A new 90-line Mirror Screw

[acl]

Please find attached the assembly code if anybody has the as much technical aptitude as Karen and feels like modifying it for 90 line operation.

Regards Chris Lewis

tc2.asm

Karen's assembly code

(25.85 KiB) Downloaded 18 times

[Viewmaster]

How Karen's system works is beyond my understanding with all that coding!
But here is a thought to solve the 90 lines into Karen's 60/120 line magic box. This is probably a non starter but might suggest to others some other ideas.

The only input into Karen's unit from the mirror screw is the frame pulses as shown on her layout in the manual.
So could 30 extra frame pulses be generated either electronically or with 30
small narrow extra slats to 'trick' Karens system into thinking it was working
on 120 lines (90 + 30)
These 'extras' would probably not need to be so polished as others as they are only reflecting back a frame pulse and not a full line for visual picture watching.

This idea might be rubbish but no harm done

[Viewmaster]

Further thoughts.......rather negative........
Because a complete picture frame occupies one complete rev of mirror screw
those extra 30 dummy frames (mech or electronic) would have to be included
in the 1 rev.
This, unfortunately, means that the 90 frames of picture would only occupy
90/120 of a complete rev., leading to picture black outs lasting 30/120 of each frame.
So I think my previous idea is rubbish unless someone can see a work around.

[Panrock]

Thank you Chris for your contributions! And thanks go to Albert for 'thinking out of the box'. Yes, I agree with the comments on your second post, Albert.

But there now seems to be some hope... just a little, mind.

First, matters improved when I troubled to ensure the phasing of the frame pulses 'off the screw' matched the beginning and end of the physical slat frame scan (rather than relying on electronic adjustments of the phase). It's all coming back to me now... I went through a similar discovery process with the 120-line screw.

I found an ancient Richard Russell miniature Test Card Generator in a drawer. So now I can set things up on Test Card 'C'!

Phew! Karen's unit now does seem to work on 90-lines. But it's a hell of a slog pushing the buttons thousands of times (yes, really) to effect meaningful timing changes over the whole picture. I would like to put my Parkinson's tremor to work in automating the process!

The above is not helped by the grotesque slat angle errors still in the screw. Yes, the screw (as an object) is good to 'look at' but the accuracy demands are intense. I think I'll need to run the setting tool past each slat again.

So, we plough on. One step back, two steps forward.

Steve O

[Panrock]

I have dismantled the screw and am now in the midst of re-assembling it for the third time! Hopefully "building back better".

Steve O

[Panrock]

Long time no speak. The mirror screw was put together for a third time, but despite taking every care, the precision of the slat angles just could not be made good enough! I have therefore got started on re-working the CAD file to make new slats that include indexing holes and pegs, so that they can only lock together in one critical position.

My first 120-line screw originally used such indexing. However, strangely, I was later able to do without it!

In retrospect, I should have used indexing here from the beginning. It will be considerable extra expense, but I am determined to see this project succeed. It seems unlikely the mirror screw will now be ready to show at this year's convention.

The precision sought here will (ideally) be ±1 minute of arc. It's become clear it won't be possible to impose this sort of figure on every two adjacent slats by simply pressing them to a setting tool or a precision protractor. We'll have to hope the laser cutting precision and lack of backlash on the pegs is good enough. If it isn't, the remaining errors should be soaked up by Karen's unit (not the case at present). Incidentally, it's been now verified that Karen's unit does work happily on 90-lines.

On another topic (Chris) could you kindly tell me what are the NBTVA subs, and how to pay them? Thanks.

Steve O

[acl]

NBTVA Membership 2025.pdf

(47.21 KiB) Downloaded 16 times

[Panrock]

That was quick!

PAID.

Many thanks,

Steve O

[Panrock]

Here's a screen-grab of the revised CAD file for the new slats. These will be 2mm wider than the old ones and will incorporate indexing holes every 4 degrees to accept locating 'pegs' (actually more like 'pins'). These pegs will actually be short lengths of solder 1.22 mm thick, that will push cold through the nominally 1.24 mm holes in staggered, adjacent slats. Using solder has the advantage it can be given a last minute squeeze across its ends to take up any slop after insertion, and it is easy to nip and neatly rub down the ends.

On Momday I'll go back to the original laser/water cutting firm and get another quote.

If this doesn't finally fix the situation for once and for all "then I'm a Dutchman"! Klaas, no offence meant - you are greatly esteemed. This is an old English saying!

Steve O

[Klaas Robers]

No problem Steve!

If you want to come over in order to be naturalized, you are welcome.

[Panrock]

If you want to come over in order to be naturalized, you are welcome.

That would be an honour. However, I don't expect to 'become a Dutchman' quite yet. I am hopeful this indexing solution will work, or at the very least it'll be interesting to see how good it is. It should be possible to get the angles within easy correction by Karen's unit.

The request for the quote went off to the fabricators today.

Can anyone think of a reason why the angle should not be precisely 4 degrees (360/90) ? Say, instead: 360/89 or 360/91? For some reason this question has been haunting me.

Steve O

[Klaas Robers]

The problem is: how precise is precisely?

Suppose you want to have 90 pixels on one line. Then an error of one pixel from one line to the next line comes with an angular error of 0,04 degree. That is already very precise.

And then: the errors accumulate over the stack. So in fact you want to have a precision of 0.00004 degrees between two adjacent slats. I don't know how to get that precision.

So you need a form of alignment from slat to slat, and from each slat in respect to slat number one. I don't know how to do that by hand. Yes of course, the Karen Orton correction helps, but even then.....

Difficult Steve.

[Panrock]

Hi Klaas, thank you for these comments.

Previously, I had been checking and correcting for accumulative errors by looking critically at the stack from above, every quarter of the way through the slat-laying task; namely after 22½, 45, and 67½ slats had been laid. At these points, looking down on the stack, the overall shape of the stack should appear bounded by a right-angle. This proved to be quite a sensitive method.

In future, when I use the CAD indexing, there will still be errors but in the nature of things I don't think they will be cumulative. The errors should be centred around a correct mean position and will be caused by manufacturing tolerances in the laser cutting.

Even so, the 'right-angle test' will still be possible. In the event of a cumulative error developing, the whole stack (this time locked together with pegs/pins) could simply be given a squeeze or stretch, like a concertina!

Steve O

[Panrock]

Unfortunately the laser cutters are taking their time on this second occasion. I suspect it's a 'small job' they're not really that keen on. I'll give it to the end of next week before looking into taking my custom elsewhere. Can you recommend any firms?

While I wait, I've stripped out the original mirror screw yet again and today I made up some cabinet sides (with rectangular holes for the connections and controls) out of MDF. Eventually I'll make up a front cover with a 'screen' shaped orifice and then I plan to veneer the lot and spray lacquer the cabinet like a vintage set.

I'd like to end up with my very own "early thirties" TeKaDe near-replica.

There's now no way this will be ready for this year's Convention. It should be on show in '26 though. Thank you all for following this and for your patience.

Steve O

[Panrock]

The sides are fitted. The mirror screw itself is currently absent.

I won't make the front screen/aperture (probably a speaker grille too) and optimise its position until I have television pictures! Unlike the original TeKaDe, this will be 'colour'.

Steve O