latest news

Custom refractors

We now offer custom built refractors with world class optics from APM in Germany. For more info please mail to: info@jtwastronomy.com

Instrumentation specialist

Thanks to the acquisition of new, large machinery we are optimising our facility towards making custom instrumentation - large telescopes, mounts, optics, etc. For more info please mail to: info@jtwastronomy.com

Astronomy news

Check our selected RSS feeds with the latest astronomy news

read more ...

other information

Removing the bayer matrix from a DSLR/CCD sensor

Convert a colour camera into a monochrome camera by debayering it

This is a tutorial that will enable you to convert your colour DSLR or CCD camera into a monochrome camera. With our modification you will be able to turn your DSLR camera into something far more sensitive and eliminate the last major hurdle between CCD and DSLR cameras. Of course the CCD will always be superior, but so many people get started with a DSLR, this modification will breathe new life into your camera, while you save for that precious large format CCD camera! With cooling modifications a monochrome DSLR is more than capable of serious narrowband imaging!

Bayer filters

Image made with deep cooled and debayered Canon 600D (c) Dr. F Hemmerich

It is important to know that this process is not perfect, and you cannot expect a perfect flat field from one of these cameras although it is close. This is not really neccessary for regular photography, but for astrophotography flat fielding becomes very important for a high quality result. The images above and below were made using debayered DSLRs, this proves that despite the process not being quite perfect, it is a very economical and very useable solution to a common problem. The low sensitivity of a DSLR and the difficulties of narrowband image are a thing of the past, even more so once you consider our cooling solutions.

Debayered sensor

Brief theory of the modification

We see colour because our eyes can differentiate between photons of different energy levels in the visual range. A silicon camera sensor can only detect photons from around 350nm to 1000nm, they are all counted as hits. The sensor cannot sense anything about a photon's energy and therefore it's colour. The CFA was invented by Bryce Bayer in 1976, each pixel has it's own individual colour filter. These are arranged in a 2x2 grid. We know where the light was detected so we also know which colour filter it passed through, thus we have a measure of intensity and energy, and therefore the photon's 'colour' can be deduced. Using this information we can reconstruct a colour image from the monochrome sensor. This information comes with a heavy price, a significant reduction in sensitivity.

For example in one 2x2 grid only one of the pixels can detect a 'red' photon. Meaning 75% of them are being absorbed by the green and blue filters. This explains the massive gains with our astronomy modification. By removing the second low pass filter, who's only function is to correct the sensor's natural high sensitivity to red light for regular photography purposes, 4x more red light is permitted to reach the sensor. By then removing the bayer array there is 4x as much surface area available to detect a 'red' photon. One could say that this equates to a 16x increase in sensitivity, we are not implying a quantum efficiency of 480% and beyond, that would be crazy! However the gains are real!

Both a CMOS and CCD (CCDs are a type of CMOS sensor) are made of silicon. In theory the quantum effiency of the two sensors should be the same (http://en.wikipedia.org/wiki/Photoelectric_effect) if it wasn't for the typically smaller active surface area due to both smaller pixel size preferred for regular photography and area lost to the pixel's readout transistors. Losses in the low pass and colour filter array filters simply compound the problem. The CMOS sensor is not less sensitive perse, rather the apparant lack of sensitivity is due to losses before the light even reaches the sensor. Eliminating these losses translates into a DSLR based system that is approaching CCD sensitivity for a DSLR price. In practice the CCD will remain king for some time, particularly as the DSLR market chases goals that are at odds to astrophotography. Look after your old DSLR cameras! If you successfully perform this modification, you will be amazed at the results!

Debayered sensor

Disclaimer

There is a high risk of damage to your sensor if you are inexperienced with this procedure. Some ability with fine work will be helpful but we at JTW Astronomy in no way, shape or form accept any responsiblity for losses due to using this procedure. It is dangerous, it is difficult and it is at your own risk. If you are not comfortable doing this, we will be happy to do it for you (guaranteed 2 week turnaround). We retain all rights to this procedure and any commercial products based on this procedure are forbidden, as is profiting from this procedure. This is for helping fellow amateurs avoid the pitfalls that we hit and to avoid wasting time and money chasing dead ends (as we also did!), if you wish to discuss a product that you intend to develop, please get in touch. This modification is a result of over a year of experimentation, countless hours and several thousand euros of tooling and broken sensors, please respect this.

Items needed for the modification (Newer sensors - 400D or later)

  • A DSLR camera you don't mind experimenting on
  • A butane pen torch or hot air PCB rework station
  • Slow curing epoxy resin, Bison brand is known to be suitable
  • A sensor cleaning swab, microfibre material removed to make a plastic polishing tool. Use one that is much thinner than your sensor, 1/3 to 1/4 of the sensor's shortest side
  • Cotton buds cleaned with isopropanol
  • Glass polish, such as from a windscreen scratch removal kit
  • 99.9%+ pure isopropanol
  • Distilled water
  • A very sharp knife, at least new Stanley blade level of sharpness
  • Powder free surgical style gloves
  • A small clean tub with a lid, to put the sensor in after step 1, preferably with anti static foam on the bottom
  • Two plastic toothpicks
  • A dustbulb to blow dust off the sensor periodically, if needed
  • 5-6 Hours of uninterrupted free time
  • Optional: Low power microscope (10-20x)
  • Optional: AR coated replacement coverglass
  • Optional: Glass cutter with new wheel and oil system
  • Optional: Inert gas (welding mix for example) to purge sensor

A note on older DSLR sensors/CCD sensors

If you are debayering an older camera, you may be lucky and able to remove the bayer filter array using solvent, this seems to work better on CCD sensors than DSLR sensors (good news for OSC CCD owners!). If this is the case, simply obtain a bottle of dichloroethane (chloroform, novolac remover and dichloromethane have also shown some promise). You can follow the procedure below, but substitute the glass polish for the solvent. WARNING: The chemicals are extremely volatile & hazardous to human health, you will need to work in a fume cupboard. We have been informed that dichloroethane is a potential CARCINOGEN! We accept no responsibility for damage to your health using the solvent method. This is not to be taken lightly, it is only a camera. You cannot buy new lungs. It may be worth using our abrasion method simply because it does not use any dangerous compounds. Every newer sensor we applied this method to has proven immune to solvent attack, anything newer than a 400D, do not even bother trying solvents.

Preparation

The key to this modification is to work slowly and deliberately. It goes without saying, but a clean work area is essential. The sensor needs to spend as little time open to the elements as possible. To this end, do not even begin this modification until you have all the items on the list and the free time to do it in two sittings. The first step will take about 1 hour and 24 hours waiting time. The second step takes around 4-5 hours and needs to be finished in one sitting. This modification should not take place in a damp or humid environment, when resealing the sensor, enough moisture will mean it is possible that the sensor will fog over internally if you fit a cooling modification.

Step 1: Removing the coverglass of the sensor

We start the tutorial under the assumption that the sensor is already out of the camera. There are already plenty of tutorials online. So we will not repeat this information. If you have an older camera you can remove the coverglass using a sharp knife, by running it around the edge of the window you will eventually get under the glass and it will pop off. Do not rush this, the coverglasses are very fragile. With newer sensors the coverglass is held far more securely. To remove this you will need our butane pen torch heat method. A hot air rework station would be slightly safer, but we assume that most hobbyists do not own one of these.

Making sure your camera is on a non-flammable surface, take your pen torch and heat one corner of the sensor's window. This will take just a second, the differential thermal expension will cause the bond to break in that spot. You will see the clear bond has turned opaque at the spot where the heat was applied. The point now is to 'chase' this break all the way around the sensor. By heating one edge of the opaque area, the break will grow. Do not apply heat for too long (anything more than a few seconds!), allow the sensor time to cool to ambient temperature inbetween heating operations or it will crack and you will get particles of glass on the sensor face which require extreme care to remove. Eventually the entire bond will have broken all the way around the sensor. The cover glass can now be removed with a sharp blade. Don't forget to don your gloves to prevent fingerprints on the coverglass. Set the coverglass in your dustfree box on the two plastic toothpicks to prevent the underside touching anything and possibly getting dirty.

Step 2: Protecting the gold wires! (Optional)

You will notice some extremely thin and delicate gold wires running to the sensor's face. You can break these almost just by looking at them! If you are afraid to break them, you can encase them in epoxy resin to protect them. Using a slow setting epoxy is very important, as it will allow you some very gentle reworking of the epoxy without it being hard enough to pull on the gold wires. Use a cotton tip with the cotton bit cut off to make a dripper tool. You should not touch the wires at all. Allow the epoxy to drip onto the inside edge of the sensor housing (not the sensor itself), do this very carefully. The epoxy will then flow over the gold wires rather than drip on them and possibly pull on them. Avoiding dripping epoxy on the sensor's face is a very good idea. If you do accidentally do this, remove the epoxy using the isopropanol before it sets. Once the wires are coated, leave the sensor in the tub with the lid on overnight to keep dust off the sensor. Don't worry if the epoxy is above the level of the mating face of the sensor, you can trim it with a sharp knife once it is very hard, but you should try to avoid this as much as possible. Any reworking of the epoxy should be kept to a minimum, and if it is not on the face of the sensor you should wait until it is completely cured before working on it, working on epoxy that is only beginning to harden up will break the gold wires very easily.

Step 3: Removing the bayer mask

This method relies on abrasion. The sensor swab should have already been stripped of its microfibre covering. Use the broad tip of the sensor swab for general polishing, you can use the thinner tip at the other end to deal with corners or stubborn spots that are too small for the broader end. It is important not to apply too much pressure, as this can damage the sensor. Be gentle and be patient! Dip the broad end of the swab in the glass polish just enough to coat the tip and begin polising the sensor face. Do not use circular strokes, as you will be momentarily (but regularly!) using the sensor swab with a very thin effective width, this will cut through the bayer mask like a bike tyre through snow. Use linear strokes and travel the whole length of the sensor in one stroke, turn the sensor 90 degrees every now and then to polish perpendicularly. It will help with uniformity. It is essential to polish the surface area equally, so that the mask is removed at the same rate over the entirety of the sensor. If you neglect one area and need to rub at it more, if the area around it is already stripped of bayer, you will be also removing the sensor surface itself.

You can use handle of the sensor swap to deal with smaller spots or corners, cut a small flat onto it with a sharp knife. There is little room for error here, if you maintain uniformity from the start the debayering will be much more successful. You should regularly stop and check every 30 seconds or so, remove the excess polish with a cotton bud (q-tip) dipped in a little distilled water. Clean, assess (preferably under a microscope), continue if necessary. If you get bored or frustrated do not just go grinding away. If it is getting tedious, take a break. Do not rush. Do not be afraid to mount the sensor in the camera to take a test shot, taking a flat and stretching it will give you a good idea where you are in the process. If you get a colour image that is dimmer in places, you are just through the microlenses. If you have an area that is completely free of bayer but has dark spots, you have gone to far and should avoid that area at all costs!

Debayered sensor

This is what you can expect to see under the microscope

The first step is to remove the microlenses, these are tougher than the bayer and you need 100% removal before you even attempt to remove the bayer filters. It is very easy to spot when the microlenses are gone, The surface will appear much flatter and shinier, and also there will be no coloured residue on the swab until you hit the bayer matrix itself. Once through the microlenses you can proceed to remove the bayer mask. Once you are working on the bayer mask you will notice the polish and tip of the polishing tool turn green. Regular cleaning of the polish and checking progress needs to happen every 10-20 seconds at this point. You will notice as it is removed there is a definite clear-ish layer underneath it. We suspect this is a coating on the sensor, but cannot be sure. What we do know is that although it is probably preferable to keep it, it is extremely difficult to do as it is much easier to remove than the bayer. By removing it you will get more uniform flats but a slight decrease in sensitivity (much less than if you go too deep into the sensor). You will notice multicoloured rings and contour lines on the sensor just like a map. You need to stop when the sensor is a golden/orange colour. You will see the contour lines can be removed easily. Once the sensor begins to look uniform it is time to avoid that area as you are now grinding into the sensor itself. If performed properly you should now have a sensor that has no contour lines and a reasonably uniform golden/orange finish. If you did this without a scratch or going too deep. Test the sensor in the body, if it still works - Congratulations! Debayered sensor

This is the sensor surface after we are finished. It is entirely uniform (the green colour is an optical artifact due to the rough and ready method used to take this photo, with your own eyes this area will appear a gold/orange colour).

Step 4: Closing the sensor

The sensor and coverglass need to be properly, carefully and thoroughly cleaned with isopropanol before attaching the coverglass. The mating face of the sensor also needs to be clean and any remnants of the original bonding agent or wire protecting epoxy need to be removed. Again you should be wearing gloves for this. Mix up a little epoxy and make a THIN line around the rim of the sensor. If you have inert gas available aim the nozzle near the sensor so the inert gas will be sealed in when the glass is replaced. Sit the glass on the mating face and ensure the seal runs all the way around the glass. Set the sensor in your dust free box and leave it for 24 hours. Reinstall it to your camera, if it still works, give yourself a pat on the back!

Step 4: Optional improvement - AR coated sensor window

You can cut the AR glass to size using the glass cutter and use it to replace the coverglass. An AR window from Edmund Optics has just a 0.5% reflectivity. A significant improvement on the original that will eliminate reflections almost entirely, especially when used in a full spectrum modified camera.

That's it! Clear skies!

If you liked this tutorial please let us know. Any suggestions for other tutorials, corrections, improvements or request for extra info in this one are gladly received. Please post on our Facebook page, or send to info@jtwastronomy.com

Rosette Nebula (c) M Woodward

Image made using uncooled debayered 600D (c) M Woodward

(c) JTW Astronomy 2012. All rights to this modification are reserved by JTW Astronomy, use is only allowed for non commercial purposes