Starting at the beginning with Mirror Screws

[gary]

They were introduced in 2000.

Actually it is only essential that the CPU supports the RDTSC instruction. However, obviously the older the computer the slower the clock. It may indeed be possible to use a slower model, but I haven't tried it on anything less than a P4 - there are so many of them floating around it should be possible to pick one up for less than 5 pounds.

Anyway, I can provide you with a routine that can test the machine for both the instruction and the speed - it will have to wait a couple of weeks though as I am in the process of moving and by tomorrow I will have anything packed away, but if you can wait that long we can at least do some tests that will only cost us a few minutes of time.

Edit: (oh, and of course I would be happy to write the software you would need to use it)

[Panrock]

Well, it's beginning to look like I'm going to be spoilt for choice as to the source. I just hope my display will do it justice...

Many thanks to all you very kind people!

Steve O

[tubesrule]

The timing I mentioned earlier in this thread for 120-lines was a suggested starting point. If I can find some data on the TeKaDe system I'll perhaps go with that rather than introduce yet another 'standard'.

Steve A.

P.S. Having a hard time of finding on-line TeKaDe data, anyone have any they can post here?

Hi Steve,
I have found excellent documentation for the RCA 120/24p standard and posted it here:
http://www.earlytelevision.org/rca_1932.html
http://www.earlytelevision.org/images/r ... l_xmtr.pdf

For the TaKaDe 120/25p, as with most of the early mechanical formats, I had to glean as much data as I could from period publications and from people like Peter Yanzcer. What I have for the 120/25p is it used the most basic format, continuous video. No frame or line blanking area was used.

One of things I think could well be a stumbling block when using a PWM system at low MHz rates may be the response time of the LEDs/Luxeons. I think the red and green ones may be OK, but I'm not sure about the blue variety.

I recall Klaas mentioning that there is a visible delay in the switching speed of supposedly white LEDs due to the UV-excited phosphor. Whether this applies to the blue-only LEDs I don't know...I guess I should bone-up on their operating mechanism and/or pour over some datasheets.

This is only true of most white LED's. For cost, most white LED's are actually much like a fluorescent lamp in that they use an ultraviolet emitter to excite a phosphor. Because of this, the phosphor adds a relatively large delay to the response, just like the phosphor on a crt. I have not seen this type of white LED work at much above about 1kHz. Blue LED's operate exactly the same as red and green, so no worries there.
There are white LED's that are actually separate RGB chips underneath, and these also work fine. Basically individual R, G, B LED's work at higher frequencies then we are concerned with. I use some OSRAM RGB packages on my full color lamps and they perform well up to the 0.5MHz I run them at.

Another thought for video generation is Kat Manton's MythTV PC based converter which I believe can produce 120 line images, or if not I'm sure can be made to produce them. There are a few threads about this over on the UKVR forum like this one:
http://www.vintage-radio.net/forum/show ... ght=mythtv

Darryl

[Steve Anderson]

I've mulled over this for quite a while today, it was pleasant outside and no rain for once.

I have to agree with Darryl, I think going the PWM route for this is going to be quite a challenge. I foresee PCBs with at least two layers, one acting as a ground-plane. The devices available needed for these speeds are finicky (read often unstable)...especially on home constructed Veroboard...

If I were attacking this I would relent and use more conventional linear driver(s). The price to pay will be a lo-tech chunk of metal (cleverly disguised as a heatink) and a beefier power supply. But in general a 4A PSU isn't that much more costly than a 2A one, and the size and assembly time not that much greater. A pre-built switching version as good as eliminates the argument of weight too...bearing in mind your back-breaking Grosvenor...

So that's my suggestion at this stage...

I also presume that the WC-01 doesn't do any Gamma-processing, I wouldn't expect it to unless specifically mentioned. This we'll have to address, Klaas's "One-device Drivers" are looking very attractive right now...

...a few more thoughts thrown in the air...

Steve A.

[tubesrule]

I also presume that the WC-01 doesn't do any Gamma-processing, I wouldn't expect it to unless specifically mentioned. This we'll have to address, Klaas's "One-device Drivers" are looking very attractive right now...

...a few more thoughts thrown in the air...

Steve A.

Steve,
With the WC-01 You can select between gamma corrected or linear output to match your driver. You can also select polarity and line/frame scan directions. If you are running a full RGB lamp, you can also select between orange, white and full color output modes.

Darryl

[Steve Anderson]

Darryl,

As a typical engineer I went straight for the specifications page of the WC-01 web-site to the exclusion of all others, unless I missed it I didn't see any reference to the features you mention. But it does, so that's excellent!

Going beyond the use of a WC-01, what should be considered is if Steve O (or anyone else) decides to build a compatible camera...yea, I know good luck on that one, but it should be thought about. Here I'm really only thinking of the Gamma issue.

Steve O used PMTs in his 30-line colour version with no Gamma processing, likewise in the monitor. Now those perverse enough to have a shot at a mechanical 120-line camera would need to consider Gamma, especially if up-converting to (say) 625/525. I know that is unlikely, but worth thinking about at what, from an electronics viewpoint, is still quite conceptual.

Me, myself and I wouldn't bother about it, but maybe others might.

What does interest me is recording this 120-line signal, natively, whether colour or monochrome. Here I invite Karen O to enter the fray and could her 405-line recorder be adapted...I'm not asking she does, just conceptually could it be done using similar technology?

So yet another can of worms...

Steve A.

[Steve Anderson]

Steve,

Done some simulations for the LED drivers as conventional linear types. Using standard components like LF356 op-amps and TIP31 driver transistors the bandwidth should exceed 1MHz. A BD139 would extend that to over 2MHz, but that's not needed here.

Power supply required +/-12V at a few milli-amps, say 100, and +24 to +30V unregulated to power the LED chains at 1A for all three. Each colour is comprised of 8 strings of six LEDs, each LED having a nominal peak current of 30mA.

This makes a total of 48 LEDs for each color rather than 50, does that matter?

Where I'm stuck is on the biasing for this, so a few questions for Darryl...

Darryl, are the mechanical RGB outputs on the WC-01 DC-coupled or AC coupled? If DC what is the nominal black-level voltage?

What (if any) is the reference (sync) signal output in this mode?

I'm trying to ensure this will work with the WC-01 as well as other sources.

Steve A.

[Panrock]

Just a word of appreciation for all you are doing Steve A. Please keep it up!

Steve O

[tubesrule]

Where I'm stuck is on the biasing for this, so a few questions for Darryl...

Darryl, are the mechanical RGB outputs on the WC-01 DC-coupled or AC coupled? If DC what is the nominal black-level voltage?

What (if any) is the reference (sync) signal output in this mode?

I'm trying to ensure this will work with the WC-01 as well as other sources.

Steve A.

Steve,
The mechanical outputs are 0 to +1V into a 75 ohm load. Black is normally set to 0V however you can adjust the output levels and gains, so you can push the black level higher than zero, but not less since there is no negative supply in the converter.

The sync output is a 0 to +3.3V signal that can be either a square wave or a sine wave. You can further select it to be either a line, frame or mains frequency reference.

Darryl

[Panrock]

The sync output is a 0 to +3.3V signal that can be either a square wave or a sine wave. You can further select it to be either a line, frame or mains frequency reference.

Darryl, are the line and frame pulses available separately or can it be arranged for the frame pulse to interrupt the line pulses?

Steve O

[kareno]

My 405 line CF player is purely a player, Steve. It cannot record. The data on the CF card is prepared on a Linux computer. The procedure is:

1. I get a movie clip from somewhere (I could mention ripping DVDs but that would be copyright infringement so I won't mention that

2. I run a Linux script that converts the movie clip into a binary file (Kat Manton is really the talent behind this step)

3. The binary file is copied to the raw CF card by accessing the card's physical device image in /dev. I do not use any file system on the card - the sectors contain raw video and audio samples.

4. The card is plugged into my player and the play button is pressed.

I'm certain my player could produce a 120 line signal, however, it would be monochrome as my player has only one video output.

Also, syncs would have to be embedded in the video like a more familiar 405 or 625 line signal. External hardware would be needed to decode separate syncs and video but it would not be complicated.

I'm sure that between now and the convention I will have something together. I just need to know the spec of the syncs and 'light up' signal for the LED array.

[tubesrule]

The sync output is a 0 to +3.3V signal that can be either a square wave or a sine wave. You can further select it to be either a line, frame or mains frequency reference.

Darryl, are the line and frame pulses available separately or can it be arranged for the frame pulse to interrupt the line pulses?

Steve O

I knew you were going to ask this

Right now the line pulse or frame pulse is strictly just that signal. I guess I need to add a line pulse with a pulse missing for the frame as is done with the NBTV standard.

Darryl

[Panrock]

The first slats are now cut and ready for collection. I'll go get them today. The suspense!

Steve O

[Viewmaster]

Just turning over my lateral thoughts as we all do whilst drinking coffee.

A test rig to try out any concave curvature of mirror screw slats without having to get samples cut each time............

On a curved slat only the first and last pixel on any line is relevant to the correct curve at any particilar illuminant/viewing distance, so my idea is to have two small pieces of mirror mounted onto a wooden rocking arm. See drawing.
In the case of a 120 line screw, the arm to be allowed to rock 3 degrees and with 60 line, 6 degrees, and so on, against stops or pegs.

The 2 small mirrors are mounted on the arm at each end which can be swivelled and locked at any position (Nuts and bolts etc). The distance between them is the slat length.

Now, say one wishes to try out a mirror screw with a slat curvature of 5 feet.
We call this 5 foot distance X.

Mount the arm firmly down horizontally so it cannot swivel and stand with your eye 5 feet from the arm.
Arrange to have a slitted LED on just beneath your eye.
(don't poke your eye out )
Make sure that you head cannot move...rest it against a stout chair/post/door etc.
Look at the reflection in one of the mirrors and get someone to swivel it so that the LED's slit is seen in the centre of it. Lock that mirror.
Now do the same with the other mirror, again getting someone to adjust this second mirror's swivel so you can again see, the LED's slit reflected in the second mirror. Again, lock it.

These 2 small mirrors are now a copy of the first and last pixels of a 5 foot radiused concave slat.

If you release the arm so it may now swivel 3 degrees you can illuminate it with the same LED at any test distance. This will be the final illuminant to Mirror screw test distance.
Now watch its reflection in each of the end mirrors as the arm moves to each end 3 degree position at another viewing test distance so that at the 3 degree extremities you can just see the LED's reflection in the centre of each small mirror in turn. You are now in the final viewing distance for that particular slat curvature X.

If these illuminant/viewing test distances are not suitable for your room and viewing requirements then do all this again at another X distance.

When a curve is established near to the requirements then a test slat properly cut as a final tryout.

The above are just my first thoughts on a concave test rig and no doubt they can be refined/improved/redesigned .........my thoughts always need to be refined !

[Panrock]

This sounds like an excellent idea Albert and your explanation is clear, but I am still finding this as hard to understand as a real mirror screw! Maybe the best idea would be for me to make one of these rigs and then it will become intuitive. This could prove a very useful tool.

First observations on my laser cut test slats:

1) They'll have to be made of something other than stainless steel! Although I have now obtained a good polish on the reflecting edges, there still seems to be a 'grain' effect disturbing the quality of the image. A mirror screw needs to exhibit a perfect, smooth, mirror finish.

2) The dimensional fits to the large central and the side (peg) indexing holes are excellent. I now have obtained the silver steel shaft.

3) As stated elsewhere, Albert is right and I'm going to have to use concave surfaces. The current convex surfaces make the 2-line 'picture' even smaller! Think the effect of looking in a magnifying shaving mirror. At any distance inside the focus, the 'picture' is bigger (which is what we want). For practical use, so that viewers at some distance can still see the picture, I think the radius of curvature will have to be pretty long. But not so long that the magnifying effect is minimal. I suspect a concave surface will mean the optimal viewing distance, though pleasantly close, will also be exceedingly critical. Yikes! Albert, any chance of a more comprehensive scan of that 1932'ish article and graph?

On another front, Dominic Beesley has now come up with some software as a possible video source. Thanks for this! I'll look at this shortly...

Steve O