After my "freeze the camera" experiment, I decided to purchase a Peltier device to play with.  These devices are semiconductors of some kind that exhibit a cold side and a hot side when powered.  They are assembled in bunches and bonded with two plates of ceramic material, one on each side to make a sandwich.  Two wires are soldered on and are used to supply a DC voltage.  I found a medium-sized one at Fry's Electronics for $20.

I tore apart the first camera box I built, and proceeded to mount the back of the ToUCam circuit board against the Peltier, in the hopes that the coldness could penetrate the a large but thin integrated device, the circuit board, and finally the CCD chip.  I found a plastic box at Radio Shack, cut a hole in the back, and mounted the circuit board, Peltier, and a heat sink/fan to it.  Then I cut a hole and mounted the T-adapter to the lid and stuffed the box with insulation.  Here's the result:

Needless to say, this wasn't entirely successful, although it did look nice!  For one, the CCD did not get as cold as I had hoped.  In a 78 deg. F room, the CCD only managed to get down to around 45 deg. F, for a delta of 33 deg. F.  For another thing, the CCD was exposed to the air, and collected a fair bit of condensation (which made for some interesting pictures!).  Finally, the plastic lid that the T-adapter was screwed on to was not too rigid, and the weight of the camera would cause the lid to flex and made for some nasty focusing.

So, with that experience behind me, I decided to try again.  This time I found a nice aluminum-alloy box for $10 at Fry's that is nice and rigid.  For the lackluster coldness problem, I decided I'd get brave and desolder the tiny CCD from the ToUCam and mount it proper-like on a "cold finger".  Finally, to keep the condensation away, I thought I'd use a small plastic makeup jar to enclose the CCD.  What follows are pictures and drawings that show how I constructed this new and improved camera box.

The first scary bit was to desolder the CCD chip, which is mounted in a tiny chip case and surface mounted to the circuit board with even tinier little leads.  I used a small 15W iron to quickly touch a lead with one hand, while using a small probe in the other hand to bend the lead up slightly when the solder melted.  After 14 leads, it was done - not too bad!  To make dealing with the CCD and the CCD-deprived circuit board easier, I took an ordinary 28-pin chip socket and carefully cut it in half - one half for the CCD and the other half was split in half again (length-wise) to use on the board.  I used two small bits of ribbon cable to solder the two halves of the halved chip socket to the circuit board, and soldered the CCD to the underside of the other half.  The CCD chip case conveniently press-fit into the chip socket with a bit of intricate sanding on the interior of the chip socket:

After connecting things back together again for a quick test, I was relieved to find that I had not damaged anything.  However, for the test I used about a foot of ribbon cable between the CCD and board (duhhhh), and noticed significant interference in the images.  I thought "oh man", but after trying a smaller length of cable, the interference disappeared (whew!)  I did notice that a small bit of the same interference was present with the camera gain turned up all the way, but was not present with more realistic gain settings.

The cold finger assembly was next.  The finger is formed by using two pieces of medium thickness angle aluminum placed back to back.  They are secured to each other with a pair of screws and nuts in predrilled holes, after coating the contact surfaces with heat sink compound.  The bottom surface is ground flat by dragging it over a file, and a little nook cut in the top edge for the CCD to sit in.  Using two small screws and washers in predrilled holes and some more heat sink compound, the CCD assembly is mounted to the finger, where the washers "grip" the sides of the chip socket, which in turn press the CCD chip firmly to the finger.  Finally, the Peltier is pasted on the bottom with more heat sink compound:

The small plastic makeup jar is prepared for use as an airtight CCD chamber by first cutting three slots in the screw-top lid.  A large middle slot fits tightly down around the cold finger while the two smaller side slots accommodate the two ribbon cables used to connect the CCD (I had to partially disassemble the cold finger assembly to get the jar lid placed on the cold finger).  For the jar body, which is the top of the CCD chamber, an almost 1" diameter hole is cut in the bottom.  A 1"x1" piece of anti-reflective glass (obtained from Edmund Optics for $11) is used for the window, and attached to the jar body using four screws, washers, rubber grommets and nuts.  In the space between the jar bottom and the glass, one of those blue Lego "siliconey" rubber bands coated with a bit of silicone is used to form an airtight seal:

The next diagrams show how the Peltier/cold finger/CCD chamber is attached to the heat sink and camera box lid (I decided to use the lid here and not the box to make it easier to fiddle with the components).  The heat sink is isolated from the box lid with a rectangular piece of blue PVC plastic the same size as the heat sink, with a rectangular hole the same size as the hole cut into the box lid (these holes are a bit larger than the sides of the Peltier/cold finger, to allow adjustment).  The box lid and spacer is bolted to the heat sink with four sheetmetal screws and washers into predrilled holes (not shown).  The Peltier/cold finger/CCD chamber is placed in the hole with some heat sink compound and secured to the lid/heat sink assembly with two sheetmetal screws and large washers, along with two medium-thick strips of aluminum.  The screws and large washers press down on the aluminum strips which in turn press down on the cold finger base to keep it and the Peltier tight against the heat sink.  This method allows the position of the entire Peltier/cold finger/CCD assembly to be adjusted:

Enough with the drawings!  Here is a real picture of the guts.  Notice the short lengths of ribbon cable that snake from under the white makeup jar lid to the single-row sockets attached to the ToUCam circuit board.  The circuit board is not really attached other than by the ribbon cables, and is firm enough.  The board is wrapped with some electrical tape to avoid problems with it touching the screw head and aluminum strips under it.

These pictures show some of the details around the base of the cold finger and the way the ribbon cables are hooked up (I used silicone around the chip socket contacts for insulation - a bit messy and not really necessary):

Here are some views inside the CCD chamber.  I used some messy silicone around the ribbon cables and cold finger to make the base airtight.  Note how the ends of the ribbon cables bend and plug into the top of the chip socket:

A few more pictures for your viewing enjoyment.  Another base shot, and one showing the PVC spacer between the heat sink and the box lid:

The final struggle is to prepare the camera box body.  A circular hole with three notches to accommodate the T-adapter is cut, and two holes for the Peltier/fan power terminals are drilled.  Finally, it's ready to connect the wires and close it all up:

Here's the obligatory cold finger temperature chart.  Not too shabby now - it gets below freezing in under five minutes, and achieves a final delta of almost 59 deg. F after 20 minutes.  I wonder if insulation would help any - I still need to try that.

Some thoughts: 

• The CCD chamber has an additional benefit I hadn't planned on - once the dust is cleared off the CCD and the chamber lid is screwed in place, there's no more 5 minutes of work trying to get rid of the dust specks before each session!  Dust specks on the chamber window aren't nearly as irritating as on the CCD itself.

• The chamber window (an anti-reflective-coated glass with transmittance greater than 99%) seems to do some filtering of the red end of the spectrum, which has the balancing effect of allowing more greens and blues into the images.

• I've not really tried to image star fields yet, having concentrated on the planets so far.  Compared to the non-cooled version of the ToUCam camera on the planets, I've noticed that the reduced noise levels allow increased gain levels, which in turn allow reduced exposure times.  Reducing the exposure times seems to reduce the "shimmers" that can blur and distort images.

• I am attempting to acquire an old grayscale Connectix QuickCam that can, through 3rd party software, support multi-second exposures.  This type of camera would probably benefit more from cooling, and when I do find one, I'll put the ToUCam back into my original "Cheap-O Cam" case and use this cooled case for the QuickCam.  Stay tuned, and clear skies!

Update - I've since replaced the guts of this monster with the guts of an old grayscale Connectix QuickCam, which basically looks the same as in these pictures, except there are fewer pins on the CCD than on the ToUCam Pro.  I added the "disable anti-blooming" switch described elsewhere that allows the QuickCam much more sensitivity for capturing the faint fuzzies.  This modification is performed by simply cutting the yellow wire from the cable to the board, and inserting a SPST switch (mounted to the case).  When switched on (yellow wire connected to board), the camera functions normally; when switch off (yellow wire disconnected), the anti-blooming feature of the CCD is disabled.  As for the cooling vs. non-cooling, it really does make a big difference with the QuickCam!  There are probably 50 or so "hot pixels" on my QuickCam's CCD, and with the Peltier and fan running, they virtually disappear.  The rest of the pixels also benefit by not producing as light an image for a given exposure time, thus allowing longer exposure times.