Starting at the beginning with Mirror Screws

Forum for discussion of narrow-bandwidth mechanical television

Moderators: Dave Moll, Andrew Davie, Steve Anderson

Postby Panrock » Mon Sep 19, 2011 3:55 am

Wise words - I agree. I think the simplest thing (for me) would be to compress the stack between a couple of collars - top and bottom. I don't have ready access to the likes of a lathe or huge die for threading the shaft. For example, with my 'Grosvenor' I had the disc mounting bosses turned by someone else.

Steve O
Panrock
Green padded cells are quite homely.
 
Posts: 870
Joined: Mon Feb 05, 2007 8:25 am
Location: Sedgeberrow, England

Postby Viewmaster » Mon Sep 19, 2011 4:37 am

Panrock wrote:. I think the simplest thing (for me) would be to compress the stack between a couple of collars - top and bottom. I don't have ready access to the likes of a lathe or huge die for threading the shaft. For example, with my 'Grosvenor' I had the disc mounting bosses turned by someone else.
Steve O


Assuming that each collar is securely fixed to the shaft, top and bottom, then the top one could have 2 vertically tapped holes to take 2 clamping biggish grub screws into the last top slat (via a washer of course). That should ensure good clamping.

Hope to see it soon, running on Utube !!

I see over 900 reads to this thread..........
.......you may have some competition as dozens start making slats by the thousands :-)
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Viewmaster » Mon Sep 19, 2011 4:53 am

Just an extra....the collars could preferably be flanged in order to bring extra locking torque. ( giving greater distance from shaft gives more effective lock on grub screws and collars)
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Panrock » Tue Sep 20, 2011 9:05 pm

I'm seeing the laser cutters this afternoon to arrange the first test pieces and get a price for 120-off.

Soon we will know how these curved beasties actually perform !

Steve O
Panrock
Green padded cells are quite homely.
 
Posts: 870
Joined: Mon Feb 05, 2007 8:25 am
Location: Sedgeberrow, England

Postby gary » Tue Sep 20, 2011 9:06 pm

Exciting times!
gary
 

Postby Panrock » Wed Sep 21, 2011 2:55 am

Back from the laser cutters. I'll have the first test slats to play with next week.

The quote for 120-off slats is fortunately going to be affordable, so - subject to it proving possible to polish their edges - it looks like this project will go ahead.

For the video, I am hoping to use again Steve A's excellent PWM circuitry firing up a stack of luxeons in R,G and B, though for this application the pulses will have to be running about 10 times as fast! I'll probably have to use something a bit snappier on the outputs than the previous 2N3055's too!

Steve O

PS. Come to think of it, Steve A previously designed for 80 lines; it was loafing along at 30-lines; so it may not be such a step change after all... however this new 120-line system has a 4:3 aspect ratio and will be rotating at 25Hz. I reckon the vision bandwidth required will be 240KHz.
Panrock
Green padded cells are quite homely.
 
Posts: 870
Joined: Mon Feb 05, 2007 8:25 am
Location: Sedgeberrow, England

Postby Steve Anderson » Wed Sep 21, 2011 6:48 pm

Hi Steve, long time no contact, I guess we've both been busy one way or another...

I agree with your 240kHz bandwidth requirement calculation for this, thus if using a similar PWM system you'll be looking at switching in the low MHz region. This may become 'Not so easy', but let me dwell on it for a day or two...

Those 2N3055s/MJE3055s will definitely have to go!!

In thinking about this, what's going to be the source of video? Does (for example) the 'Aurora' standards converter provide this? And do you have one? Or do you have something else in mind?

This definitely does interest me (and Jeremy Jago via previous e-mails) though I will admit I'd take the easy cop-out and use a CRT...of course monochrome in my case, but if there's a requirement for a signal source at these standards, then I'm interested...

to wit, I'm in the process of trying to pull together an item for the newsletter, not the one usually due this month, but the December issue. If Jeremy decides to publish it, here's a sneak preview...

it has 16 Club Standard test patterns encoded within it, in fact 32, 16 being linear and 16 being Gamma-encoded. Once I'm finished with the last of the software it's done...a few more hours of work....there's more to it, but my reason for mentioning it here is if there's no source for your 120/25p video I'm willing to uprate it to suit....colour might be interesting (read a challenge) though...

Steve A.

P.S. NBTG = Narrow Bandwidth Test Generator.
Attachments
Photo 1.jpg
Photo 1.jpg (180.71 KiB) Viewed 15803 times
Photo 2.jpg
Photo 2.jpg (269.21 KiB) Viewed 15803 times
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Postby Panrock » Wed Sep 21, 2011 10:43 pm

Hi Steve,

Steve Anderson wrote:Hi Steve, long time no contact, I guess we've both been busy one way or another...


Yes, it's nice for me to once again have a little time available for NBTV... though this project won't be quite so "NB" as usual, this time...

Steve Anderson wrote:I agree with your 240kHz bandwidth requirement calculation for this, thus if using a similar PWM system you'll be looking at switching in the low MHz region. This may become 'Not so easy', but let me dwell on it for a day or two...


Thanks! I'll be building this from the ground up and not re-using any parts from the old system, which is now gathering dust and taking up a fair bit of space in the store room.

Steve Anderson wrote:Those 2N3055s/MJE3055s will definitely have to go!!


Thought that would be the verdict. I recall you weren't too keen on them before either, but they were 'legacy' parts on my rig and you allowed me to retain them.

As for the new 'line of light', I wonder what driver transistors you will recommend? I'll need to go all out here for maximum power and a fair number of luxeons, since (as I think Karen O mentioned) at 120 lines picture brilliance (or rather the lack of it) could become a problem. I'm hoping though that things will be helped here by the sheer size of this mirror screw.

I'm thinking of building this line of light around a vertical perspex rod with one side flattened and frosted. The lines of luxeons would be housed inside a vertical 'wedge' with reflecting internal surfaces, and they would be beamed onto the inward facing flat frosted surface of the rod for colour mixing.

Steve Anderson wrote:In thinking about this, what's going to be the source of video? Does (for example) the 'Aurora' standards converter provide this? And do you have one? Or do you have something else in mind?


I have been thinking it would be worth my while investing in one of Darryl's 'World Converters'. This could certainly provide the RGB 120-line video (amongst many other standards). However, I have also received a couple of kind offers. Dominic Beesley has offered to provide software that outputs from my sound card (not sure if this could be colour though) and Jeffrey Borinsky has offered to lend me his own World Converter.

Steve Anderson wrote:This definitely does interest me (and Jeremy Jago via previous e-mails) though I will admit I'd take the easy cop-out and use a CRT...of course monochrome in my case, but if there's a requirement for a signal source at these standards, then I'm interested...


Speaking personally, I got the CRT urges out of my system by building the 405-line 'Argus'! :) But It seems to me that if one wants a mechanical display that can also produce a visually large, relatively high definition, colour picture - the curved-face mirror screw is the only realistic option. Once it's built and working, I'd like to see if it holds any potential for 3D too... though the answer is 'probably not'.

Steve Anderson wrote:to wit, I'm in the process of trying to pull together an item for the newsletter, not the one usually due this month, but the December issue. If Jeremy decides to publish it, here's a sneak preview...

it has 16 Club Standard test patterns encoded within it, in fact 32, 16 being linear and 16 being Gamma-encoded. Once I'm finished with the last of the software it's done...a few more hours of work....there's more to it, but my reason for mentioning it here is if there's no source for your 120/25p video I'm willing to uprate it to suit....colour might be interesting (read a challenge) though...


Wow! This would be great if you could crack it. I can envisage needing a sharp-edged waveform when I'm finally 'tapping' the slats to (try to) remove any remaining jaggedness from the picture. I'll probably just feed in a fast square wave...

Steve O
Panrock
Green padded cells are quite homely.
 
Posts: 870
Joined: Mon Feb 05, 2007 8:25 am
Location: Sedgeberrow, England

Postby Steve Anderson » Thu Sep 22, 2011 4:50 pm

OK Steve, I have a few ideas rolling around in that vacuous space between my ears...but first I think we need to work backwards...

Let's start with the light emitters, presumably Luxeons? Have you made a choice as yet? If you have then please pass on their Farnell/RS (whoevers) stock No. so I can download the datasheet(s)...

Furthermore, have you made any choice as to how many of each colour (I presume RGB) and in what arrangement you might connect them, series, parallel or a combination of both?

I'm undecided at this point as to whether to use MOSFETs as the drivers or standard bipolars, both have advantages as well as drawbacks.

Conceptually I think I have 'cracked it' as far as a test-pattern generator would go, but at some point you're gonna want to interface this with a real RGB analogue source. Therein lies my dilemma, the LM311-type comparators are way too slow for this bandwidth, but let me dwell on this further...

The mini-NBTG above could be adapted to produce RGB in the standard you require, but as it's only a 14-pin package (deliberately chosen) the range of colours would be limited. Remember Windoze 3.1 in 16 colours? Eeek...

Now I guess we need to set some 'standards', for 120-lines, the only reference I have is the RCA Victor data from the 30's by Engstrom, Beers (wonderful name), Zworykin and Kell...yes that one.

This was based on a mechanically-scanned film system at the studio end at 24 fps...rather perverse for those of use who are used to some multiple of 10ms for a frame. But the concept is sound, so here's my thoughts...

Let's up the frame (non-interlaced) rate to 25 fps, at 120-lines that yields a line rate of 3kHz, or 333.33us (recurring) per line. within that line we'll allow some 15% for front porch, sync and back-porch with a view to adding some form of burst in the back-porch for composite colour in the future as well as black-level clamping.

So as a nice round figure line blanking comes out at 50us...active video period then is 283.33us. This then divided by 160 (4:3 aspect ratio) is 1.77us/pixel, or a bandwidth of around 280kHz....please, someone, check my math.

In keeping with established proportionality a line sync duration would be something like 20us...seems reasonable to me...

That's the line-time specification dealt with, now the frame...keeping it simple a single broad frame sync pulse of approximately half a line in line 1 (or line 0 for the binary-heads), say 150us?

These syncs could be totally separate (RGBS) or encoded on the Green channel (ala GVG in the US), or even on all three signals...overkill IMHO.

Thoughts, suggestions, even flames sought!

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Sart at the begining Mirror Screw

Postby Alan » Thu Sep 22, 2011 8:21 pm

I noticed from the last posting by Steve that some questions had previously been raised about the increasing luminance required for ever larger, and higher definition displays.
Reminiscing back to the dim and distant '50's, I remember that the Phillips Back Projection TV with about a 2' screen, used a (approx) 2" CRT with 25kV EHT - the then normal was about 8kV - which gave a blindingly bright picture on the CRT screen - needed to compensate the losses of the light path via mirrors and the viewing screen.
I seem to remember that Klaas has visited this area quite extensively.
The conclusion seems that the greater the viewing area, and faster the scan, the higher the intensity of illumination required.
Now we come to drivers. From experiments done by Steve, it would seem that PWM is favorite, as this requires no dissipation in the drivers - a great saving skin and protective gloves! It also allows for the use of high intensity LED's. For Screws, it would seem that the favorite is still a line, driven in parallel, and parallel to the axis of the screw, fitted with a diffuser.

Alan
User avatar
Alan
Laboratory Assistant
 
Posts: 34
Joined: Sat Jul 25, 2009 9:56 pm
Location: Glastonbury, UK

Postby Panrock » Thu Sep 22, 2011 9:26 pm

Hi Steve,

Steve Anderson wrote:OK Steve, I have a few ideas rolling around in that vacuous space between my ears...but first I think we need to work backwards...


And first, I'd also like to thank you once again, for freely offering your time and in getting involved. This is not at all taken for granted, but it is very much appreciated. Of couse it's still early days. I haven't even confirmed that it will be possible to polish the laser-cut slats to be reflective enough yet, or whether the viewing distance for correct aspect ratio will work out as I want it. Those things should be known in a week or two. But anyway...

Steve Anderson wrote:Let's start with the light emitters, presumably Luxeons? Have you made a choice as yet? If you have then please pass on their Farnell/RS (whoevers) stock No. so I can download the datasheet(s)...


I've had a look this morning at both their web sites. The most desirable luxeons with lenses seem to be 'discontinued' but in any case it might be easier and as effective here, where colour mixing over a narrow line is required, to use strings of narrow angle superbright LEDs such as RS 713-3961, 496-6178, and 713-3964 - facing into the frosted 'mixing line' on my acrylic rod. I reckon there should be room for 50 of each in the housing for each colour channel. Others may wish to advise on the appropriate proportions of these to achieve white. The LEDs do differ somewhat both in luminous intensity and beam angle though, which will complicate the calculation. Yes, 150 of these LEDs should be pretty bright, but I don't think we should underestimate what will be required here.

Steve Anderson wrote:Furthermore, have you made any choice as to how many of each colour (I presume RGB) and in what arrangement you might connect them, series, parallel or a combination of both?


No. That's still up for grabs! Any required power supply voltage and current can be provided to order.

Steve Anderson wrote:The mini-NBTG above could be adapted to produce RGB in the standard you require, but as it's only a 14-pin package (deliberately chosen) the range of colours would be limited. Remember Windoze 3.1 in 16 colours? Eeek...


My advice would not to worry too much about modding your test generator at this stage. Would you be bringing yours to the convention or would I be building one? Once the mechanical side of the mirror-screw rig has been constructed, my priority will be to get 'pictures' onto it as soon as possible. What follows is that I'll be using Darryl's converter or another similar solution. Test waveforms could then simply be fed into it at 625 from my own bench generator.

Steve Anderson wrote:Now I guess we need to set some 'standards', .....


Yes, and at this point I'm going to try to bring Darryl into this, so he can advise. He doesn't seem to be a member here though. The standard I have in mind appears in his World Converter handbook as "120/25p 4:3 progressive mechanical (TeKaDe/Fernseh-A.G., 1932)".

Steve O
Panrock
Green padded cells are quite homely.
 
Posts: 870
Joined: Mon Feb 05, 2007 8:25 am
Location: Sedgeberrow, England

Postby Viewmaster » Thu Sep 22, 2011 11:50 pm

Panrock wrote:Hi Steve,

. I haven't even confirmed that it will be possible to polish the laser-cut slats to be reflective enough yet,
Steve O


The best edge quality is obtained when laser cutting and using Nitrogen assisted gas. (see this link)
http://www.ackermanengineering.co.uk/laser_cutting.htm

This adds to cost though. Maybe discuss this with your
cutting firm, although you may have already done this ?
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Panrock » Fri Sep 23, 2011 12:30 am

No, I didn't know about this - always learning... :) I've just phone my laser cutters. They use nitrogen as standard.

Steve O
Panrock
Green padded cells are quite homely.
 
Posts: 870
Joined: Mon Feb 05, 2007 8:25 am
Location: Sedgeberrow, England

Postby kareno » Fri Sep 23, 2011 9:09 am

Hi Steve,

For most of the summer I worked on my PIC-based 405 line Compact Flash player and it is now working really well. I have a Linux script to prepare data for the card (for which I had invaluable assistance from the UKVRR forum).

On my TODO list is a SD/MMC version for high end NBTV and 120 lines is one of the standards I am aiming at. To give you an idea, a 2GB SD/MMC card could store an hour of 120 line video!

Still in the research stage at the moment but I think I can do it.

One trick I employed to avoid buffering - my CF player issues instructions to the card during the frame sync period. The frame blanking period (around a millisecond) then gives the card time to set up for the actual data transfer, which is raw video.

That time space between issueing the read instruction and starting the video data transfer is key to avoiding buffering and I will doubtless need to do the same for 120 lines. In practise this will mean one line of the display will be black and given over to 'transfer preparation time' for the card.

Have you thought about what the video waveform will look like? Will there be a frame sync pulse? If so I would assume it will be a wide pulse taking up a large part of a line? If this is the plan then I will use this line as the transfer preparation time.
Attachments
CF player 3.jpg
My 405 line CF player
CF player 3.jpg (75.67 KiB) Viewed 15655 times
kareno
 

Postby tubesrule » Fri Sep 23, 2011 11:17 am

Steve notified me of this thread and asked me to add what I could, so here goes...

I have not built a mirror screw myself although this has always been on my todo list. After viewing the one at the ETF and some of Peter Yanzcer's, I would also like to experiment with them.

That said, I was also zeroing in on a 120 line set but quickly realized the difficulty in precisely aligning the mirrors. Reading through this thread, I would only add two other considerations. First, accumulated error. I like your idea to use alignment pins as this will greatly ease construction, but if there is any cumulative error in their locations, this could quickly ruin the screw.

Which ties into the second consideration. For a vertical line in the image to appear straight, there can be no error in the mirror positioning. For practical purposes, lets say we'll accept a +/-5% positioning error, line to line, which would cause the vertical image to appear wiggly. 5% represents just 9 minutes of error. If this is non-cumulative from line to line, we should be fine. If however there is any cumulative error, and we don't want more than 5% error from the first line to the last line, that represents just 4.5 seconds of error in each mirror. The lower this number, the straighter things will look.
The 60 line screw at the ETF was an original one from ATT from the 20's or 30's and even it has noticeable error from top to bottom.

My thought for making the screw, although no where near as large as what you are working on, was to have the entire screw machined out of a single piece of aluminum which could then be polished and flash chromed or aluminum plated like a mirror. This would result in a very light assembly with mirror positioning as accurate as the CNC that cuts it. Of course I have not followed through on this as of yet. Just thinking out loud.

With the WC-01 converter I have thought about allowing the user to shift the position of each line individually to correct for mirror position errors, but this would certainly be a tedious process, and would have to be done and stored for each screw. You would also end up with shadowing at the beginning or end of each line depending on which way you shifted it.

The weight was another big concern since it will take a sizable motor and control electronics to spin and sync this assembly. One thing you can do to cheat this with the WC-01 is to rather than sync the motor to the video, sync the video to the motor. By generating a signal from the screw, preferably frame or line, you can use it to sync the WC-01 output, as long as the speed is within about +/-5% of true. This way the motor speed and stability is not critical. Even if you do run a servo loop on the motor, the phasing of the video output on the WC-01 will never change regardless of the input, so once you have it framed correctly, the motor will never have to resync easing the control loop design.

For a video signal, I don't think an audio output will be sufficient as 20kHz would only represent about 12 pixels on each line at 25fps. Minimally you need about 270kHz of bandwidth. The WC-01 provides 540kHz of bandwidth on the 120/25p standard since an analog system with a given bandwidth will appear sharper than a digital system that provides just the Nyquist sample rate.

Because of this high video bandwidth, a PWM controller for the LED's would be very challenging. What I decided to do in the WC-01 is to use a MOSFET current controlled output driven from a PWM switching voltage supply. On first powerup, the unit applies full power to the LED's for a few msec's and reads the voltage drop across them. It then uses this information to keep the dissipation across the MOSFET to a minimum. It continually reads back how it's doing and adjusts the switching supply as needed to account for the temperature coefficient of the LED's and any other changing drops in the system. There are three independent 1/3A outputs that can be wired together to provide a single 1A monochrome output, or three 1/3A RGB outputs at 20V maximum. This allows for up to 20W of power to the LED's. There is also a 20W class D amplifier that can be used to drive a phonic coil, or a speaker if you just want to pass the audio through.

For the video standard itself, the 120/24p from RCA in 1931 isn't directly applicable as it was intended for a crt set. The signal looks relatively familiar to today's composite signal with line and frame blanking areas and line and frame sync pulses. Mechanical video signals can not have vertical blanking areas since there is no break from the end of the last line to the start of the first line, regardless of disk, drum or screw. This could be done if "blank" holes or mirrors were added or if you sacrifice some of the mirrors or holes, but since image resolution was at such a premium in these early days, nothing was wasted on blanking or syncs. There can be a short line blanking area at the beginning of each line as this was used in some original systems to help lock to the fundamental line frequency. Baird originally used this in his broadcasts.
No mechanical standard that I have found used any kind of frame sync, again as this would show up in the image unless there were some "blank" mirrors or holes added. Basically, the video for any mechanical standard has to be continuous with only a possibly short line blanking area.

Darryl
User avatar
tubesrule
"Fester, enough of the light-bulbs!"
 
Posts: 16
Joined: Fri Sep 23, 2011 3:38 am
Location: Michigan, USA

PreviousNext

Return to Mechanical NBTV

Who is online

Users browsing this forum: No registered users and 32 guests

cron