Page 1 of 5 123 ... LastLast
Results 1 to 10 of 41

Thread: DIY Drum Scanner?

  1. #1

    Join Date
    Apr 2012
    Posts
    16

    DIY Drum Scanner?

    Hi,

    since some years, the DiY DSLR scanner project is popping in my mind again and again, but actually never headed of to start it. I really like the idea, I can imagine I like the handling. I am a bit concerned regarding the sharpness / DOF thing and mainly to get a good light source to scan color sheets / negatives.

    During reading another post regarding the v850 scan quality (again thinking about buying one...), somebody mentioned advantages of drum scanners again... Sure, old, used, nobody knows how long you get spare parts. List goes on and on. No option for me. Definitely.

    Then it struck me. Is it possible to build a drum scanner on your own as open source project?

    What do you need:
    • A rotating glass tube of good optical quality (expensive for sure).
    • three photomultiplyers
    • a light source (in the center of the tube?)
    • a photomultiplier moveable along the center axis of the glass tube
    • electronics to read out the stream of the data from the photomultiplier
    • software to display the data and control the whole thing


    For sure, this a completely naive view of the whole project. I have no idea where one could get a glass tube of needed quality and the price point the same goes for the photomultiplier... Nether have I used a drum scanner, nor do I know exactly how one works...

    As I am software developer and have some experience with tinkering around, I could cover this parts of the project.

    Does anybody here know enough about the technical details about the glass tube, the photomultiplier and electronic side of the story to decide whether this whole idea is crazy or feasible?

  2. #2

    Join Date
    Jun 2015
    Location
    Glasgow
    Posts
    1,009

    Re: DIY Drum Scanner?

    Quote Originally Posted by mongole View Post
    Hi,

    since some years, the DiY DSLR scanner project is popping in my mind again and again, but actually never headed of to start it. I really like the idea, I can imagine I like the handling. I am a bit concerned regarding the sharpness / DOF thing and mainly to get a good light source to scan color sheets / negatives.

    During reading another post regarding the v850 scan quality (again thinking about buying one...), somebody mentioned advantages of drum scanners again... Sure, old, used, nobody knows how long you get spare parts. List goes on and on. No option for me. Definitely.

    Then it struck me. Is it possible to build a drum scanner on your own as open source project?

    What do you need:
    • A rotating glass tube of good optical quality (expensive for sure).
    • three photomultiplyers
    • a light source (in the center of the tube?)
    • a photomultiplier moveable along the center axis of the glass tube
    • electronics to read out the stream of the data from the photomultiplier
    • software to display the data and control the whole thing


    For sure, this a completely naive view of the whole project. I have no idea where one could get a glass tube of needed quality and the price point the same goes for the photomultiplier... Nether have I used a drum scanner, nor do I know exactly how one works...

    As I am software developer and have some experience with tinkering around, I could cover this parts of the project.

    Does anybody here know enough about the technical details about the glass tube, the photomultiplier and electronic side of the story to decide whether this whole idea is crazy or feasible?
    The most difficult bit is the software - how to correctly reassemble the sampled data across (potentially) multiple simultaneous files while preventing errors & resolving questions of how many 'samples' are taken & averaged for every pixel in the final file - all at a rate of several thousand samples per inch. The rest is basically some fairly high precision mechanical, optical & electronic engineering. You'd need access to some pretty serious machine tools to do it right. We're talking about spinning a thick wall perspex drum at up to quite high RPM's with repeatable precision. Power supplies should be as standardised as possible. Pretty much any computer today will run rings round any onboard hardware they used in the 90s to try & speed things along - add a thunderbolt or USB-C after the Analogue-Digital-Conversion stage & you'd have something much more future resistant.

  3. #3
    Resident Heretic Bruce Watson's Avatar
    Join Date
    Sep 2003
    Location
    USA, North Carolina
    Posts
    3,362

    Re: DIY Drum Scanner?

    Quote Originally Posted by mongole View Post
    Does anybody here know enough about the technical details about the glass tube, the photomultiplier and electronic side of the story to decide whether this whole idea is crazy or feasible?
    It's acrylic, not glass. And the whole idea is completely crazy, but of course it's feasible. Somebody did it in the first place, no reason you can't do it over again now.

    An accomplished mechanical engineer would be a big help. Without one I doubt you can make a functional drum scanner.

    The name of the game is accuracy and precision (these are not interchangeable concepts). That is, you have to read each pixel in real-time as they pass in front of the light beam going through the drum, the film, and the overlay to the PMTs. Each pixel has to be read by each of the PMTs, passed through the log amps to the A2Ds, and the digits have to make it into the file, all in real-time as the drum turns. The drum has to index to the next line (that is, it has to move *and* settle) in the time it takes for the drum to make the partial turn from the end of the film back around to the start of the film. There's no going back, no do-overs, no repeat scans. One and done. Each and every pixel. Thousands of pixels each turn of the drum. Just as fast as the drum can turn.

    Really, the amount of money you'd spend on just the mechanics would be 10x what you can buy a used drum scanner for. Then at least you have all the mechanical parts. If you want to redesign all the circuits and the software, have at it. But it's probably at least a couple of orders of magnitude more difficult than you seem to think.

    First thing you do is, make a formal functional specification for it. If you can't do that, you can't make a drum scanner. You can't solve a problem that you can't define.

    All that said, I think a modern drum scanner would be a pretty interesting thing to do. It would be especially interesting to replace the light source with an LED. My drum scanner uses xenon bulbs, very expensive, very bright (I have no idea how many lumens), and very hot, but a beautiful clean/clear balanced white light. But I would imagine one could do just about the same thing with an LED these days, if one can manage the flicker.

    [EDIT]
    Better yet, why not use a laser, or three. RGB lasers directed to converge at the pixel (spot on the film you want to measure) and then diverge to three PMTs. This could get interesting...
    [/EDIT]
    Last edited by Bruce Watson; 28-Feb-2018 at 13:27.

    Bruce Watson

  4. #4
    Pali K Pali K's Avatar
    Join Date
    Aug 2014
    Location
    MD, USA
    Posts
    1,397

    Re: DIY Drum Scanner?

    Bruce covered most everything I can think off but I'll add that another complexity would be the fine precision that they require for everything. The drum spins at ridiculously fast speed on my Scanmate and Tango and the focusing system has to be microscopically precise. If you really want a drum scanner today, there is nothing better than getting one from Karl Hudson (http://hudsongrafik.com/) and letting him take care of it for you. You can alternatively get one from Michael (http://www.scansolutionsonline.com/) if you are in the market for a table-top drum scanner or a high-end flatbed that is near drum scanner quality. IMO, both options will cost you significantly less than trying to make one from scratch.

    Pali

  5. #5

    Join Date
    Jan 2011
    Posts
    123

    Re: DIY Drum Scanner?

    Quote Originally Posted by Bruce Watson View Post
    You can't solve a problem that you can't define.
    What an awesomely pragmatic approach. Bravo !!

    Great project though to sink one's teeth in ...

  6. #6

    Join Date
    Apr 2011
    Location
    Hemel Hempstead, UK
    Posts
    543

    Re: DIY Drum Scanner?

    Oddly enough, I was thinking about the very subject only this month (something to distract me from the day job)... and it occurred to me that the critical point of a drum scanner is, basically, an arbitrarily small spot of very bright light and a single (I was thinking monochrome as a first step) sensor. It doesn't actually need a drum(!) since as far as I can see the purpose of the drum is to generate a raster.

    The rotation provides a scan in one axis by spinning, and a scan in the other by either the light source or the drum moving axially, perhaps on a screw as the drum scans?

    (I've never seen a drum scanner; this is extrapolation based on descriptions I've read and good knowledge of broadcast technology.)

    One could replicate both the scan axes with a flat image and a stepper system, as in a standard flatbed scanner. The pixel count would be set by the minimum step size.

    A laser LED and focussing lens could be arranged to produce a variable spot size; the sensor needs only to be large enough to accept the largest spot. Timing for the ADC is not critical since the spot doesn't move until the ADC is complete - though there may be a time limit on how much energy is transferred to the film by the laser. It would be a shame to set fire to your image!

    The software is not that complex (but then, that's the sort of stuff I do in the day job). The data appears one sample per pixel, just dump it into a file in real time. Post process the file if desired. I'd probably aim for a 16-bit sample and adjust the brightness of the LED to give as close to a full-scale sample across the image as possible.

    I'd probably arrange things so that the laser and sensor stay put, and the film moves on a platten, but it could be done the other way around.

    Just ideas...

    Neil

  7. #7
    jp's Avatar
    Join Date
    May 2009
    Location
    Maine
    Posts
    5,628

    Re: DIY Drum Scanner?

    There is probably CNC stepper/sensor accessories for controlling generic lathes that could operate the mechanical aspects of a diy drum scanner. Probably a lathe could be the basic of the mechanics. The hardware for controlling those is not going to go away; it will probably get more interesting. I'd have a index mark parallel to the film that started the recording for each row. I don't use a drum scanner but I would certainly at least inspect a broken one before proceeding to build one as it's probably full of clever tricks you won't want to reinvent. If the film were rotated the same amount as the pitch of the lead screw, it would never have to stop feeding and could turn slowly and continuously while scanning if the sensor were in place of the tool post/turret.

    Honestly, I'm pleased with my epson flatbed for LF. Something better for MF would be nice though if it were as convenient.

  8. #8

    Join Date
    Apr 2012
    Posts
    16

    Re: DIY Drum Scanner?

    Yes, it's crazy! I realized in meantime as well. If you want to have a somehow future proof solution, buying a drum scanner and rebuild the electronics and write the software would be the way to go.
    Though I have no idea how much a photomultiplier costs, up to now and where to get it.

    This page showed me how a drum scanner is build: http://www.terrapinphoto.com/drumscansaga/
    Also this thread gave some real valuable information: https://www.rangefinderforum.com/for...d.php?t=134187

    What I do not understand up to now is: Why needs the drum rotate that fast? Is it out of rpm consistency? The faster it rotates, the less the rpm fluctuates?

    Is it really necessary to read the data in a constant stream, if the AD conversion is not fast enough? Would it not be enough to measure the time of one rotation, take the desired resolution per perimeter, divide by let's say 10 and sample every 10th pixel in the first round. Then shift one pixel and sample every 10th+1 pixel and so on. You get the idea. I guess it's not as accurate as do it in one stream.

    But I don't think you need to reposition the light in the gap where there is no slide. Just move on one step and wait some rotations. Would be beneficial to reduce vibrations due to reposition anyway...

    I am a bit amused Neil, you as software develope also do not seem to see it as undoable. Maybe it's a trait of software developers (as I am one myself) to be ignorant enough in the other technical fields to even think about to try it!

    Let's see where this will lead. If it even will lead to somewhere.

    Thanks for all your valuable input!

    Greets,
    Andreas

  9. #9
    Resident Heretic Bruce Watson's Avatar
    Join Date
    Sep 2003
    Location
    USA, North Carolina
    Posts
    3,362

    Re: DIY Drum Scanner?

    Quote Originally Posted by mongole View Post
    What I do not understand up to now is: Why needs the drum rotate that fast? Is it out of rpm consistency? The faster it rotates, the less the rpm fluctuates?
    At 600 rpm, my scanner takes an hour+ to scan a 5x4 sheet of film at around 11x enlargement. IIRC (and it's been years so I might not remember correctly) it was something less than 4000 dpi. Slower rpms mean longer scan times. How long are you willing to take?

    BTW, this is what I mean by writing a formal functional specification. Specify what you are trying to do before you try to decide how to do it.

    Quote Originally Posted by mongole View Post
    Is it really necessary to read the data in a constant stream, if the AD conversion is not fast enough?
    If you want to complete the scan in a reasonable amount of time, then yes, it really is necessary. But it's a meaningless question. If they could already do this in 1996, why couldn't you do it now? I guarantee that today's tech is faster than 1996 tech.

    Bruce Watson

  10. #10

    Join Date
    Apr 2011
    Location
    Hemel Hempstead, UK
    Posts
    543

    Re: DIY Drum Scanner?

    Quote Originally Posted by mongole View Post
    I am a bit amused Neil, you as software develope also do not seem to see it as undoable. Maybe it's a trait of software developers (as I am one myself) to be ignorant enough in the other technical fields to even think about to try it!
    Greets,
    Andreas
    Call me a generalist, Andreas; I'm a software developer working on the bare metal for domestic products - but I also develop the hardware and previously developed deep-drilling robots for bore-hole guidance and logging working five kilometres underground... and before that, I worked as a broadcast engineer (at many levels) in the BBC for over thirty years.

    Photomultiplier tubes are available cheaply - under thirty pounds - second hand parts on eBay; you can probably pick up a high voltage supply at the same time - but the design of such isn't difficult. But as I said in my post, I think that current optical sensors are probably good enough to use directly. The trick is to get the sensor biased correctly to ensure as large and as linear an output as possible between solid silver and clear film base. Everything after that you can do in software. And since you control the speed of the scan, you also control the speed at which data need be output and stored.

    At a first thought, in order of difficulty:

    - analog design for a clean signal of sufficient resolution and dmax, and a suitable DAC
    - mounting and moving the image
    - handling, processing, and storing the data

    Neil

Similar Threads

  1. Replies: 14
    Last Post: 21-Feb-2018, 22:33
  2. DIY DSLR Scanner vs Drum Scanner
    By JasonF in forum LF DIY (Do It Yourself)
    Replies: 39
    Last Post: 20-Jan-2015, 16:35

Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •