This part of the forum has been pretty quiet of late. I thought I'd inject some life into it. I had a couple of old Canon DSLRs lying around so I decided to do some radical mods. This is informed largely by two excellent threads in Stargazerslounge, linked below:
https://stargazerslounge.com/topic/2173 ... ra/?page=1
and
https://stargazerslounge.com/topic/1663 ... ix/?page=1
The idea is to de-Bayer a sensor, and then perform a radical modification to fit the camera into a sealed box for easy moisture management. This will allow the sensor to be cooled without condensation. Hopefully. The following thread will chronicle the work I have done and will embark on. Stay tuned.
DIY Cooled Astronomy Camera - In progress
Re: DIY Cooled Astronomy Camera - In progress
Step 1: Removing Sensor for De-Bayering
In order to remove the sensor from the camera to perform the debayering, I consulted the excellent guide here: http://dslrmodifications.com/rebelmod450d1.html
This allows one to perform a full-spectrum mod (if you want a dedicated astro camera) by simply removing ALL the filter elements from the optical train. It's extremely easy to do. All it takes is a set of tools (screwdrivers, tweezers, etc) and about 30 mins of your time. You'll end up with the following:
![Image](https://farm2.staticflickr.com/1691/25671444254_866efc163a_m.jpg)
As it turns out, Bayer layer removal is an inherently hazardous activity. I should know. I went through 3 sensors before getting a half-decent one. Of course, one could send the camera in for professional debayering (~ US$400 including shipping), but I am adventurous that way. As it turns out, many of these old, obsolete cameras have relatively cheap sensors that can be had off auction sites for US$40-60, depending on your luck. So on balance I'm still out ahead by some 200 bucks. Not too shabby.
In order to remove the sensor from the camera to perform the debayering, I consulted the excellent guide here: http://dslrmodifications.com/rebelmod450d1.html
This allows one to perform a full-spectrum mod (if you want a dedicated astro camera) by simply removing ALL the filter elements from the optical train. It's extremely easy to do. All it takes is a set of tools (screwdrivers, tweezers, etc) and about 30 mins of your time. You'll end up with the following:
![Image](https://farm2.staticflickr.com/1691/25671444254_866efc163a_m.jpg)
As it turns out, Bayer layer removal is an inherently hazardous activity. I should know. I went through 3 sensors before getting a half-decent one. Of course, one could send the camera in for professional debayering (~ US$400 including shipping), but I am adventurous that way. As it turns out, many of these old, obsolete cameras have relatively cheap sensors that can be had off auction sites for US$40-60, depending on your luck. So on balance I'm still out ahead by some 200 bucks. Not too shabby.
Re: DIY Cooled Astronomy Camera - In progress
Step 2: Opening up Sensor Cover Glass
The next step you have to do is to remove the sensor cover glass from the sensor assembly. Every sensor that you see in the market comes packaged so that there is a cover glass over the actual silicon. This helps to protect the array from damage, as well as the very delicate wire-bonds that interfaces the silicon to the chip packaging and ultimately the pins.
![Image](https://farm2.staticflickr.com/1603/25673574503_2f6fd16b4d_m.jpg)
Here's the sensor. It's already been opened once and loosely re-attached. That's why the borders around the sensor glass look funky. To remove this glass, you can either pry it open (bad idea) or use a butane torch, like this one from Dremel:
![Image](https://farm2.staticflickr.com/1606/25673741713_da92760da7_m.jpg)
I started off using an Xacto knife to try to pry open the glass. However, the lever action is too great and you will inevitably end up with a chipped or cracked glass. This is very dangerous because there are, as I mentioned, some very fine wires bonded to the silicon chip. An errant glass shard can break one of these wires, ensuring you end up with a dead sensor. Instead, what you do is run a naked flame from the torch around the edges of the sensor glass. As it begins to attach, you can see a "white" border appear, where an air gap has appeared within the epoxy layer. Chase this white border, taking care not to heat the glass (or sensor chip) for too long. Otherwise, the expansion might be too great and the glass will crack. When you're done, remove the sensor glass and keep it in a safe place.
Next up, protecting the sensor wires. Later!
The next step you have to do is to remove the sensor cover glass from the sensor assembly. Every sensor that you see in the market comes packaged so that there is a cover glass over the actual silicon. This helps to protect the array from damage, as well as the very delicate wire-bonds that interfaces the silicon to the chip packaging and ultimately the pins.
![Image](https://farm2.staticflickr.com/1603/25673574503_2f6fd16b4d_m.jpg)
Here's the sensor. It's already been opened once and loosely re-attached. That's why the borders around the sensor glass look funky. To remove this glass, you can either pry it open (bad idea) or use a butane torch, like this one from Dremel:
![Image](https://farm2.staticflickr.com/1606/25673741713_da92760da7_m.jpg)
I started off using an Xacto knife to try to pry open the glass. However, the lever action is too great and you will inevitably end up with a chipped or cracked glass. This is very dangerous because there are, as I mentioned, some very fine wires bonded to the silicon chip. An errant glass shard can break one of these wires, ensuring you end up with a dead sensor. Instead, what you do is run a naked flame from the torch around the edges of the sensor glass. As it begins to attach, you can see a "white" border appear, where an air gap has appeared within the epoxy layer. Chase this white border, taking care not to heat the glass (or sensor chip) for too long. Otherwise, the expansion might be too great and the glass will crack. When you're done, remove the sensor glass and keep it in a safe place.
Next up, protecting the sensor wires. Later!
- Airconvent
- Super Moderator
- Posts: 5787
- Joined: Tue Sep 30, 2003 11:49 pm
- Location: United Federation of the Planets
Re: DIY Cooled Astronomy Camera - In progress
Good project, geyes30!
The Boldly Go Where No Meade Has Gone Before
Captain, RSS Enterprise NCC1701R
United Federation of the Planets
Captain, RSS Enterprise NCC1701R
United Federation of the Planets
Re: DIY Cooled Astronomy Camera - In progress
Thanks!Airconvent wrote:Good project, geyes30!
Part 3: Protecting Gold Wires
As mentioned, there are tiny wire-bonds that are connected to the silicon, and which, if broken, will render the chip useless. In the picture below, you may also notice a bluish border around the sensor chip. Those are patterned connections that also cannot be broken. One slip of the hand and you will end up losing signal from half your chip!
![Image](https://farm2.staticflickr.com/1647/26205298182_dfa869ffef_m.jpg)
To protect these fragile components, you can either be extremely careful, or seal them up with epoxy. Regular epoxy cannot work because there is enough shrinkage to rip the wires off their bonding pads. Instead, use one of these:
![Image](https://farm2.staticflickr.com/1642/26231436311_33b4be271b_m.jpg)
Slow-setting epoxy takes hours to cure, but because it cures so slowly there is minimal shrinkage. These were bought from a hardware store for 6 bucks, and will last forever!
So, after removing the sensor glass, use a toothpick to pick up a bead of epoxy, and let it drip slowly onto the wires. Since I had some bad experience with scratching the blue features, I actually applied epoxy there too. However, take note that the epoxy flows over time, and there is a good chance you will get some of it on the sensing parts. It's not a deal breaker, but you do end up with unmodified regions around the edges.
Next up, debayering!
- joachim.ong
- Posts: 96
- Joined: Wed Jun 08, 2011 3:40 pm
- Favourite scope: 150cm Telescope at Gunma Observatory
- Contact:
Re: DIY Cooled Astronomy Camera - In progress
Great stuff here. Looking forward to the next parts. A project I hope to embark on soon too!
Re: DIY Cooled Astronomy Camera - In progress
Thanks Joachim! A lot of hits-and-misses here. Hopefully this will increase the hits and decrease the number of misses!joachim.ong wrote:Great stuff here. Looking forward to the next parts. A project I hope to embark on soon too!
Step 4: Debayering
This is THE most important part of the whole exercise. While some have had success with tape (Canon 20D's color filter array, or CFA, comes off if you apply tape to it, and rip it out!), solvents (no one has yet come up with anything conclusive), the most successful method seems to be abrasion. It's not trivial to perform this because from the outset it's unclear how much force should be applied. In addition, the type of polishing compound also matters. For me, I tried out the following polishing compounds.
![Image](https://farm2.staticflickr.com/1544/26320116175_d1ed156cca_m.jpg)
The coarseness decreases from left to right. I used Goddard's counter polishing material on the left (from HomeFix) on my first sensor. My main gripe is that the size of the particles are not even. As a result, you can leave pretty deep gashes on your sensors if you are not careful. Since my first sensor died completely, I cannot confirm the suitability of this polishing compound. It definitely did remove the CFA, though. Next I got the TurtleWax from Shell. It is almost purely liquid and wax (?) and has no perceivable abrasive property. Save your money. Don't bother with it. Lastly I got the PlastX from Qoo10. This was actually recommended on one of those SGL threads. It was very homogeneous, but definitely does the trick.
Next up, tools for polishing. A number of tools have been suggested, including sharpened wood, toothpicks, sharpened rivets, etc. Here are the ones I used.
![Image](https://farm2.staticflickr.com/1478/26227711332_dd4ee74b49_m.jpg)
All of these performed okay. The advice on the forums is to go for long, straight strokes, and rotating the sensor 90 degrees "every so often", whatever that means. I've personally found that unnecessary, and use pretty much random strokes throughout, though YMMV. The strength applied depends on the tool used. I actually liked using the steel tweezers, because the fine point and strength of steel allows you to control where you are scratching very well. However, the small area means that you will spend a long time scratching at the sensor. The force I used can be described as normal handwriting strength, without leaving deep impressions on the paper.
The scraping procedure is something like this. I start off with the chopsticks, apply lots of PlastX and scrape pretty hard. You won't reach the naked sensor surface soon, so don't worry too much. The top layer consists of the micro lenses, which is very hard. PlastX is an opaque, off-white suspension so you cannot really see what is happening very well. The way you know that you've made progress is that you will start seeing clear liquid and residue appear. Then, as you continue working on it, you will see green material rubbing off. At this point you will need to proceed with great caution, because that indicates that the CFA is coming off. Beyond that appears to be an additional organic layer of unknown purpose, which overlays the naked sensor. The naked sensor looks gold. Because I generally start from the middle, and miss the edges, I will use the tweezers on the edges of the sensor. This gives me good control and prevents me from over-polishing the sensor, which will result in dead pixels or lines.
For my latest (and most successful) sensor, I used an additional step. I exposed the sensor to strong UVC for about 5 hours, which I had hoped to make the micro lens layer brittle, and easier to scrape off. I am not sure if I actually succeeded, but I think it made a difference. This was inspired by another SGL post here: https://stargazerslounge.com/topic/1663 ... x/?page=79 (see the posts by the aptly-named user, Ultraviolet). After the exposure, I used the tweezer with a light touch, which scratched up the top layer over the whole sensor. I cleaned that up with ethanol and repeated 3 or 4 times.
What I ended up with is a sensor that is monochrome. Here's what an image of my living room wall looks like, after flats and darks processing.
![Image](https://farm2.staticflickr.com/1457/26047599430_8de288af8b_m.jpg)
It's not perfect, but very good as far as I am concerned. Next up, cooling mod.
Re: DIY Cooled Astronomy Camera - In progress
Step 5: Sealing Sensor
After a long hiatus, I am updating this post. In order to use the debayered sensor in a camera, it must first be re-sealed. Since we will be performing cooling on the camera, the space between the sensor cover glass and the silicon surface itself will need to be very dry. To achieve this, I purge the space with dry argon. I work in a lab, so I have access to some glassware to assemble the following:
Argon drying rig by Cyrus Beh, on Flickr
This is called a cold trap, for solvent removal in chemistry. The way I do it is run the already somewhat dry argon through a bed of silica gel, which is topped off with some molecular sieve. The silica gel works well for high moisture, while molecular sieves pick off moisture even at low humidity. This rig drops humidity to around 10%, as determined by a digital hygrometer. All I did was then to blow this dry argon over the sensor surface, and then applying the epoxy-lined cover glass over the sensor. Unfortunately I used a fast setting epoxy, which is not ideal since I had to work very fast and that meant that the epoxy was not evenly applied. Next time I will use the slow setting epoxy.
After a long hiatus, I am updating this post. In order to use the debayered sensor in a camera, it must first be re-sealed. Since we will be performing cooling on the camera, the space between the sensor cover glass and the silicon surface itself will need to be very dry. To achieve this, I purge the space with dry argon. I work in a lab, so I have access to some glassware to assemble the following:
![Image](https://farm2.staticflickr.com/1602/25949755403_be212043b5_m.jpg)
This is called a cold trap, for solvent removal in chemistry. The way I do it is run the already somewhat dry argon through a bed of silica gel, which is topped off with some molecular sieve. The silica gel works well for high moisture, while molecular sieves pick off moisture even at low humidity. This rig drops humidity to around 10%, as determined by a digital hygrometer. All I did was then to blow this dry argon over the sensor surface, and then applying the epoxy-lined cover glass over the sensor. Unfortunately I used a fast setting epoxy, which is not ideal since I had to work very fast and that meant that the epoxy was not evenly applied. Next time I will use the slow setting epoxy.
Re: DIY Cooled Astronomy Camera - In progress
Part 7: Hot and Cold Fingers
Cooling the sensor will inevitably cause condensation to occur. To address this, a 'hot' finger is prepared in addition to the cold finger (for cooling). This will heat up the sensor cover glass and prevents condensation on it. (I had two designs in mind. The other uses a Nichrome wire, but because it has yet to arrive, and it seems silly to waste precious battery power to generate hear, I have put it on the back burner.) The hot finger will draw heat from the Peltier and transfer it, via a copper thermal connection, onto the sensor glass. This is what it looks like.
![Image](https://farm2.staticflickr.com/1483/26542998706_62cae249c4_m.jpg)
Heating frame (or hot finger) for sensor front cover glass by Cyrus Beh, on Flickr
I used a Dremel to cut the copper, and though not perfect, it is pretty good. Note that it is important to remove all the metal burrs along the cuts to prevent scratching the sensor cover glass. It will be mounted onto the cover glass with some thermal paste (though I do not rule out the use of thermal epoxy if necessary to secure the hot finger). The cold finger is similarly prepared, but lies between the PCB and the sensor chip back. Thankfully there is enough room to slot in a copper plate, though do make sure that the copper plate does not short the legs of the sensor chip. I also added a layer of neoprene from an old laptop bag to thermally isolate the sensor chip from the PCB. This prevents inadvertent cooling of the PCB, which can cause condensation on the electronic components. I know that the thermal insulation is working well since the camera temperature sensor (located on the PCB) reports 35degC even when the sensor itself is cooled to 10 degC (confirmed by thermal camera).
Cooling the sensor will inevitably cause condensation to occur. To address this, a 'hot' finger is prepared in addition to the cold finger (for cooling). This will heat up the sensor cover glass and prevents condensation on it. (I had two designs in mind. The other uses a Nichrome wire, but because it has yet to arrive, and it seems silly to waste precious battery power to generate hear, I have put it on the back burner.) The hot finger will draw heat from the Peltier and transfer it, via a copper thermal connection, onto the sensor glass. This is what it looks like.
![Image](https://farm2.staticflickr.com/1483/26542998706_62cae249c4_m.jpg)
Heating frame (or hot finger) for sensor front cover glass by Cyrus Beh, on Flickr
I used a Dremel to cut the copper, and though not perfect, it is pretty good. Note that it is important to remove all the metal burrs along the cuts to prevent scratching the sensor cover glass. It will be mounted onto the cover glass with some thermal paste (though I do not rule out the use of thermal epoxy if necessary to secure the hot finger). The cold finger is similarly prepared, but lies between the PCB and the sensor chip back. Thankfully there is enough room to slot in a copper plate, though do make sure that the copper plate does not short the legs of the sensor chip. I also added a layer of neoprene from an old laptop bag to thermally isolate the sensor chip from the PCB. This prevents inadvertent cooling of the PCB, which can cause condensation on the electronic components. I know that the thermal insulation is working well since the camera temperature sensor (located on the PCB) reports 35degC even when the sensor itself is cooled to 10 degC (confirmed by thermal camera).
Re: DIY Cooled Astronomy Camera - In progress
Quick Update to Step 7:
It became apparent fairly quickly that using the hot finger was going to be a challenge. Firstly, the finger directs an inordinate amount of heat back to the sensor, and since Peltiers are notoriously inefficient, the net effect was that the sensor glass was getting heated to about 70 degrees C, and the cooling simply couldn't keep up. The high temperature triggers a shutdown routine on the camera, which is probably a good thing!
This happens partly because the area on the hot finger was too large, so the thermal transfer through the copper finger was far greater than the transfer through the aluminium heat sink. One simple remedy was to cut down the area of the finger, so that the thermal resistance becomes higher. This is fairly successful. I am able to get some cooling to the chip, while maintaining the sensor glass at a toasty 39degrees, thereabouts. Unfortunately, I found that the cooling remains fairly inefficient. My fan was able to get the Peltier to only about 70 degrees, and with my applied voltage I could only get the cold finger to get down to about a couple of degrees below ambient! I bought a water cooling unit but before I did the radical modification of my enclosure to incorporate the very large water cooling unit (not ideal!), I decided to do a Peltier stack.
Using an external Peltier to get the hot finger to around 5-10 degrees above room temp, I was able to finally have all the thermal components working correctly - a slightly warm sensor glass and a chilled sensor chip. When I assembled the whole thing in the camera, I found that the hot finger is too thick, and causes the chip to tilt!!!!!!!!!!! Argh~~!!
The functional but ugly cooled CMOS chip by Cyrus Beh, on Flickr
At this point I've decided to do away with the hot finger, and instead decouple the heating and cooling functions. That will give me much more flexibility with my design. Next up - Nichrome heating wires to the rescue!
It became apparent fairly quickly that using the hot finger was going to be a challenge. Firstly, the finger directs an inordinate amount of heat back to the sensor, and since Peltiers are notoriously inefficient, the net effect was that the sensor glass was getting heated to about 70 degrees C, and the cooling simply couldn't keep up. The high temperature triggers a shutdown routine on the camera, which is probably a good thing!
This happens partly because the area on the hot finger was too large, so the thermal transfer through the copper finger was far greater than the transfer through the aluminium heat sink. One simple remedy was to cut down the area of the finger, so that the thermal resistance becomes higher. This is fairly successful. I am able to get some cooling to the chip, while maintaining the sensor glass at a toasty 39degrees, thereabouts. Unfortunately, I found that the cooling remains fairly inefficient. My fan was able to get the Peltier to only about 70 degrees, and with my applied voltage I could only get the cold finger to get down to about a couple of degrees below ambient! I bought a water cooling unit but before I did the radical modification of my enclosure to incorporate the very large water cooling unit (not ideal!), I decided to do a Peltier stack.
Using an external Peltier to get the hot finger to around 5-10 degrees above room temp, I was able to finally have all the thermal components working correctly - a slightly warm sensor glass and a chilled sensor chip. When I assembled the whole thing in the camera, I found that the hot finger is too thick, and causes the chip to tilt!!!!!!!!!!! Argh~~!!
![crying [smilie=crying.gif]](./images/smilies/crying.gif)
![Image](https://farm8.staticflickr.com/7280/26372731364_26eeff28d8.jpg)
At this point I've decided to do away with the hot finger, and instead decouple the heating and cooling functions. That will give me much more flexibility with my design. Next up - Nichrome heating wires to the rescue!