By detector I mean the webcam minus its lens (which would come under optics, cf. Optics). My second webcam is a Philips ToUcam Pro which I may use at VGA size (640x480 pixel) or at SIF size (320x240 pixel). My first webcam is a Logitech QuickCam VC for parallel port. Its size is SIF. I treat the two sizes of the ToUcam as different detectors here. The following table shows the characteristics of these three detectors:
|ToUcam Pro VGA||QuickCam VC||ToUcam Pro SIF|
|pixels||640 x 480||320 x 240||320 x 240|
|pixel size||5.6 µm||8.2 µm x 7.6 µm||11.2 µm|
|detector size||3.6 mm x 2.7 mm||2.6 mm x 1.8 mm||3.6 mm x 2.7 mm|
|diagonal size||4.5 mm||3.2 mm||4.5 mm|
A major difference between 35 mm film and a CCD camera (webcam or proper amateur astronomer CCD camera) is that the CCD is almost ten times smaller. The resolution is also slightly better at 5 or 10 µm.
It is necessary to mention Steve Chambers and the webcam modification he pioneered. See his web site on the matter at http://www.pmdo.com/wguide.htm. Many webcams - including Philips Vesta and ToUcam models but excluding the Logitech QuickCam VC - use the same integrated circuit chip to control the image readout from the detector. Steve has rewired several of the contacts of this chip with the effect that electrons are collected in the detector as long as the astronomer likes, controlled by a switch or a software-controlled wire to the parallel or serial port. Equivalent modifications for the QuickCam VC are described by Robert Pigott and by Michael Mommert.
I have not done this to my ToUcam Pro. Hence I can say nothing in this document about how that is done, how it works, and what you do to acquire and reduce images. The modification is for deep sky imaging, and it seems to have a dramatic effect. It has become so popular among astro webcam'ers that they call their modified webcams Something or Other SC for "modified à la Steve Chambers".
|The Philips ToUcam Pro out of the box.|
My second webcam is a Philips ToUcam Pro. Like virtually all webcams these days it connects to the USB (Universal Serial Bus). The Philips Vesta Pro and ToUcam Pro are said to have good sensitivity. They also feature more pixels (640x480 or VGA size) on the detector, and they are supported by Linux. The Philips webcams are not so easy to find in this country. So I had to go to a mail order company, which brought the price to a bit over £ 70.
I use the ToUcam Pro at VGA size for bright extended objects, because here its advantages count and its disadvantages don't matter:
- It is well supported under Linux.
- It can do colour with good results.
- The compression used in transfer across the USB does not cause significant artefacts for this sort of object.
- The size of the pixels is ideal for the optics that I use for these objects.
- The time required to take enough data is short enough.
For star fields this detector is not suitable. The bright pointlike objects are not well represented in the images, they have bad compression artefacts in the wings of the point sources.
To look up the specification of the ToUcam Pro go to http://www.pc-cameras.philips.com, then follow the Manuals : ToUcam Pro : Techical Specification.
|The Logitech QuickCam VC out of the box.|
My first webcam is a Logitech QuickCam VC for the parallel port. I went to some length to find this in the catalogue of a mail order company and paid a premium for that: Instead of £ 50 for the USB version available on the high street, I had to pay almost £ 70. The hope was that a parallel port camera might be supported not only by Windows systems, but also by Linux systems. In fact the support on Linux has come about since then and is probably better for the USB version than the PP version.
Parallel port cameras also connect to the PS/2 or old-style keyboard port, which is their source of power. The QuickCam VC comes with all the bits to plug in a keyboard as well as the webcam.
This webcam has 320x240 pixel (SIF size). It is better for pointlike as well as faint objects, so I would this for star fields and deep sky objects. Its advantages over the ToUcam Pro at VGA size are:
- Compression is not used in the data transfer from the webcam. Hence even stars do not cause compression artefacts in the images.
- The frame exposure can be up to an amazing 1.8 s. This means that per accumulated stack exposure time 40 times fewer frames need to be collected. Hence the noise level is lower.
- Virtually no time is lost between frames. Other webcams like the ToUcam Pro have a longest exposure that is five times less than the frame rate, leading ot a loss of 80 per cent of the elapsed time.
There are also disadvantages:
- Support under Linux is not quite as good as for the Philips webcams.
- The webcam has fewer pixels, reducing either the field of view or the resolving power.
- It is not very good at colour, which does not matter for grey deep sky imaging.
- The pixels are not square, but slightly smaller vertically than horizontally. This, too, does not matter very much.
The QuickCam VC has the following specification, as collated from a document "QuickCam Technical Information", which used to be on http://www.logitech.com, and from Michel Collart's AstroCam web site (http://www.astrocam.org/qcamvcUS.htm):
- CCD, 352x288 pixels, CIF sensor format
- 24 bit colour
- up to 30 frames per second AVI
- still formats BMP, TIFF, JPG
- CCD 362x291 pixels in total, diagonal size 0.2 inch
- operation at 0°C to 45°C, storage at -10°C to 70°C
- humidity 0 to 90 per cent non condensing
- lens is f/2.0 with 46° field of view
- shutter speed 0.1 ms to 20 ms
- pixel size 8.2 µm x 3.8 µm
- CCD size 2.9 mm x 2.2 mm
- sensibility 170 mV/lux
The detector diagonal quoted by Logitech is an overstatement, Collart's numbers make more sense. In the vertical direction there are probably two half-height pixels for different colours. This still leaves the vertical pixel size 7 per cent smaller than the horizontal pixel size. So circles will not be exact circles in the image output. I confirmed this experimentally: Images of Jupiter appear circular, indicating that the pixels deviate from squares by the same percentage as Jupiter deviates from being spherical (0.065). The Jupiter observation - assuming the focal length is known - also confirms the size of the pixels to be as listed by Collart.
This detector is in fact not very important to me. An advantage over the ToUcam Pro at VGA size is that at 5 Hz frame rate there is no compression on the USB and the image quality in star fields is indeed better. However, there are still artefacts around the brightest stars.
Since I also have the QuickCam VC working under Linux it is unlikely that I will use the ToUcam Pro at SIF size. Although sometimes it might be an advantage to use the smaller size on Sun, Moon and planets simply for reasons of data volume.
Under Windows the ToUcam Pro SIF has the additional disadvantage that 80 per cent of the time is lost between exposures. The eye-opener for me was when I imaged a star field with the ToUcam Pro (VGA) and compared the limiting magnitude with that of another star field taken with the QuickCam VC. The exposure times were broadly similar at 600 s and 800 s resp. and the limiting magnitudes were in proportion to that (11.2 mag and 11.4 mag resp. at f = 50 mm, f/1.8). Similar time gives similar result, right? Wrong! The QuickCam VC can collect the exposure time in under 15 min, while the ToUcam takes 50 min. The reason is the limited frame rate combined with the limited maximum frame exposure time.
Things look a bit better for the ToUcam when it is run at SIF size. Four VGA pixels become one SIF pixel, which reduces the noise twofold. Where the QuickCam VC exposes close to 100 per cent of the elapsed time, the ToUcam Pro exposes only 20 per cent of the time. However, at SIF size it should achieve the same limiting magnitude two times faster, reducing the time disadvantage from fivefold to 2.5-fold.
Under Linux it appears that the qastrocam application, which is written for Philips Vesta and ToUcam webcams, manages to increase the exposure time to 0.2 s, the same as the frame rate. So the time loss argument does not apply.
Copyright © 2003 Horst Meyerdierks
$Id: detector.shtml,v 3.3 2004/02/21 18:13:39 hme Exp $