter is limited to some 100.000 exposures and you reach this number in a few months when capturing all the night.

It is encouraging that you can use long (how long?) video cables. The limitation of short USB cables is frustrating.

The length of the video cables from the quad to the cameras is about 10-15m each. Thats no problem. You can use 50m or even 100m without significant loss of signal quality.

I would prefer to avoid the four cameras and quad splitter - it reminds me too much of the hassle I currently have with running two webcams on one computer.

Four cameras on one quad splitter is not a problem. In fact it operates like one camera and I need only one capture device. I found it is not possible to use two USB capture devices simultanously on one computer because of limited bandwidth of the USB bus. But it is possible to use one USB capture device and a PCI capture card simultanously with one computer.

In stead, I am considering a single downward-looking camera with an upward facing convex mirror. I often have trouble translating the image position to altitude/azimuth, and expect this to be simpler with a all-sky mirror camera.

Hmm, I don't think so. The mirror constructions are to complicated, hard to install and it is difficult to find a good mirror. One has to clean the mirror often from dust and snow or use a heater. Also the image resolution is very low when using only one video camera above a convex mirror. I want to get maximum image resolution and live video. I found switching between four cameras with the quad splitter gives the best performance. I also tried a combination of one video camera for live video and a digital SLR for high resolution. When there was a meteor detection the camera shutter was released automatically. But there is a lack of about 2 seconds from the command by the computer to a shutter release of the Nikon D70 I used - to slow for meteors.

When your more compact station is complete, do you consider making more copies for other stations, or provide detailed building instructions?

If everything works fine then I will likely set up more stations by myself and give detailed instructions how to set it up for others who are interested in using the system.

Mark

[Anton Norup Sorensen]

A very small step forward since last post:
After getting hold of a Pinnacle Dazzle DVC 100 and finding a patched version of the v4l library, I have the frame grabber running fine on Fedora Core 6 and Ubuntu 7.1 Linux installations.
A borrowed video camera showed disappointing performance - the faintest signals seems to get clipped, so the sky background appears completely dark. My webcams with poorer optics perform better than this video camera. See the example with 1 hour integration below.
I think that the clipping is caused by the camera and not by the frame grabber, and hope to get this confirmed.


The moon peeking through clouds during 1 hour. Left: Video; right: webcam.

[Anton Norup Sorensen]

I currently use low light 1/3" b/w cameras:
CS-Mont: http://www.rfconcepts.co.uk/2046xai.htm
and
SK-1004X (32 x 32mm, Sony EXview 0,003 Lux board camera with 3.6mm lens)

Dear Mark,

Can you give an estimate of the minimum back focal distance for a lens to be used with the 2046XAI camera? I am considering fisheye-like lenses and these mostly have a very short BFD that may not allow room for e.g. a built-in window in front of the CCD. In particular, I am interested in using the DSL-215A with 4.31mm BFD found here:
http://www.optics-online.com/dsl_fisheye.asp

Thanks,
Anton

[Mark Vornhusen]

Dear Anton,
I don't think the minimum back focal distance is a problem.

This is a board mount lens. So I would use a cheap SK-1004x camera instead of the CS-Mount version. The board mount version has an even lower lux rating and will show more stars than the CS-Mount camera. The only reason I use the CS-Mount camera is, that I want to use very fast lenses (f0.75) and these lenses have all CS-Mount. Also I want to use the camera at day and night so I need auto iris lenses. I guess you can use the fisheye lens day and night with good image quality and SK-1004X camera, because the f-ratio is not very fast.

Mark

[Anton Norup Sorensen]

Just wanted to report a little progress:
Now a CB1004X with a Sunex F=1.55mm f/2.0 fisheye lens has been built into a simple outdoor house made from a dummy security camera dome and some plastic tube.
It was installed outdoor this weekend and has already been thoroughly tested for watertightness


The lens is very good, but the dome distorts the image somewhat. Faint star performance is not as good as hoped for, maybe only one magnitude better than webcams. Maybe it will be better when it gets darker and colder. So far only Arcturus and Vega are easily visible, as seen below:



But at least the sky coverage will be good, when installed in a place with less obstructions.

[Mark Vornhusen]

Hallo,
vielleicht reicht schon eine einfache Canon Sucherkamera als Feuerkugel-Überwachungsstation. Man kann diese Kameras so programmieren, dass sie ein Foto aufnehmen, wenn sich etwas im Bild bewegt. Vielleicht geht das auch für die Videofunktion. Jedenfalls finde ich das eine spannende Sache.

http://www.howtofixcomputers.com/forums ... 71388.html

http://chdk.wikia.com/wiki/CHDK



Grüsse,
Mark