Image Sensor From Discrete Parts Delivers Glorious 1-Kilopixel Images

Chances are pretty good that you have at least one digital image sensor somewhere close to you at this moment, likely within arm’s reach. The ubiquity of digital cameras is due to how cheap these sensors have become, and how easy they are to integrate into all sorts of devices. So why in the world would someone want to build an image sensor from discrete parts that’s 12,000 times worse than the average smartphone camera? Because, why not?

[Sean Hodgins] originally started this project as a digital pinhole camera, which is why it was called “digiObscura.” The idea was to build a 32×32 array of photosensors and focus light on it using only a pinhole, but that proved optically difficult as the small aperture greatly reduced the amount of light striking the array. The sensor, though, is where the interesting stuff is. [Sean] soldered 1,024 ALS-PT19 surface-mount phototransistors to the custom PCB along with two 32-bit analog multiplexers. The multiplexers are driven by a microcontroller to select each pixel in turn, one row and one column at a time. It takes a full five seconds to scan the array, so taking a picture hearkens back to the long exposures common in the early days of photography. And sure, it’s only a 1-kilopixel image, but it works.

[Sean] has had this project cooking for a while – in fact, the multiplexers he used for the camera came up as a separate project back in 2018. We’re glad to see that he got the rest built, even with the recycled lens he used. One wonders how a 3D-printed lens would work in front of that sensor.

35 thoughts on “Image Sensor From Discrete Parts Delivers Glorious 1-Kilopixel Images

    1. Depending on whether you are trying to sense UV that can pass through an LCD screen, would it help to have a selectively masked lens and fewer receptor pixels? A bit like a resin 3D printer. Scan the pixel mask across the front of the camera and map the sensor readings at the rear to produce the image.

      1. You could make a pin hole camera and have an LCD on one end so you can scan the aperture around the sensor array. Then you do some Fourier transform magic and you can have a high resolution image from the low resolution sensor. I did that with a few LDRs a while back but I moved the camera because my sensor was only 16 pixels.

    2. What wavelength of UV? There are ccd and cmos sensors that are sensitive down into the 2-300nm range. The issue has always been optics. Standard optical glass becomes almost useless below 400nm so everything has to be made from fused silica. Nikon makes a nice UV-Nikkor lens that sells for about $6k… used. Good UV filters are needed as well which are expensive too.

      There are some enlarger lenses that can be used with various results.

      So, unless the board uses bare diodes and you rig up a silica optics train and filter you are better off using a modified dslr.

      1. I think the idea is that a pinhole camera doesn’t have any lenses. Unfortunately, the pinhole didn’t work for this project, because it didn’t let enough light through for his sensor, but perhaps a crazy bright UV source and/or a different sensor would let a UV pinhole camera work.

    1. +1 to the thermal camera idea, but thermistors react too slow, you’d be sitting there 10 mins getting the thermal mass of their encapsulation to match the subject temperatures. You’d need those spendy old PIR sensors, which still ain’t full motion video quick but would do.

      1. Yep, I agree. I built a 10X10 array of 0402 size thermistors and the frame rate ended up being about 1 frame every 10 seconds due to the thermal mass (the electronics could do about 80fps). It was a neat project though. I did not know about the 32 bit analog multiplexers this dude used at the time, so I may go back and re-design the project.

        Perhaps a thermo-chromic pigment bonded to a temperature controlled (thin) glass plate imaged by a standard HD webcam?

        1. In a 1960s Mechanix Illustrated, there was a brief piece in one of the “what’s new” pages about a thermal camera under development for the forces, that said it had a thermal sensitive liquid crystal, which was illuminated and then had output fed to image tube through image intensifier. Wish I could find that article again, it mentioned what the chemicals were, seemed enough to start experimenting. I was not terribly motivated at the time because it was mid/late 90s when I saw it and video equipment, especially stuff you eventually wanted to be portable, still cost enough to put it into the range of things to do when you had a LOT of spare money, vs on a whim.

  1. For those that remember, in the 70’s there were memory chips, that if you de-capped them and put a small lens over the die focused correctly, that you could actually read the memory as a pixel camera. Maybe a 2114? Not that this was a major feat considering what we have today, but back then the high sensor pixel array’s either didn’t exist, or were tens of thousands of dollars. Actually, I’m not sure they even had production digital cameras back then for any price. The biggest problem is they didn’t have any high capacity memory chips that I remember for this hack. Even the 2114 was 1024×1, they used those in the AD3M terminal – and they produced a LOT of heat.

      1. Yeah, I missed looking it up first. My experience was that I was in my teens back then and doing electronics for fun. This was an inexpensive item to play with then – although the price for #1 of those was $9.95 at Radio Shack.

    1. The Cromemco Cyclops, described in the February 1975 issue of Popular Electronics described this.

      Steve Ciarcia some years later did the same thing in Byte, but better density, and memory says he sourced the
      sensor from.a company that was making the modification and maybe selecting RAM that was better suited.

      When this came up here some years back, I did a bit of reading, and I vaguely remember reading then that Harry Garland or Roger Melen had been involved in research on sensors, but I don’t find anything now. I think professional work was already underway on primitive digital cameras, but I’m not sure. I do remember reading something in Popular Science or the like in the early seventies about very low density digital.cameras, and when “Westworld” hit tv, when they showed how the robot gunman saw the world, it was broadly pixelated and looked like the digital photo in the article.

      But I’m not certain if the Cycloos happened independently, or derived from professiinal work. Somehow Cromemco made the leap to uncover RAM and use it as a sensor

      1. I used to have all of the beginning issues of Circuit Cellar and many, many issues of the different electronic magazines like Popular Electronics, and like a moron – lent them to my nephew who was starting electronics. My Brother threw them all out on a house move. It really sucked. I was also a die hard fan of Popular Electronics at the time.

        1. @BabbyMac99, my Dad also had a lot of those, but I was the culprit in this case. I wish I’d been more careful with his mags. However, I came across these:
          https://www.americanradiohistory.com/Popular-Electronics-Guide.htm
          http://www.pong-story.com/mags.htm

          Unfortunately, Circuit Cellar and Elektor are not on the above sites, but with a bit of Googling I’ve been able to find them as well

          NB: If anyone has any of the missing issues, PLEASE scan and send to this guy!!

          1. Elektor was.at that first site a few months ago, but further down the page from fhe US hobby magazines.. Probably in the section about UK magazines. But maybe it’s been removed, that has happened.

            Byte is there, and Steve Ciarcia wrote a monfhly column there for years before starting his own magazine.

  2. Now incrasing the resolution: try to add an accuator to the array to make littles movements on X and y to put the pixel betwen the previous pixel , you will get 4 pixels insted of one. If it’s fast enough you can have a bigger resolution for only the time to take 4 pictures.

  3. The Incredible advantage of this single photo diode detector would be the enormous possible dynamic. If connected to (a lot of) DDC2256A, some can get 24-bit dynamic without any additional pre-amp.

    1. The Ti DDC2xx chips are amazing, but sooo expensive (have worked with DDC118 so far). The most important take-away from that line of products might be charge integration and photovoltaic mode instead of the photoconductive one demonstrated in the build.

      I’ve also been tempted to mess around with photodiode arrays, ever wondering whether CCD frontends (e.g. VSP5610) can be adapted ($5/1k).

Leave a Reply to NiHaoMike Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.