Arduino Televisor by Keith Colson

[Robonz]

Hi Guys, I started my build on an "Andrew Davie's Arduino Televisor" today. I am at step 1 of 10,000 or whatever. I thought I would start by drawing up the Nipkow disc as my first step. I download Gary's Nipkow Disc DXF Generator V1.1 and put in these settings

The aspect is 1.3 - I thought I would make the screen a tad wider as it's already really small. Yes, distortion added.



I put it into my cad program Coreldraw and drew some circles over the cross hairs of 3 apertures and put 3 apertures of 0.659mm stacked to see how they would line up, but they don't. Is this an error in the software or do I misunderstand the disc?

Here is a screen shot. The red line(circle) is the aperture to the left and the green line is the one to the right. You can see my yellow rectangles do not line up. Any ideas? I wonder if its a metric issue as it was inches to start off with.

[gary]

Sorry Keith, I can't for the life of me follow what you are doing there, can you elaborate?

[Robonz]

Sorry Gary, hopefully I can explain it better with this picture. I am a bit dumb at explaining things sometimes.

I have drawn red and green centric circles that map perfectly on to the left and right crosshairs generated by your program. We are looking at the top 3 apertures on the Nipkow disc. I have drawn yellow rectangle to represent the actual aperture to show line overlap etc. When zoomed in (picture in first post) I would have assumed the apertures would be perfectly spaced edge to edge. I hope that makes sense.

e.g. I would expect the yellow rectangles to line up perfectly in the vertical plane

[gary]

Ok, I think I understand, but just confirm that you are talking about the overlap in the vertical direction and that the rectangle positions in the horizontal direction are just where you happened to draw them for clarity?

I would assume that the error in the vertical direction would be due to round off error but would be well within the tolerance of my CNC machine (for which the software was written).

However it's worth my looking into.

[Robonz]

Thanks Gary

You got it! The error is only .03mm or so but that is about 5% of a 0.6mm line width, if I calculate it right. My CNC is good for .01 if I hold my breath during the cut. Am I being too fussy? Before I commit the energy to make something I like to measure twice as they say. If this is the expected error then I will need to work the calculations out in Excel which I don't know how to do right now and am avoiding it haha. Its so much easier to click your button.

Cheers
Keith

[Andrew Davie]

Thanks Gary

You got it! The error is only .03mm or so but that is about 5% of a 0.6mm line width, if I calculate it right. My CNC is good for .01 if I hold my breath during the cut. Am I being too fussy? Before I commit the energy to make something I like to measure twice as they say. If this is the expected error then I will need to work the calculations out in Excel which I don't know how to do right now and am avoiding it haha. Its so much easier to click your button.

Cheers
Keith

I should do you one with OpenSCAD and see how it compares.
Will do in a few hours!

[gary]

Thanks Gary

You got it! The error is only .03mm or so but that is about 5% of a 0.6mm line width, if I calculate it right. My CNC is good for .01 if I hold my breath during the cut. Am I being too fussy? Before I commit the energy to make something I like to measure twice as they say. If this is the expected error then I will need to work the calculations out in Excel which I don't know how to do right now and am avoiding it haha. Its so much easier to click your button.

Cheers
Keith

Well, I am using double precision for everything so it is hard to see how it could be less precise than, say, excel.

I am not, at present, in a position to reproduce your "test" in order to confirm an error, however, if you are using a CNC your should actually NOT be using the crosshairs to locate the aperture, but the points on the "bead" layer. I am sorry that is so confusing but I think you will understand knowing that I wrote this programme to create my bead disks (shown elsewhere on this forum). The crosshairs were added later to allow people to create disks manually by printing the disk out on paper.

To test this I have created a file using DXFNipkow with a circle through every aperture, and a circle at every aperture with a diameter twice that of the actual diameter, this means that the outer limits of the aperture should just touch the big circles either side - as far as I can tell this is spot on (given the cad software I have).

The points on the "bead" layer are, of course, the centre point of the aperture/bead hole. Now the crosshairs *should* bisect those points exactly, but it may be worth your while to do your "test" with the points rather than the crosshairs (if you haven't already twigged to that), and certainly, when converting to gcode (if that is what your CNC supports) those points would be the correct objects to use.

I attach the test file and would appreciate it if you could confirm my results.

NipkowDXF1.dxf

(12.38 KiB) Downloaded 291 times

[Robonz]

Thanks Gary I totally believe you have it right and thanks. I cant believe you responded so fast. Something I cannot fathom is happening. If I load the new DXF over your old one it is a perfect match. The circles all measure symmetrical. When I select one of those bead circles and rotate it from the origin 11.25 degrees a small alignment error shows but I cannot measure why it would happen. If I rotate 90 degrees or 180 it lands directly in the right spot. I was using those cross hairs to center my apertures that I have designed. I hope you haven't burned time on this, I already owe you a beer and I'm only at step 1.

Anyway I have made a disc from scratch and it rotates in the cad okay, so my laser cutter is going to be happy. I kind of enjoyed creating the disc too,

ANDREW, if you don't mind checking what I have made to go with your system. It has one encoder slot 180 degrees from the start aperture. The attached pdf has specs.

Construction of the disc will be 4.5mm thick clear acrylic with black paint on the viewing side. The laser will remove the paint where the yellow rectangles are. This should equate to a super thin aperture say 50 microns of paint at a guess. I have made the apertures thin in the scan direction to hopefully gain some vertical resolution at the loss of brightness, I have a 30 watt led to make up for it haha.

[Robonz]

I have ordered this motor from RS part 834-7613
986 rpm, 12.8 W. 6mm shaft It has max efficiency around 700 rpm and I hope to direct drive off the motor shaft with little wobble. I might be dreaming though. I better turn the hub up well I guess. I am not too practiced on the lathe, fingers crossed



I have ordered this light source on Ebay. It is 30 watts. I do not need to run it at full power of course. I have a really nice collection of diffuser material I can laser cut and put on it.

30W 4X CREE XM-L2 LED 12V

[Andrew Davie]

ANDREW, if you don't mind checking what I have made to go with your system. It has one encoder slot 180 degrees from the start aperture. The attached pdf has specs.

Construction of the disc will be 4.5mm thick clear acrylic with black paint on the viewing side. The laser will remove the paint where the yellow rectangles are. This should equate to a super thin aperture say 50 microns of paint at a guess. I have made the apertures thin in the scan direction to hopefully gain some vertical resolution at the loss of brightness, I have a 30 watt led to make up for it haha.

Should be OK. Moving the IR hole to the inside and putting your scanlines towards the rim will give you an ever-so-slightly larger image. Personally I'd err on the side of more light and less resolution - after all, vertical resolution is limited by the 19200Hz bandwidth. That works out to 19200/32(lines)/12.5(fps) pixels/line = 48. So you're not going to do a lot by having tiny apertures.

[Robonz]

Thanks Andrew, great feedback. I can cut another disk pretty easily if I need to increase the apertures. I will be lucky if I get it right first try anyway. I purposely put the image 20mm away from the edge of the disk so the light wont end up going around the edge. Encoders are also way easier to mount on the edge.... well unless it wobbles a lot.

[Andrew Davie]

I have ordered this light source on Ebay. It is 30 watts. I do not need to run it at full power of course. I have a really nice collection of diffuser material I can laser cut and put on it.

30W 4X CREE XM-L2 LED 12V

I am including a pre-built 30 LED matrix board for you in my package to you, so you can try that too.

[Andrew Davie]

Construction of the disc will be 4.5mm thick clear acrylic with black paint on the viewing side.

If you silver the back of the disc, in theory the 99% where there isn't a hole will reflect the light back into the LED reflector and it will bounce back and forth until there IS a hole to escape through and you could increase your brightness. In theory, anyway - always wanted to try this

[Robonz]

Andrew, thanks for the extra leds. If I wanted to get fancy, a lens disk would be awesome, but they look really hard to make. I think I would go for a mirror drum if I wanted to move to the next level. I have acrylic mirror that can be laser cut. Its only 1mm thick so it would have little double imaging. I only just worked out how a mirror drum works, its brilliant and you get a nice big picture. We could use your pcb to drive that too. But first steps first. I am easily distracted.

[Andrew Davie]

Hi Guys, I started my build on an "Andrew Davie's Arduino Televisor"

Let's go with the name "ArduinoVisor" from now on.