Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
10-25-2020, 05:49 PM - 7 Likes   #1
Loyal Site Supporter
Loyal Site Supporter




Join Date: May 2007
Location: Flagstaff, Arizona
Posts: 746
Converting (part of) a Pentax *istD Camera Sensor to a Black & White Sensor

And, ruining a few other cameras along the way!

As part of my amusement during the pandemic, I made a simple spectrograph, using a plastic diffraction grating from Ebay for the dispersing element (for about $1). One of the reasons for doing this was to enable me to measure the spectrum coming from various light bulbs that one might use to photograph art works (or stuff for Ebay). The spectra of most modern light bulbs (CFLs and LEDs) are not very uniform across the visible spectrum. CFLs produce most of their light at a few specific wavelengths, and LEDs tend to have “holes” in their spectra.

Of course, I could have just bought a fancy digital spectrometer on Ebay or Amazon for between $500 and $1000, but where’s the fun in that?!

Once you have the spectrograph, you need a detector. Well, we’ve all got one (if not many!) of those - our digital color cameras. Turns out, though, that if you want to be able to make semi-accurate measurements of spectra, a DSLR is not really the way to go. The problem is the color filtering deposited on the surface of the CCD or CMOS sensor in the typical camera - the Bayer filter. The R,G,B response due to the filter results in significant variations in the sensitivity for different colors. Also, there tends to be “leakage” between the filters - for some examples, see the curves here: Spectral Response .

OK, so what I really want is a camera with more-or-less equal sensitivity across the visible light spectrum. A Black & White camera!

Well, you can buy one of those, too - but get out your AmEx Platinum card! Here are the current digital camera options:

Sony A6000 Monochrome - this one is a mere $2,250

Leica M Monochrom - this’ll set you back $8,000 (but it IS a Leica)

or, go all out for the Phase One XF IQ3 100MP Achromatic, at $50,000 (yikes!)

That’s it! Even the cheapest car backup cameras (you can get one from Walmart for $12) and web cams all have the color filters in them.

As an avid do-it-yourselfer and a cheapskate, I figured I could do better - by removing the Bayer filter from the sensor in some camera. A bit of web searching indicates various folks have done just that. You get inside the camera, all the way down to the sensor chip, and literally scrape off the Bayer filter layer.

Even after you’ve gotten inside the camera, there is still a problem: all the sensor chips have a glass layer glued to the chip carrier, just in front of the chip. Getting past this glass turned out to be the biggest challenge.

I asked around locally and got donations of a couple of older point&shoot cameras on which to practice my skills. Thus, my first attempt was on a Sony DSC-W830. Ifixit.com has nice instructions on how to get inside this camera, and it was rather easy to get down to the sensor - with its protective layer of glass. I made a little plunger to serve as a glass smasher to break the cover glass. I got the sensor glass out without too much apparent damage, but managed to seriously scrape some circuit board traces next to the sensor (see picture below) and probably did damage one or two of the very fine wire which connect the sensor chip to the rest of the camera electronics. I scraped off some of the color filtering material. When put back together, the camera still functioned to some extent, but did not make useful images - many of the lines of pixels were missing. I’m not sure if the problem was from the scraped traces or the sensor wires (or both). At any rate, one camera dead.




Next try was with an old IBM web cam. Again, it looks like I broke one of the very tiny wires that connect the sensor chip to the rest of the camera circuitry in the process of breaking through the cover glass. This camera was completely non-functional when put back together again. See the next picture, showing the shards of glass that landed on the IBM camera sensor (for scale, the central black square is about 5 mm by 6 mm, and darker interior rectangle - the actual sensor, I think - is only about 2x3 mm). Second camera dead.




OK - next up, a Canon SD800 (again with Ifixit.com assistance). This time, I used a diamond-tipped scribe to get into the cover glass, and removed it in pieces with no obvious damage to the sensor wires. However, in my hurry to put it all back together again, I left the battery in the camera - BIG MISTAKE! During re-assembly, I must have shorted something out: one of the itty-bitty surface mount components glowed very red, very briefly, followed by an audible “pfft” and a wisp of smoke! I put it back together anyhow, but the camera is totally non-functional. Dead camera #3. (Actually, I still have hopes for this camera - you can buy them on Ebay for less than $10, and I can probably create something working by combining parts from two of them.)

However, by now, I was ready to try a “real” camera - a Pentax *istD . This one came via Ebay, where it was listed as probably not working, but the price seemed cheap enough ($35 for this one) and there was a fair probability that it was “user error” on the part of the seller, not understanding some of the subtleties of getting a Pentax DSLR to shoot with a non-A lens. Turns out, it worked just fine! (A bit of this quest is recounted here: Is this ist D likely to work? - PentaxForums.com and Questionable *ist D works just fine! - PentaxForums.com )

This time, I used the excellent instructions put out on the web by Vincenzo (Enzo) Miceli: Pentax *ist D IR Mod - Enzo's Home . I had several email exchanges with Enzo - very friendly and helpful. The biggest problem with this camera turned out to be getting out one of the little screws, as recounted in this thread: Getting out teeny, tiny seized screw - PentaxForums.com . Enzo’s goal was solely to replace the IR filter in the camera, so I was once again faced with getting off the sensor cover glass, for a Sony ICX413AQ CCD.

Also this time, rather than destroy yet another camera, I decided to check with some experts in such things - companies that provide cover glass removal as a service. I wasn’t sure whether I would get any answers for a non-revenue-producing inquiry, but was most pleasantly surprised and grateful for the helpful information provided by Hugh Garvey with Salvo Technologies. He pointed out that they have had good luck removing Sony ICX series cover glass with heat. (If you need some commercial work on image sensors, please consider an inquiry to Salvo!)

So, out with the heat gun. After several minutes of heating the glass cover (at which point it was too hot to touch), I could see the adhesive holding the cover glass starting to release its grip. I pried under one end of the glass, and off it popped! Well, except that it broke at the other end, which had not gotten hot enough. But the chip was completely exposed and NO wires were broken this time. Here’s a shot of the naked sensor (with some glass shards and local dust):




Application of a wooden toothpick was adequate to scrape a central area of the filtering from the CCD. The filter material came off fairly easily. I was very careful not to get too close to the edges where those fine wires are. Here’s a shot of the scraped area:



and a close-up:




Notice that there is a fair amount of crud left behind, from breaking the glass and scraping (as well as a house that has the inevitable dust and some dog hairs!). I cleaned this up about as well as I could, figuring I’d put the camera back together and see what I had achieved. I can live with this camera if its only problem is some dust spots in my images.

And, TADA - the camera still works!!!!! Here is a shot of blue sky with the modified camera. You can plainly see the area where the Bayer filter has been scraped off. The scraped spot is flipped left to right compared to the view above of the sensor.




Further testing shows that indeed that CCD response in the scraped area responds as I hoped - there is no particular color sensitivity from pixel to pixel - they all respond fairly uniformly across the optical spectrum, with whatever their intrinsic quantum efficiency is. The absolute data values for the pixels have different scaling factors in the PEFs from the camera, left over from how the camera interpreted what used-to-be their color values.

Here’s a cropped shot of the spectrum of an incandescent light bulb, taken with the modified *istD and my spectrograph. The spectrum is that of a black body, and relatively smooth across all wavelengths (stronger in the red and tailing of to the blue). In the top half of the image is the normal color response of the camera (where the Bayer layer still exists) to the spectrum. Note the non-uniform intensity thanks to the R,G,B filtering - this is especially obvious in the transition from red to green, where yellow is somewhat shortchanged. In the bottom of the image is the spectrum in black and white - here it is position along the spectrum that tells you what wavelength you are observing.




The above image was created from the camera PEF file, processed by RawDigger software, which can output a TIF. I can read that TIF file into a program called ImageJ (ImageJ ; freeware from the US National Institutes of Health - your tax dollars at work!). ImageJ will make amplitude cuts across the image frame. When doing this for color TIF files, ImageJ actually processes the R,G,B image subsets and can display them separately or in a 3-color plot.

Here is the three color plot for the spectrum in the top half of the image above. You can plainly see the filtering effects of each of the R,G,B filters. The valid spectral range is around 650 to 2400 pixels (probably between about 700 nm (red) to about 400 nm (blue)).




And, here is the same plot for the bottom half of the image, the “black&white” version. Now the response is basically uniform across the image (except for some blips where there is crud on the sensor) and just about the same for all 3 “colors” (to within a scaling factor for each “color” - remember that the camera and software know nothing about the modification), and more properly indicative of the spectrum of the incandescent bulb - which should be that of a black body at around 2700-2800K. The spectral range is the same as the above plot, except at the high pixel (blue) end, where you can see the edge effect of where the sensor has been scraped.




I am quite happy with the performance of the modified *istD. I still need to make some more measurements of spectral uniformity across the frame, and figure out how to do a wavelength calibration. The latter won’t really be very hard - I can use CFL bulbs and a neon bulb that I have to stamp the spectra with wavelength.

I will probably leave the *istD as it is, but may well try to modify some other point-n-shoots since I am now more confident that I can remove that darned cover glass from the sensor chips. I will also keep my eye out for another cheap *istD on Ebay.

Soon, I hope to be able to report on the color characteristics of various light sources.


Last edited by AstroDave; 10-27-2020 at 06:37 AM. Reason: clarify: Sony ICX series cover glass
10-25-2020, 06:12 PM   #2
PJ1
Site Supporter
Site Supporter
PJ1's Avatar

Join Date: Jul 2012
Location: Mosquito Creek, Australia
Photos: Gallery | Albums
Posts: 4,390
Great work. I am amazed at what some people get up to. The B&W spectrum image really says it all. My hat is off to you.
10-25-2020, 06:30 PM   #3
Emperor and Senpai
Loyal Site Supporter
VoiceOfReason's Avatar

Join Date: Jun 2010
Location: Mishawaka IN area
Photos: Albums
Posts: 6,011
There have been a few of the guys at Cloudy Nights who have done this. Here's one Canon 7D Monochrome Mod - Imaging Test Log - DSLR, Mirrorless & General-Purpose Digital Camera DSO Imaging - Cloudy Nights
10-26-2020, 12:27 AM   #4
Loyal Site Supporter
Loyal Site Supporter
BigMackCam's Avatar

Join Date: Mar 2010
Location: North-East of England
Posts: 15,500
Excellent work, Dave, and a very interesting read for my morning coffee time... thanks for posting

10-27-2020, 04:28 AM   #5
Site Supporter
Site Supporter




Join Date: Sep 2006
Location: midwest, United States
Photos: Gallery | Albums
Posts: 4,751
QuoteOriginally posted by AstroDave Quote
And, ruining a few other cameras along the way!

As part of my amusement during the pandemic, I made a simple spectrograph, using a plastic diffraction grating from Ebay for the dispersing element (for about $1). One of the reasons for doing this was to enable me to measure the spectrum coming from various light bulbs that one might use to photograph art works (or stuff for Ebay). The spectra of most modern light bulbs (CFLs and LEDs) are not very uniform across the visible spectrum. CFLs produce most of their light at a few specific wavelengths, and LEDs tend to have “holes” in their spectra.

Of course, I could have just bought a fancy digital spectrometer on Ebay or Amazon for between $500 and $1000, but where’s the fun in that?!

Once you have the spectrograph, you need a detector. Well, we’ve all got one (if not many!) of those - our digital color cameras. Turns out, though, that if you want to be able to make semi-accurate measurements of spectra, a DSLR is not really the way to go. The problem is the color filtering deposited on the surface of the CCD or CMOS sensor in the typical camera - the Bayer filter. The R,G,B response due to the filter results in significant variations in the sensitivity for different colors. Also, there tends to be “leakage” between the filters - for some examples, see the curves here: Spectral Response .

OK, so what I really want is a camera with more-or-less equal sensitivity across the visible light spectrum. A Black & White camera!

Well, you can buy one of those, too - but get out your AmEx Platinum card! Here are the current digital camera options:

Sony A6000 Monochrome - this one is a mere $2,250

Leica M Monochrom - this’ll set you back $8,000 (but it IS a Leica)

or, go all out for the Phase One XF IQ3 100MP Achromatic, at $50,000 (yikes!)

That’s it! Even the cheapest car backup cameras (you can get one from Walmart for $12) and web cams all have the color filters in them.

As an avid do-it-yourselfer and a cheapskate, I figured I could do better - by removing the Bayer filter from the sensor in some camera. A bit of web searching indicates various folks have done just that. You get inside the camera, all the way down to the sensor chip, and literally scrape off the Bayer filter layer.

Even after you’ve gotten inside the camera, there is still a problem: all the sensor chips have a glass layer glued to the chip carrier, just in front of the chip. Getting past this glass turned out to be the biggest challenge.

I asked around locally and got donations of a couple of older point&shoot cameras on which to practice my skills. Thus, my first attempt was on a Sony DSC-W830. Ifixit.com has nice instructions on how to get inside this camera, and it was rather easy to get down to the sensor - with its protective layer of glass. I made a little plunger to serve as a glass smasher to break the cover glass. I got the sensor glass out without too much apparent damage, but managed to seriously scrape some circuit board traces next to the sensor (see picture below) and probably did damage one or two of the very fine wire which connect the sensor chip to the rest of the camera electronics. I scraped off some of the color filtering material. When put back together, the camera still functioned to some extent, but did not make useful images - many of the lines of pixels were missing. I’m not sure if the problem was from the scraped traces or the sensor wires (or both). At any rate, one camera dead.




Next try was with an old IBM web cam. Again, it looks like I broke one of the very tiny wires that connect the sensor chip to the rest of the camera circuitry in the process of breaking through the cover glass. This camera was completely non-functional when put back together again. See the next picture, showing the shards of glass that landed on the IBM camera sensor (for scale, the central black square is about 5 mm by 6 mm, and darker interior rectangle - the actual sensor, I think - is only about 2x3 mm). Second camera dead.




OK - next up, a Canon SD800 (again with Ifixit.com assistance). This time, I used a diamond-tipped scribe to get into the cover glass, and removed it in pieces with no obvious damage to the sensor wires. However, in my hurry to put it all back together again, I left the battery in the camera - BIG MISTAKE! During re-assembly, I must have shorted something out: one of the itty-bitty surface mount components glowed very red, very briefly, followed by an audible “pfft” and a wisp of smoke! I put it back together anyhow, but the camera is totally non-functional. Dead camera #3. (Actually, I still have hopes for this camera - you can buy them on Ebay for less than $10, and I can probably create something working by combining parts from two of them.)

However, by now, I was ready to try a “real” camera - a Pentax *istD . This one came via Ebay, where it was listed as probably not working, but the price seemed cheap enough ($35 for this one) and there was a fair probability that it was “user error” on the part of the seller, not understanding some of the subtleties of getting a Pentax DSLR to shoot with a non-A lens. Turns out, it worked just fine! (A bit of this quest is recounted here: Is this ist D likely to work? - PentaxForums.com and Questionable *ist D works just fine! - PentaxForums.com )

This time, I used the excellent instructions put out on the web by Vincenzo (Enzo) Miceli: Pentax *ist D IR Mod - Enzo's Home . I had several email exchanges with Enzo - very friendly and helpful. The biggest problem with this camera turned out to be getting out one of the little screws, as recounted in this thread: Getting out teeny, tiny seized screw - PentaxForums.com . Enzo’s goal was solely to replace the IR filter in the camera, so I was once again faced with getting off the sensor cover glass, for a Sony ICX413AQ CCD.

Also this time, rather than destroy yet another camera, I decided to check with some experts in such things - companies that provide cover glass removal as a service. I wasn’t sure whether I would get any answers for a non-revenue-producing inquiry, but was most pleasantly surprised and grateful for the helpful information provided by Hugh Garvey with Salvo Technologies. He pointed out that they have had good luck removing Sony cover glass with heat. (If you need some commercial work on image sensors, please consider an inquiry to Salvo!)

So, out with the heat gun. After several minutes of heating the glass cover (at which point it was too hot to touch), I could see the adhesive holding the cover glass starting to release its grip. I pried under one end of the glass, and off it popped! Well, except that it broke at the other end, which had not gotten hot enough. But the chip was completely exposed and NO wires were broken this time. Here’s a shot of the naked sensor (with some glass shards and local dust):




Application of a wooden toothpick was adequate to scrape a central area of the filtering from the CCD. The filter material came off fairly easily. I was very careful not to get too close to the edges where those fine wires are. Here’s a shot of the scraped area:



and a close-up:




Notice that there is a fair amount of crud left behind, from breaking the glass and scraping (as well as a house that has the inevitable dust and some dog hairs!). I cleaned this up about as well as I could, figuring I’d put the camera back together and see what I had achieved. I can live with this camera if its only problem is some dust spots in my images.

And, TADA - the camera still works!!!!! Here is a shot of blue sky with the modified camera. You can plainly see the area where the Bayer filter has been scraped off. The scraped spot is flipped left to right compared to the view above of the sensor.




Further testing shows that indeed that CCD response in the scraped area responds as I hoped - there is no particular color sensitivity from pixel to pixel - they all respond fairly uniformly across the optical spectrum, with whatever their intrinsic quantum efficiency is. The absolute data values for the pixels have different scaling factors in the PEFs from the camera, left over from how the camera interpreted what used-to-be their color values.

Here’s a cropped shot of the spectrum of an incandescent light bulb, taken with the modified *istD and my spectrograph. The spectrum is that of a black body, and relatively smooth across all wavelengths (stronger in the red and tailing of to the blue). In the top half of the image is the normal color response of the camera (where the Bayer layer still exists) to the spectrum. Note the non-uniform intensity thanks to the R,G,B filtering - this is especially obvious in the transition from red to green, where yellow is somewhat shortchanged. In the bottom of the image is the spectrum in black and white - here it is position along the spectrum that tells you what wavelength you are observing.




The above image was created from the camera PEF file, processed by RawDigger software, which can output a TIF. I can read that TIF file into a program called ImageJ (ImageJ ; freeware from the US National Institutes of Health - your tax dollars at work!). ImageJ will make amplitude cuts across the image frame. When doing this for color TIF files, ImageJ actually processes the R,G,B image subsets and can display them separately or in a 3-color plot.

Here is the three color plot for the spectrum in the top half of the image above. You can plainly see the filtering effects of each of the R,G,B filters. The valid spectral range is around 650 to 2400 pixels (probably between about 700 nm (red) to about 400 nm (blue)).




And, here is the same plot for the bottom half of the image, the “black&white” version. Now the response is basically uniform across the image (except for some blips where there is crud on the sensor) and just about the same for all 3 “colors” (to within a scaling factor for each “color” - remember that the camera and software know nothing about the modification), and more properly indicative of the spectrum of the incandescent bulb - which should be that of a black body at around 2700-2800K. The spectral range is the same as the above plot, except at the high pixel (blue) end, where you can see the edge effect of where the sensor has been scraped.




I am quite happy with the performance of the modified *istD. I still need to make some more measurements of spectral uniformity across the frame, and figure out how to do a wavelength calibration. The latter won’t really be very hard - I can use CFL bulbs and a neon bulb that I have to stamp the spectra with wavelength.

I will probably leave the *istD as it is, but may well try to modify some other point-n-shoots since I am now more confident that I can remove that darned cover glass from the sensor chips. I will also keep my eye out for another cheap *istD on Ebay.

Soon, I hope to be able to report on the color characteristics of various light sources.
Amazing project and information. Didn't know Bayer filters could be scraped away. Fascinating. Interested to see where your projects goes next. Appreciate all the illustrations and explanations.

Thanks for sharing,
barondla
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, chip, color, cover, ebay, filter, glass, image, photography, photoshop, sensor, spectrum
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Does Sensor Size REALLY Matter? Camera Sensor Size Comparison - Part Two Winder Non-Pentax Cameras: Canon, Nikon, etc. 12 11-08-2020 10:09 PM
Converting Black and White file to Color TPMartin Troubleshooting and Beginner Help 3 06-05-2016 11:57 PM
Converting to Black-and-White K McCall Digital Processing, Software, and Printing 16 08-08-2009 08:28 PM
Converting to Black & White using Apple Aperture Gabe Digital Processing, Software, and Printing 0 06-26-2009 10:40 PM
Converting to Black and White with my K200D Mark21 Photographic Technique 3 07-13-2008 10:11 AM



All times are GMT -7. The time now is 08:41 AM. | See also: NikonForums.com, CanonForums.com part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top