Starting at the beginning with Mirror Screws

[Panrock]

A few brief comments.

There is the same number of reflective slats as there are lines in the picture. One thing that puzzled me at the start was that mirror screws all look like they are 1½ turn corkscrews! They aren't really. This is appearance you always get when you take a series of reflective slats/edges and stack them round a full 360 degrees.

My system was conceived as dual standard from the start: either 60 or 120 lines. At 60-lines, the slats were bunched together in pairs (giving half the number of picture lines of twice the thickness and an overall picture the same height) and at 60-lines each pair was displaced with respect to its neighbour by twice the angle of that at 120-lines.

I also used a pin and hole idea to index the positions of the slats. However, because my system needed dual standard indexing, there were three equi-distant holes to take the pins in each slat. To minimise any angular positioning errors caused by hole/pin positioning tolerances, I placed the holes toward the outside of each slat, rather than nearer the centre as Andrew has done.

Steve O

[albertMunich]

Sorry -my mistake. I took your comment about the 180 line screw as a comment on my drawing. And you meant the photo of the historic screw. I Screwed up there. (Sorry can't resist the pun..)
I remember reading somewhere that one gets 120 lines for 60 slats because there are two full turns of the screw. But that was probably mis representation of your 60/120 line double system layout.

Sorry for that!

The TKD patent explains the fact that you are seeing double images as you get closer to the mirror screw. I think I begin to get around on this....

BTW your website is marvelous.

[Panrock]

And you meant the photo of the historic screw. I Screwed up there. (Sorry can't resist the pun..)

You've clearly got a pretty impressive command of English!

The TKD patent explains the fact that you are seeing double images as you get closer to the mirror screw. I think I begin to get around on this....

Yes, and you get more and more images side-by-side (with horizontal scanning) as you go closer...

BTW your website is marvelous.

Thank you! Glad you enjoyed it.

Steve O

[albertMunich]

Hi Andrew,

thanks for your explanations. I have tried openScad and I say I'd rather sculpt a mirror screw out of clay than trying math formulas to make anything. I Was always the worst in maths!
But I will look into it, I must have a copy of OpenScad somewhere on my PC.
I will look ASAP into your Arduino build log -. I started with the standard -no computer- system because I wanted to start from the beginning and be as analog as possible. I felt cannot run before I can walk!
I have looked at the NBTV website years ago and downloaded the various programs for making/simulating NBTV signals from a PAL source -and had some fun making simulated 30 line pictures. Now I feel its time to give things a whirl- I'll order a nipkow disk from the club when I get back home- I write from my holiday in France- thats why I have the time to do some reading. I have a lot of other stuff waiting at home so it might be slow going at first.
I have 2 UltiMaker Original 3 D printers at home. Your help will be most welcome! I think I'll need it.

I include for all interested parties a picture of the Emil Mechau lens crown scanner from the late Thirties- it could run film (two projectors for switchover) and scan a person.

and the car wheel sized lens wheel using 270 (!!) Leitz microscope lenses and spun at 3000 rpm. It was "duck and cover" at the lab when they spun this monster up for the first time.
But it worked flawlessly.

mechau-buch-bilder-106-Linsenkranz.jpg (41.76 KiB) Viewed 14207 times

The gears inside...

mechau-buch-bilder-108-getriebe.jpg (70.05 KiB) Viewed 14207 times

Just to show to what lengths they went with the all mechanical approach. Of course all of this was wiped out by the all electronics systems only a few years later. The more I look into this the more I respect the engineers and craftsmen of the time.

The images come from Gert Redlich's excellent German website
http://www.fernsehmuseum.de
where I have spent hours looking thru the various stories he tells. But the images alone are interesting too.


Albert

[Klaas Robers]

Andrew, how do you think to get an optically flat edge to the 3D-printed slats? Before metalizing they should be optically flat, that is a surface roughness of about 10 nm or better. A used method is to place all slats in the same orientation on a spindle, and then polish the block that you made. But with your positioning stubs it is no more possible to make a polishing block of them. And they should become flat as a mirror.

[Panrock]

And they should become flat as a mirror.

That is true. I learned the hard way! Moreover, if you look at a plain white raster, you can see every tiny imperfection in the surface, exactly as when the slats are locked together in 'polishing block' form.

Steve O

[Andrew Davie]

Andrew, how do you think to get an optically flat edge to the 3D-printed slats? Before metalizing they should be optically flat, that is a surface roughness of about 10 nm or better. A used method is to place all slats in the same orientation on a spindle, and then polish the block that you made. But with your positioning stubs it is no more possible to make a polishing block of them. And they should become flat as a mirror.

I can think of two solutions. Firstly don't print the pegs, but just holes - and add the pegs later. Or, secondly, print a "spacer" module which provides holes and pegs in suitable positions to mount all of the slats "flat" for polishing. That is, you'd have slat/spacer/slat/spacer/slat/spacer. That way you could join them flat, polish, discard the spacers and assemble in the spiral.

[albertMunich]

Well I think the demand for optical perfection cannot be fulfilled by an ordinary 3d printed surface. Klaas- You just hammered home the fact that this has to be precise-period.Or it's junk.

And the stuff I use (PLA) cannot be polished at all. It just heats up and smears all over the place. A mess! There are metal filled PLA variants, where a certain amount of metallic powder is added to the original mixture. But I have never tried it because it tends to clog up the printer nozzle. You have to have special ones for this purpose. My own printers are the first series Ultimaker brought to market, their newer ones have provisions for changing the nozzle in a few seconds, in mine its a chore and you risk damaging the machine. So I'm not ready to try that.
This leaves one with the idea to replace the surface in question with another one. Small metal strips glued to the 3d surface in question...?and if its not done perfectly, if the glue is too thick on one side and too thin on the other, there goes your alignment. Same Catch 22 situation.

My original "shot from the hip" idea was that in order to profit from 3d printing at all, the critical junctions on the thing should be done in one piece. This way you can have the alignment done by the computer. If that does not work then its a serious question why to use 3d printing at all? Only for the weight reduction...? The other question is if a professional service such as Shapeways could do the thing in metal. Laser sintering comes to mind -And then at what price? Could it stand the centrifugal forces?
The other problem that bothers me is that I still cannot get a simulation of the light rays done in the computer. A question to The Guardian Of The Screw: if you have the time, could you perhaps PLEASE make a short video from the right viewing distance, light line turned on without the modulator, and rotate the screw a few turns by hand? If I can see what happens in slo motion my slow mind might be able to better understand it....

Oh...The appropriate soundtrack for this could be "The Turn Of The Screw" by Benjamin Britten (ouch...!)

[Andrew Davie]

And the stuff I use (PLA) cannot be polished at all. It just heats up and smears all over the place. A mess! There are metal filled PLA variants, where a certain amount of metallic powder is added to the original mixture. But I have never tried it because it tends to clog up the printer nozzle. You have to have special ones for this purpose. My own printers are the first series Ultimaker brought to market, their newer ones have provisions for changing the nozzle in a few seconds, in mine its a chore and you risk damaging the machine. So I'm not ready to try that.

Personally I think it's possible but you have to do it slowly and carefully. In any case, there's always ABS, and your printer should handle that fine, no? And ABS can be given a very smooth surface with a brushing of acetone. That's what I'd be trying - although abrasives might be OK too - it has a higher melting point than PLA (not by much though). The reason to use a 3D printer at all is that you do it at home

[albertMunich]

Another patent by TKD

This one details the construction of a mirror screw for interlaced direct viewing. They had the same problem and propose making a screw in one piece by creating polished steps with a profile. This way there is an interlace or a overlap between lines. Or at least having two steps of the screw in one block with dynamic balance. CNC instead of 3d printing?
Fig. 8 is the body of a mirror screw where the mirror consists of either tightly wound thin steel wires or of separate small polished straight steel bars. Fig. 9 shows the cross sections.
They felt the crunch of electronic TV at their heels and were ready to try everything.

Mirror screw construction.pdf

(816.14 KiB) Downloaded 550 times

The original THE TURN OF THE SCREW was a horror(!) novel by Henry James.....

328px-The-Turn-of-the-Screw-Collier's-1A.jpg (47.47 KiB) Viewed 14188 times

[albertMunich]

Hi Andrew,

my printer does not have a heated bed and the nozzle will only go to 240-250 degrees. So I have not tried it yet.

[Andrew Davie]

my printer does not have a heated bed and the nozzle will only go to 240-250 degrees. So I have not tried it yet.

The heated bed is an issue - but try making a "slurry" by mixing acetone and small (rice-size) cutoff pieces of ABS filament - and mix to a thickish liquid (constantly stir with a small brush). Then when mixed, brush onto the surface of your print bed. This will form a binding layer onto which printing ABS could/should stick very well. To remove, use pure acetone and a cloth. Be careful - acetone is highly flammable. Wear gloves, facemask, eye protection!

I find that ABS prints extremely well around 220-230 degrees, so you're good there.

[albertMunich]

Bingo- I found the patent that looks like they used this process for the screw I photographed in the museum.
The text says that they ran into problems when glueing or soldering the slats to each other. Too much lateral pressure caused the slats to warp- So they propose a way to insert a small screw into a hole thats threaded on the outside of the slat. They say the threads can be made while the slats are still a packet. The lateral pressure under which the entire assembly is held is not so high as to cause warping of the slats. The other idea seems to have a peg inserted at the outside that will touch the edge of the preceding slat. By keeping low lateral pressure the slats would settle against the pegs while rotating.I hope I have correctly interpreted the text. Machine translation would probably cause mumbo jumbo because there are too many special words here.

Mirror screw with adjustment.pdf

(383.74 KiB) Downloaded 542 times

[albertMunich]

Thanks Andrew for the ABS hints - i still have a roll somewhere that was mis-ordered, will give it a try.

[Panrock]

A lot going on here...

It's good to see the expertise being bandied about re. 3D printing. This is something I know nothing about.

Some amazing German patents there. I am still trying to get my head around the 'interlaced' stack method shown.

I used 'spreader' discs with raised rims clamped by central collars on my screw to get round the lateral warping problem.

Of course they didn't have today's option of electronic angle correction using a Timing Corrector. Using mechanical 'screw' correction on all 180 slats must have been laborious. Mind you, their mirror screw was much smaller and lighter than mine, so running it up and stopping it 180 times (or more) for adjustments maybe wasn't such a big deal.

As I have said, I am writing a summary of the project for the club magazine. This will involve at some point getting the whole caboodle working again. I'll then try to make a better, longer, video for YouTube.

When you rotate the screw by hand and view at the correct distance, all you see is a single light spot slowly describing a raster, left to right, line by line. On my avatar, you can see the effect - with more light spots - when viewing too close.

Steve O