The Philips SPC900NC seems a good CCD for CS modification. Here the story I found it in the Net with pictures and comments. No assurance that this works.

In order to send the correct signals to the camera, the capture software must be told to use the correct COM port (from the computer control panel -> system -> device manager) and the correct long exposure settings. In my case I always use K3CCDTools and the serial command signals must be set as shown below.

 

 

 

Unfortunately some USB-serial converter proved not usable to me right out of the box, due to non stable and too low signals at the output pins of the serial port. In order to stabilize these signals and to make the connector compatible to the much wider used parallel port design, one to modify the USB-serial converter to be used with a standard long exposure modified camera with a parallel control connector.

I do not guarantee that ALL USB-serial converters need this stabilizing circuit, but it can be very handy and is not difficult to build.

In the schematic below is the circuit I used to modify the USB-serial converter.

 

On both sides of the camera you find a button. Around that button a silver shield is glued. With a small knife you just lift that off. Repeat at the other side.
Then you find two notches. With a small screwdriver you easily push them a bit down and the case pops open. Repeat at the other side.
Now the only things that hold the two black parts together are the ring around the lens and the USB cable.
The ring around the lens comes off easily when you use a small screwdriver to pry it off.
The cable is connected with a socket. With a small screwdriver you can pop it out.
What you have then is the inner housing, which swivels up and down inside the black housing. To make the swivel action a bit stiff, two rubber bands are placed around the edges. Remove these first. Then you can push down the two lids on either side of the housing to remove it. You can also reverse this and the previous step. So remove the USB cable after you have popped the inner case open.
Then you see this. The two small screws attach both boards to the front of the inner case. The two big screws hold the two circuit boards together.
The boards are connected with a large detachable connector. The S shaped metal construction on the right is the ground connection between the two boards. It is not soldered, so separating the boards is easy. Remove the two big screws and gently pull the boards apart.
Around the CCD is a chamber to make it dark. It has a very strong feel to it, so it will hold the camera with webcam adapter very well.
It effectively shields the light from the white LED (yellow rectangle with D8 label) so for normal. i.e. not long exposure captures, it need not to be removed I think (further tests will prove right or wrong). The microphone is the round cylinder in the bottom left, where we also find our beloved 16510 CCD driver chip. 
Now with the CCD chamber removed.The two big screws hold two metal separators that keep the boards apart.The area around the mounting holes on the top and bottom, as well as where the P/N number is written are pefectly flat and ideal to support the boards inside a new housing.
The inside of the CCD board. The CCD is mounted with the legs through the board, so the standard SC amp-off mod will require to cut one of the legs of the CCD in two on the front side. The two small screws hold the CCD chamber in place.For cooling the CCD, the camera is a bit different from the original ToUcam. The CCD sits on the board without any space to put a metal shim underneath (as was possible with a ToUcam Pro 740 or 840), but you do not have a large chip directly behind the CCD. So indirect cooling by placing a cold finger to the circuit board is a good option.
The inside of the back board. Here we find all the processing electronics. Good old friends like the TDA8787 and 24C04N.
b And on the back board the SAA8116.
Not very clear, but my microscope (Intel Play QX3) reveals that it is indeed a ICX098 BQ CCD. BQ because of the blue edge around the photosensitve area.
An old ToUcam Pro 740 board (the same as a 840) next to the new SPC900NC boards. As you see the physical size is smaller. The new board measures 38mm by 36mm with a maximum thickness (from top of microphone to the cable connector with cable attached on the back) of 25mm, while the old board measures 43mm x 38mm.

 

Opening the camera is a bit more complicated than a ToUcam Pro 740 or 840, because it requires more steps. The benefit of the new case is that there is much room inside it, so you can easily put the SC circuits for long exposure and amp-off mod inside it. This was certainly not the case with the 840. And the design of the two boards give a greater versatility in mounting the camera in a project box. The new boards are nicely symmetrical with respect to the placement of the CCD and mounting holes and has some good large flat area’s to give the construction support. Also around the CCD there is quite a large empty area, making it easier to mount a larger CCD for the SC3 mod.

 

 

First two images, taken with the SPC900NC (left) and ToUcam Pro 840 (right). Images are clickable. Both were taken from the same distance with the standard lens. As you can see the SPC900NC lens gives a much wider field of view, much less barrel distortion and less vignetting than the 840. 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).
The lenses side by side. On the left the SCP900NC lens, on the right the ToUcam 840 lens (with the yellow or silver ring removed).
The general build quality of the SPC900NC lens seems much better. Larger elements, and an exit pupil at the back of the lens that is at least twice as large in diameter than the 840 lens. That explains the lack of vignetting, as more light will pass unobstructed.
A very interesting feature is the manual tunable sharpening of the SPC900NC. Unlike the 740 and 840 which needed a firmware hack to set it in a fixed position, the SPC900NC lets the user adjust it to his needs with a simple slider (beeldverbetering means picture enhancer). The Trimension demo mode lets you see the original and enhanced image is a split image screen.
On the left you see the main tab from the driver settings for the SPC900NC. Here you can adjust the shutter speed, gain, white balance, etc. No big differences from the ToUcam 740 or 840 here. The only thing that is different is the contrast slider. Now you can adjust the contrast in manual mode, something that could only be done in auto mode with the 740 and 840.
Unfortunately numerical indications or stops on the sliders are omitted, so it is more of a guess where you have put the slider. Fortunately WcCtrl still works with the SPC900NC so you can use that to fine tune the parameters.
The SPC900NC introduces a new codec to the list. You can now choose the YUY2 codec, which delivers a higher quality image (and a larger file for the same amount of data).
A 100 frame AVI at 640×480 is about 65Mb with the YUY2 codec, compared to 45Mb with I420 or IYUV.
Unfortunately the camera is no real USB2 device in terms of bandwidth and throughput of data. USB view shows that it is recognized as a USB1.1 device. To test if it is not my USB port that is wrong, I tested a real USB2 device in the same USB port. See the screenshots. Left one is the SPC900NC, right one a USB2 external harddisk. USB view can be downloaded here: http://www.ftdichip.com/Resources/Utilities.htm
   

 For how I did these tests I refer to the explanation on this page.
In the table below you can see the results. As there are no indications in the settings with respect to the shutter speed, I indicate only the steps. 1 is the slider to the left, 11 the slider to the right. The slider sticks at each fixed position in between. There is a strange thing however. For some reason at 5fps the slider not stick to setting 8 and 11. So I could not measure these exposures. I did try my best to measure the shortest exposures, but I do not think they are of any use in astronomy, and the method of measuring the shortest exposures may not be very accurate, because it was very hard to determine the end of the line on the oscilloscope screen. So from setting 8-11 the read-off error may be in excess of 20%.


the measured exposures of the SPC900NC at different exposure settings, framerates and 640×480 resolution.

 

5. Long exposure mod

To prepare for the long exposure mod several measurements must be done to know which signals have to be controlled. To illustrate it I have prepared some pictures. For a great deal, the long exposure mod for the SPC900NC was inspired and based on the work of Steve Chambers. You can read about it here http://www.pmdo.com/wintro.htm

Front of the front board with some important connections.
Same for the back of the front board.
And the inside of the back board.

Because of the two board design with the processing and control components on different boards, there were high hopes that the long exposure signals could be connected to the large black connector, so that the lifting of pins and cutting of traces could be avoided. And indeed it can. See the long exposure circuit layout and pictures below. Please read it carefully, I get many questions which are already answered in the text below.

Furthermore: the modification I describe below is one of the many possible ways to do the long exposure mods. For example you can use the 74HC00 version from Steve Chambers’ website too. There is no ‘best’ mod. They all do the same, but sometimes with different components used.

The circuit I used is based on the 4066 chip, which is a bilateral analogue switch. It acts just as a normal flip switch, by making or breaking a contact on demand. To make the connections as compact and easy as possible the so called ‘dead bug’ approach is taken when connecting.When you turn the chip upside down there are a some legs that must be connected. Because unused inputs on any digital circuit will show erratic behaviour when not connected to a fixed value (i.e. they are not 0 or 1), all unused inputs are tied to the supply voltage of 5V. It is very handy to bend the legs together and to solder them. Also I broke off pin 10 because it can be in the way and because it is an unused output, you do not need it.
When wired together correctly, the view will be as on the left. Two 10k Ohm resistors are soldered to provide pull-up to the control pins, and 5 wires for positive voltage (power of the 4066 chip), ground, LX signal, pad 8/13 and pin 8/13 connections.
The connections to pin 8/13 are made to the large solder points of the black connector and not to the tiny pins of the 16510 on the front side of the front board (you can make them there, but I would strongly advice against it) 
And on the inside of the back board the connections to pad 10 and pad 8/13 are made to the solder points of the black connector there. In the picture below the connections to the ‘outside’ world are shown.To choose between normal and long exposure mode, the connection between pin 10 and pad 10 must be made or broken. In order to do that, you place a small switch between the pin 10 and pad 10 connection points. Also add a 100k Ohm resistor between pin 10 and +5V; it ensures that pin 10 always has a defined signal on it.
You do not want the webcam to control the exposure and shutter anymore, because you are going to control that with the long exposure circuit.
To break the connection of the appropriate pins inside the connector, the pins are bent, as shown in the picture. And to make the connector fit during reassembly, a small amount of plastic is scraped away, so the pins can stick out sideways. In case of total falure or if you need to reassemble the camera to the original state, you simply bend the pins straight as they were before.
The best place to accommodate the circuit will be on the back of the back board. There is plenty of room on top of the SAA8116 chip, or you can place it next to it, whatever you choose.
Finally the power connections and the connection to the LX signal are made. The  +5V and ground are directly derived from the USB connector. +5V is the red wire, and you can solder the wire to the 4066 to the pins of the USB connector where they stick through the board.
The ground connection is soldered to the connection of the black USB wire. There are 3 of them as you can see, but the top black wire is the real USB ground, the other two are shielding wires. They are all connected internally, but as the top black wire is the appropriate one to solder to, I decided to make the connection there. The LX signal that arrives at pin 6 of the 4066 is controlled via the parallel port by software like K3CCDTools. The default pin of the parallel port that delivers the LX signal is pin 2. For protection of the parallel port you can add a 220 Ohm resistor. See the picture above. If you encounter much interference or other noise in the image, it might help to also add a connection from the camera’s USB ground to pin 21 of the parallel port. Normally, that connection is not necessary, because the ground connections are also made inside the computer, so you can leave it out. Please see Steve Chambers’ website about the SC mods and the website or documentation of your capturing software for more details.Alternatively you can use a normal serial or USB-serial port, but you need an extra circuit to provide the correct voltages, see here for the USB-serial adapter, or here for a normal serial port (I have not tested this myself for a normal serial port)
And this is the result: a dark frame of 10 seconds with high gain.
To control the long exposures you need to use special software like Astroart which provides LX control to the camera via the parallel or serial port.
   

Amp-off mod: I have not done the amp-off mod myself, but Paul Hardwick has a very good description on his website. The mod involves cutting one pin of the CCD, so it is not as easy as the long exposure mod, but not too difficult either, provided you work precise and take your time and care.
The amp-off mod circuit description on Paul’s website is slightly different from the standard amp-off mod on Steve Chambers’ website, but the procedure is generally the same.