CNC & CAD in Buildling a Camera - Tri-Color Specifically

holmburgers · Mar 5, 2012

Lately I've been pipe dreaming about building a 3-color camera. This is not something I intend to do tomorrow exactly, but I still think there's some good that could come from talking about it and discussing the feasibility.

3-color cameras require a very precise optical path and placement of the reflectors/film-planes needs to be very exact. Since Computer Aided Design and Computer Numerical Control were in their infancy (if not still in the womb) when the last "one-shot camera" was made, it seems reasonable to assume that making one today might be much easier than it was then.

As I'm imagining it, the camera body could be assembled from a number of separate pieces, cut on some kind of CNC machine (?) and bolted together. The design would accomodate film holder placement; accurately spacing their distances.

The reflectors would then need to be mounted in frames that allow for fine adjustments and calibration after installation, with a provision to fit these frames in the aforementioned body.

The lens board & focusing standard would be fairly easy to make; requiring no great feat of precision. A well made bellows & focusing track would suffice. It'd be nice to connect this to a rangefinder as well.

Considering the need for low thermal expansion, excellent rigidity & an ability to be easily machined, what would be the best material? Steel, stainless steel, aluminum, brass, YbGaGe?

Knowing what kind of machines would be needed to do this work would help you choose a machine shop with the capability. Finding someone versed in CAD shouldn't be too difficult either. I've got at least 1 engineer friend I could bribe!

E. von Hoegh · Mar 5, 2012

holmburgers said:
Lately I've been pipe dreaming about building a 3-color camera. This is not something I intend to do tomorrow exactly, but I still think there's some good that could come from talking about it and discussing the feasibility.

3-color cameras require a very precise optical path and placement of the reflectors/film-planes needs to be very exact. Since Computer Aided Design and Computer Numerical Control were in their infancy (if not still in the womb) when the last "one-shot camera" was made, it seems reasonable to assume that making one today might be much easier than it was then.

As I'm imagining it, the camera body could be assembled from a number of separate pieces, cut on some kind of CNC machine (?) and bolted together. The design would accomodate film holder placement; accurately spacing their distances.

The reflectors would then need to be mounted in frames that allow for fine adjustments and calibration after installation, with a provision to fit these frames in the aforementioned body.

The lens board & focusing standard would be fairly easy to make; requiring no great feat of precision. A well made bellows & focusing track would suffice. It'd be nice to connect this to a rangefinder as well.

Considering the need for low thermal expansion, excellent rigidity & an ability to be easily machined, what would be the best material? Steel, stainless steel, aluminum, brass, YbGaGe?
Knowing what kind of machines would be needed to do this work would help you choose a machine shop with the capability. Finding someone versed in CAD shouldn't be too difficult either. I've got at least 1 engineer friend I could bribe!

Try Invar. The alloy you mentioned is probably useless for structural stuff. Carpenter sell Invar in rounds, I don't know where or if you could get sheet and other shapes.

holmburgers · Mar 5, 2012

Thanks E.,

I was totally joking about YbGaGe of course... but I appreciate you mentioning this alloy.

E. von Hoegh · Mar 5, 2012

Found a link for Invar

http://www.thomasnet.com/products/alloys-invar-36r-1198803-1.html

paulrocon · Mar 5, 2012

Invar would be great, but it'll be expensive and wont come in many forms. Im also going to take a wild guess and say that itll be a bit difficult to machine since as far as I know nickel isnt known to improve machinability unlike lead in 12L14, etc.

I would recommend a good engineering plastic. Ultra-high molecular weight polyethylene (UHMW PE), nylon, or acetal come to mind. There are cheaper plastics like fiberglass reinforced resins and ABS but they aren't meant for machining.

If you get serious about this defiantly do your homework and compile a lot of data. I may be interested in helping from the CAD standpoint as Im a mechanical design drafter by trade and have access to both 3D and 2D CAD systems.

richard ide · Mar 5, 2012

I think that there is no need to use esoteric alloys with the machining problems and costs involved. If you use the thermal coefficient of expansion for a metal; calculate the change in dimension of your longest piece say from 20 degrees celsius to 35 degrees. There will be such a small dimensional change that it probably is not a factor. I had an enlarger with a 12' aluminum column (bed) and did not need to take temperature into account between summer and winter even with precision work.

holmburgers · Mar 5, 2012

I really like the idea of a plastics. Low weight would be a boon and machining would likely be easier and cheaper. I'll look into those..

This question probably deserves a huge answer, but just for the sake of asking, what kind of data would you need to design something like this? If we understood the optical needs, then I could hand-draw something encompassing the basic design once I had reasoned it out in my noggin. It'd be nice to get a more in-depth look at the inside of one of these cameras and work from that.

FYI, the reflectors are not actually glass mirrors, but are ultra-thin pellicles that have been vacuum deposited with some kind of reflective coating. National Photocolor, a company who actually manufactured these cameras back in the day, still makes such pellicles to this day. You'd need 2, and a 5x7" one in a frame is about $250. A smaller one might even be less.

I think a camera designed to take 120 film backs would be an awesome first start; though 4x5" might make more sense at first.

polyglot · Mar 5, 2012

Front-surface partial mirrors get cheaper as you go smaller; I'd suggest that medium-format is a good place to start. In particular, using RB backs gets you a simple (graflok) mounting system for your films.

The real problem (unless I'm missing something obvious here) is that the mirrors will take up a lot of room in the optical path (about twice as much as in a normal SLR, and that's without any SLR viewfinder!), which means you're limited to quite-long lenses. It will be difficult-to-impossible to build a wideangle camera, and you'll want to have lenses with a registration distance as long as possible. You may only be able to support quite narrow fields of view.

Quit worrying about thermal expansion; such details are an irrelevancy.

You can get free CAD programs. Hell, you can use google sketchup or Inkscape with zero training and a little practise. Design the camera to be built entirely from shapes laser-cut from flat material, make yourself some drawings, send them off to ponoko and they'll send you all the parts in a couple of weeks. If complex 3D stuff (cams, etc) are required, they can be 3D-printed.

michaelbsc · Mar 5, 2012

What I have thought about, when looking at the old patents to find the angles and dangles for the reflectors, is the majority of the camera body requires very little precision, at least not for prototyping.

A large plastic clsmshell with a mounting pad, then a holder assembly with precision turnscrew adjustments.

We'll need three film backs, also with some kind of 3-point adjustment for planar perpendicular adjustment. But the 3 film gates can be rough with the precision limited to the mounting plate.

What I was sort of hoping would materialize is a community effort - and I nominate Chris as the librarian - to put together an Open Source Hardware Project. Hopefully one that will live on well beyond all of us into the future.

I suggest the project name K-15, as the successor to true three layer color process using B&W emulsions.

MB

michaelbsc · Mar 5, 2012

richard ide said:
I think that there is no need to use esoteric alloys with the machining problems and costs involved. If you use the thermal coefficient of expansion for a metal; calculate the change in dimension of your longest piece say from 20 degrees celsius to 35 degrees. There will be such a small dimensional change that it probably is not a factor. I had an enlarger with a 12' aluminum column (bed) and did not need to take temperature into account between summer and winter even with precision work.

My suggestion, once you get beyond the initial design experiments, would be to use a vacuum formed plastic housing.

A determined tinkerer can do this in a kitchen.

Google vacuum forming and check out a few you-tube links.

michaelbsc · Mar 5, 2012

holmburgers said:
I really like the idea of a plastics. Low weight would be a boon and machining would likely be easier and cheaper. I'll look into those..

This question probably deserves a huge answer, but just for the sake of asking, what kind of data would you need to design something like this? If we understood the optical needs, then I could hand-draw something encompassing the basic design once I had reasoned it out in my noggin. It'd be nice to get a more in-depth look at the inside of one of these cameras and work from that.

FYI, the reflectors are not actually glass mirrors, but are ultra-thin pellicles that have been vacuum deposited with some kind of reflective coating. National Photocolor, a company who actually manufactured these cameras back in the day, still makes such pellicles to this day. You'd need 2, and a 5x7" one in a frame is about $250. A smaller one might even be less.

I think a camera designed to take 120 film backs would be an awesome first start; though 4x5" might make more sense at first.

Chris. I found the old patents to some of these once, and the drawings are pretty complete. I'm not sure where I dug them up, but it was with search engines.

I found those drawings very instructive. I'll see what I might have when I get a chance. But search away.

The National Photocolor connection is excellent. Much better than the stuff I tracked down. $250 is awesome.

holmburgers · Mar 5, 2012

Thanks guys, glad to have such input!

Google sketchup... ponoko.. what is this, the 21st century?! That's some crazy stuff; amazing what you might be able to build from an armchair. Duely noted.

Michael, you're thinking exactly as I am about mounts with precision turn-screw adjustments. The body can get you in the ballpark, the fine adjuster get you on the money.

Thermal expansion might've been a red herring, but I just meant to ask about materials in general, and CMB did once mention that Bermpohl tri-color cameras (made of teak) did have this problem. Plastic's the way to go...

Good call on focal lengths. This is an issue that these cameras tried to address in the past. I'll have to do some digging, but there were retrofocus lenses to somewhat overcome these difficulties... I'm gonna post a list of cameras and their lenses here in a few minutes...

. . .

Ok, here's a list of cameras and their lenses from Issue 15, Volume 3 (1942) edition of The Complete Photographer

Curtis Color Scout - 2.25x3.25" - 7.5" Goerz Dogmar f/4.5 (190mm)
Curtis Color Master - 5x7" - 14" Eastman Ektar f/6.3 (355mm)
McGraw Hand Camera - 6.5x9cm (2.5"x3.5") - 140mm Goerz Dogmar f/4.5 (5.5")
McGraw Professional - 5x7" - "none"
National Daylight (or Tungsten) Feather-Weight - 3.25x4.25" - 8.25" Goerz Dogmar f/4.5 (210mm)
National Studio Deluxe - 5x7" - 12" Goerz Artar f/9 (305mm)

A much earlier model:
Bermpohl Naturfarbenkamera - 9x12cm (3.5x4.75") - Hugo Meyer & Co. Gorlitz Doppel Plasmat 21.5cm f4

Sure enough, these are some pretty long lenses. Anyone care to find the film-diagonal:focal-length coefficient?

update:

This is promising, a Curtis Color Stellar 133mm f/5.3 on 4x5". here and here

michaelbsc · Mar 5, 2012

I tried to think through the focal length issue once.

The geometry is difficult to conceptualize in my head. And I'm afraid I've never had time to really analyze it. But you could get a kinda short lens by bringing two of the film gates forward. But you have to stay out out of the image field. The camera would be unwieldy and awkward. And short is a relative term. You probably can't get 90mm.

willrea · Mar 5, 2012

At the camera swap meet up here someone was displaying a Curtis. I'm about 90% sure it was a 4x5 and looked pretty different than the Stellar shown

holmburgers · Mar 5, 2012

Was it the lens or camera that looked different?

polyglot · Mar 6, 2012

That Curtis Stellar is the obvious arrangement with two mirrors in series, everything close to being at right-angles. Easy to build without much calculation.

However if you're happy for it to be "crooked", you can do a couple of things:
- bring the film planes forward toward the lens and at an angle, which means that the mirrors are steeper, which means less optical path length consumed, which means you can have shorter lenses
- put in three mirrors, which gives you four outputs and therefore a TTL viewfinder!

Of the first point, consider the Red film plane of the Curtis above, swinging it around to lie immediately to the left of the lens. You can imagine that the first mirror is steeper, which means you can bring the second mirror closer. Getting equal path length (for equal focus) means that the red film will actually be a long way out the front, next to the lens.

Of the second point, consider three mirrors in a tree topology. First mirror splits light into two beams, each secondary beam hits a second mirror, which splits it into two further beams. Net result is four copies of the light from the lens: one for each colour and one for the viewfinder. Each beam of light has been through exactly 2 mirrors and so has lost (assuming 50% silvering) 2 stops. Given that you need two mirrors in series for a basic tricolour camera anyway, this design shouldn't have any added flange distance (it makes the scrunching process described above harder though) and means you get a viewfinder.

Steve Smith · Mar 6, 2012

I'm not sure how much help I can be from the other side of the planet but I do use CAD and CNC. Our CNC machine is intended for drilling and routing PCBs but I use it mainly to cut plastic sheet up to 10mm thick to make test fixtures and assembly jigs.

I use it to make cameras too - see the link below.

I'm happy to do what I can if it is of any help.

Steve.

michaelbsc · Mar 6, 2012

Another point to remember, which isn't really big, is that if the images do not have the same number of reflection then odd will be reversed from even.

Also the filter factors will not be the same so 50% may not be our friend. 67/33 might prove useful.

holmburgers · Mar 6, 2012

polygot, I think Bermpohl utilizes a slightly "crooked" design like you're referring to. http://www.vintagephoto.tv/bermpohl_img.shtml Would you agree?

Steve, thanks for your offer of assistance. I think we're a long ways from actually doing anything, but this kind of group effort is just the type of thing that could make this project a reality someday.

Michael, I hadn't thought of that... does that mean in the typical arrangement, 2 of the images will actually not be mirrored, and the remaining plate, being struck by the % of light that was never reflected, only transmitted, would give a typical mirror image negative? I guess so..

If we pretend filter factors are all equal, then a 33/66 mirror at the front, followed by a 50/50 mirror would give equal exposures on all plates. 33% to the red (in the case of the Curtis above), and the remaining 66% split down the middle to green & blue. However, there might be a more novel arrangement if we consider filter factors.

Tom Kershaw · Mar 6, 2012

Chris,

Would this camera be used for making negatives for colour alternative processes?

Tom

holmburgers · Mar 6, 2012

That's right Tom. It seems we're going back in time with regards to color film, and so someday I figure this will be the only way to obtain analog color capture.

By the way, here's some info on UHMW-polyethylene and other plastics. Sheet sizes are available, and machining appears to be easy:
http://www.garlandmfg.com/plastics/sendcad.html
http://k-mac-plastics.com/uhmw-sheet-general-purpose.htm
http://www.plasticsmag.com/features.asp?fIssue=Sep/Oct-03&aid=3862
http://www.redwoodplastics.com/products/uhmw-polyethylene/recycled-uhmw-repro-uhmw/

Steve Smith · Mar 6, 2012

holmburgers said:
I think Bermpohl utilizes a slightly "crooked" design like you're referring to. http://www.vintagephoto.tv/bermpohl_img.shtml

They look fantastic.

There are enough ways to go wrong with just one sheet of film. I can't begin to calculate the opportunities for error with three!

Steve.

holmburgers · Mar 6, 2012

Let's not let the potential for errors deter us. Basically the only things we have to worry about are distances and angles. If we design-in the ability to affect these things, it's not that daunting.

A note about Bermpohls; since they use glass mirrors, refraction errors are introduced when the light strikes these at angles. You'll notice that two of light paths have slanted glass in front of the film-holders to correct for this. Pellicles though, are so thin that they don't introduce any appreciable error.

holmburgers · Mar 6, 2012

Ok, some more on the Curtis Color Stellar from the author of that website.

-> Dead Link Removed

Curtis' solution to this problem was to design a retrofocus wide angle lens. This lens was named the Curtis Color Stellar. It has a focal length of 133mm (5.25 inches), but its back focal distance is approximately 210mm (8.25 inches). Thus the lens could be positioned 8.25 inches away from the focal plane, but it possesses the angle of view of a 133mm wide angle, which in this case is 60 degrees. Prior to introduction of the Color Stellar lens the standard lens on Curtis 4x5 cameras was 8.25 inches, with an angle of view of 38 degrees. It's obvious that 38 degrees is too tight for a variety of news images.

holmburgers · Mar 6, 2012

US Patent - Des. 115,165 - Curtis Color Camera

Not very exciting though. Haven't found the goods yet.

edit: Adding some as I come across them; haven't read any yet though.
USP 1,895,555 - Emil Reckmeier 1933 - (Dead Link Removed)
USP 1,951,896 - Bermpohl 1934

CNC & CAD in Buildling a Camera - Tri-Color Specifically

holmburgers

Attachments

E. von Hoegh

holmburgers

E. von Hoegh

paulrocon

richard ide

holmburgers

polyglot

michaelbsc

michaelbsc

michaelbsc

holmburgers

Attachments

michaelbsc

willrea

holmburgers

polyglot

Steve Smith

michaelbsc

holmburgers

Tom Kershaw

holmburgers

Steve Smith

holmburgers

holmburgers

holmburgers

CNC & CAD in Buildling a Camera - Tri-Color Specifically

Attachments

Attachments

Privacy & Transparency

Privacy & Transparency