@@adriangaleron3293 I agree fully, @Stuff Made Here , you should totally shout some channels you see producing cool content in a video some time, to give a boost to people who do cool stuff.
Superb ! I built my first mechanical TV camera/monitor in the early 1980s. It was monochrome, used red LEDs as a light source and I recorded the images on cassette tape !
Man, Nipkow would be proud to see his ideas are still studied and replicated. Really makes you appreciate the leaps in technology moving image projection has one thru.
Very cool, ever since I learned about these mechanical TVs I had been hoping to find a good quality video of one. I certainly didn't expect it to be a color version!
I'm no expert, but I imagine that if you could get enough of blackout paper or plastic, you could replace the large disk with a tape loop that encircles the LED panel. In that way, you could make the display as wide as you like without drastically increasing the width of the device. As a bonus, you'd lose the distortion you get from right to left. The only caveat is the length of the tape is going to be a function of the operational frequency of your DAC and the amount of columns in your display. If it was too long, you'd have to use a tape loop mechanism like those used in older telephone equipment. You might think about making an analog source for your device. You'd need several light sensors and colored film to filter them with, but then you could have fun building a frequency division duplexing encoder/decoder pair to transmit the signal with. As long as you use shielded wire to transmit the signal, you shouldn't have to worry about interfering with electrical equipment in your area. The encoding would be a simple amplitude modulated multi-stream with a frequency domain for each color of light, possibly with a clocking signal if you have need of it. I don't remember for sure, but I think something similar was used in early television sets, except without the FDD, as you only needed one amplitude signal, as there was only one color of lamp. Tune your set into the proper frequency, and the light brightened and dimmed in time with the spinning of the disk.
With greater width you will have less fps. Limit is in how fast you can change color from "pixel" to "pixel". But yes, loop tape design would take less space and eliminates distortion. Yes, that was exactly earlier tv setup. Camera records analog signal and transmits via radio where this device receives it and plays analog light signal in sync. You just would need to catch both speed and phase of rotation to properly display an image. This approach was later evolved in ray tube tv's, they do basically the same, but signal already contains blank parts to sync receiver phase with transmitter. But same approach: direction of electronic ray changes from pixel to pixel line by line, and analog brightness value is taken from the signal.
i just ran through the numbers, and a 16dpi 2x2" display really wants about 64 inches of belt, running at ... 55 mph. each scan is 2 inches. 16 scans per inch, for 32 scans per frame, for 64 inches a frame. 16 times a second, 1024 inches... 93km/h. ~55mph. scan with two projectors at an inch high, halve that. its doable. as for scanning... um, i just ordered a bunch of the photodiodes... bw61 or something? "spectral response equivalent to human eye". available with RGB filters. for extra.... sigh. use tinted glass. took me years to understand the "vertical" aspect of them, that its just an analog signal for the duration of a "scan". too much stuck in the "traditional" electronic method, distinct elements, pixels..."quanta" so. RGB photodiodes poked where one would be for B&W, 3 signals of intensity, and three RGB leds masked by dot appropriately. pretty simple. getting levels right is tricky? with digital stuff you can easily bust your scan up into "pixels" as you simply sample each diode in sequence... mhz!... the response time of the diode becomes the limiting factor. probably a way to hack a CCD module to just output RGB with no "pixels". you arent interested in that, just the value of what the entire sensor is being hit with at any given time.
😀Thanks! I assembled this TV and thus realized my ten-year dream!! Unfortunately, it is impossible to add sound, since the frame rate is variable and the sound will lag or rush.
Have you tried a cylinder instead of a disc? Imagine you add a wall to the outer edge of disc about 1.5" wide, You put the holes on the wall and you get a perfectly rectangular frame, The disc is placed horizontally so you get the holes spinning horizontally for horizontal scanning like CRT tube, I would like to see a working model of this.
Thanks for the comment - yes, a rotating drum is a possibility to achieve a more rectangular picture. I did some tests a while ago, but the disk was easier to handle in the end.
@@makermac70 You can make the edge small enough like 20mm and have even bigger resolution like 108 lines (1/10 of HD), You would have to use a magnifying lens projecting the small frame and a more powerful RGB light source, the holes will be like 0.5mm each. I can help you with the design of the disc if you take care of the electronics.
I have been waiting for someone to try building a display using two hexagonal spinning prisms (one vertical and one horizontal) like are used for laser displays…
I have been thinking about the use of a Nipkow display for use in a pseudo-steampunk (very) post-apocalypse text mode terminal. It's essentially a bit of worldbuilding where civilization collapsed, and is in the process of recovering. Some group, expecting a collapse, put a bunch of information and interactive educational stuff into very very durable and long-lasting computing devices. The eggs are meant for the owner to bootstrap the computing peripherals with each new part adding capabilities and access to more data. The Nipkow disk display is meant to be a Victorian-accessible text-mode display that could start out very simple with a hand-cranked disc and eventually be improved upon to the point it could act as a very simple remote terminal display. Your display is very encouraging.
Need to put that disc motor and LED timing together in a servo loop. Then add more lines and enlarge the view field. Can't wait for the full HD version.
Probably it is, the jitter might be because the motor is just a simple DC motor and the angular velocity is not stable. It can be improved by putting sync holes for every position on the disc so any jitter would not affect the whole frame, just one scanline. You'll need some means for homing, another IR, or double the slits at the home point then you are at the start of frame when "two slits came way faster than usual". Can be synced in init stage to avoid wasting time during displaying, just ignore the extra slit there.
Although it might not be in the spirit of the original invention, but with a faster board, you could have used PWM to control RGB output. That would mean less pins, less heat as you are switching, and most importantly a more linear color and brightness reproduction. With LEDs, doubling the current will not necessarily produce double light output. While doubling a PWM duty cycle will actually and exactly double the light output. Not using an R2R network DAC also means you avoid non-linearity introduced by resistor tolerances. You can also improve contrast by applying a gamma curve, which is an easy operation on more powerful boards. Edit: when doing the math, it turns out that the ATmega clock at 16 MHz is actually more than fast enough for PWM. For the original 6 bit per pixel with 1024 (32x32) pixels per frame and 30 frames per second, 2 MHz are already sufficient. So you could even push for 8-bit PWM, which would require 8 MHz.
Hi, thanks for your interesting comment on the DACs. I fully agree with the non-linerarity and power consumption issues R2R DACs bring. But I had thought through this and my conclusion was that PWM-DACs would not work so well since the max PWM frequency is only 62.5 kHz at 8 bits on the Mega (and other Aruidnos) by setting the clock prescaler to 1. So the trade-off between a necessary high PWM frequency (in the MHz range), the required AC output signals frequency (around 30kHz) and the LPF characteristics were the blocking point for me. See my own analysis of this on hackster in the comments section. But I may be wrong ! Let me know if I made a wrong assumption...
@@makermac70 Because of the low pixel rate, it should at least theoretically be doable even with the Mega, but I do agree that a bit more headroom would go a long way. You'd basically have to sync the rotation speed to the internal clock of the Mega. My calculation was 1024 pixels * 30 fps * 64 steps (6 bit) = 1.966 MHz. But then you'd have to manually reset the timers with every pixel, and the picture would probably still pump because every cycle the exact duty cycles would vary. With a CPU running at for example 200 MHz, you could easily do 8 bit or more, and still have a lot of headroom, as each pixel would have multiple on-off cycles. While only running at 84 MHz, I have particularly good experiences with Arduino Due. In addition to being a lot faster clock-wise, you also get 32-bits, so calculations like gamma can be a lot faster. Maybe for V2... Anyway, and unbelievably as it is, but modern DLP projectors also use hard on-off PWM, as the mirrors only have two stable positions. As modern color wheels have 6 segments, and each segment requires 8+ bits, and they typically display up to 120 frames per second, you get millions of mirror flips per second.
@@DouglasWalrath MicroPython is usually not the best solution for timing-critical applications, however that is the main language for the Pi Pico. It's of course possible to program in C++ for the Pi Pico, but it's rather complicated to setup the cross-compilation toolchain and a project, including having to interact with make files. I'd rather stay inside the Arduino universe, where you just download the Arduino IDE and be good to go. Generally, ESP32 would be the best solution, as it'd also make it easy to transfer pictures and videos to it via Wifi. It also runs at 240 MHz, has two cores, so you can block one completely for the timing-critical stuff, the ESP32-WROVER module comes with a whopping 8 MB RAM and 4 MB flash, and it provides 16 independent PWM channels with up to 16 bit resolution, 8 channels of which are called "high speed" and 4 high speed clock modules - although like most peripherals in the ESP32, it runs at "only" 80 MHz, which gives a PWM modulation frequency of 160 kHz at 8 bits resolution. You also have the ability to do DMA, which means you can modulate the PWM channels with only little CPU utilization, although DMA is a bit more involved to setup.
To be complete, one could also implement the nipkov camera. Just add a lens and photo diode. The needed ADC could be harder than the resistor-based DAC, because the arduino adc is too slow. One would need to use a dedicated adc chip.
I see how it 'scans' as the holes pass the light... How hard would it be to adjust timings for increasing the number of holes, or even using an vey narrow slit?
The pixel timing is no big issue. Currently with 30fps, each rotation (one complete picture frame) takes 33.3ms with a 32-hole disk we'll get 32x32=1024 pixels per frame so each pixel duration is 32.55us. This could be improved easily for higher resultions. Pixel hole size is a different story - it can not be reduced significantly due to the small amount of light which can pass through. In this design in use approx.1mm^2 which seemed a good value for 3d printing. So if the resolution would need to be increased, the disk needs to get - a lot - bigger...
Would interlacing be possible on something like this? The only reasons I can think of is that the flicker would be unbearable at 30fps, and you wouldn't have the phosphor decay that CRT's have.
thanks for the comment. A flying-spot camera would surely be the perfect addition, but here I was more interested in the display side of things... but a nice idea for a future project
Some ideas: (1) Add reflective tape to the back of the disk as well to minimize lost light (to a blocked aperture) (2) Decrease pinhole spacing such that it's just enough for the LEDs to color transition (increases brightness) (3) If you revert this idea, you can use a fast thermal sensor to build a thermal camera.
The Nipkow design was a great contender up until CRTs. Of course today we can rapidly modulate light intensity but it would have been much harder in the 1930's. Have you researched the light modulation methods from the era? What would they have used, mirrors on galvanometers to switch light toward and away from the spinning holes?
Yes, the "television light bulb" was a real challenge in the beginning. In early mechanical television sytems a neon glow lamp was typically used which produced a dim orange image at 7.5 fps. The signal came from a radio AM receiver unit and was amplified using triodes. Later on, Kerr Cells were used which could produce brighter images in combination with more advanced amplifier systems.
They also had more advanced scanners, for the sets anyway. Starting with mirror drums and eventually getting to the Mihaly-Traub system. Mirror screws are also an interesting design. If you had a small image with sufficient brightness you could project onto a screen, either a large one or something closer to tv set size
I have an idea, instead of using a disk make an conveyor belt like mechanism around led, and instead of round holes make square holes of size of pixel. With conveyor belt mechanism you can make display with square shape and less distorted .
Ever thought of making a barrel version of the nipkow disk, like instead of a plate what if you used a spinning barrel that covers over the whole assembly. I'm wondering if that would make a bigger screen based how wide the ring was and how fast it traveled
Thanks for the suggestion. I had actually thought of this before and even built a prototype. However, the construction was quite bulky, so I discarded it. Balancing the drum also caused problems at high speeds. But the advantage is that there is less distortion in the image.
Any reason a belt with holes couldn't replace the disk with holes for a more compact and rectangular viewing surface? Should be easy enough to test with a belt sander. The closest thing from the time was mirror drums but those don't sound too similar, and they had mirrors for some reason.
if you can speed up a belt enough that should work as well. I never saw this variation, perhaps not that easy. A while ago I did some tests with a rotating drum with holes in the outside which provided a rectangular view, but the disk was easier to handle in the end.
Thanks ! The horizontal viewing angle is around 70°, so similar to an old LCD display. This mostly depends on the thickness of the disk and the hole size.
@@makermac70 This is a great project. I've been wanting to do something similar for a while but you've went all out having all 3 colour channels. Props to you!
I'm curious if there's any reason you didn't try rotating the display window 90° to the top (or bottom) to get more area for the same disc (and in turn, pinhole) coverage. That was my first thought when I saw the original design.
I tried a top aligned configuration in the beginning, but I did not like the aspect ratio, so I moved it to the side to get closer to a 4:3 picture. The area remains the same on either side.
Thanks for the comment Neil, in fact the final image size in one of the issues with this application (esp. at higher resolutions). A lens for magnification was commonly used, but I wanted to preserve the original character of the picture.
Hi, thanks for the question. The disk and the motor frame are both 20cm, so you could perhaps try to scale everything down to 90% . However then the dimensions for the screw holes no longer fit (maybe try to drill them open to original size after printing). Good luck and let me know alter if it worked..
Thanks for the recommendation, I hadn't heard of the Early Television Foundation. The museum is not exactly around the corner, but it would certainly be interesting to have a look. But I'm not sure if my setup is really interesting, though, since it's not an authentic replica of an old TV.
Yes, I even used that configuration in the beginning (at the top). But I did not like the aspect ratio, so I moved it to the side to get closer to a 4:3 picture. BTW it is impossible to see the vertical scanning direction when the disk is spinning.
The problem with this is that you can't make the disk as big as you want it to be without running into significant mechanical difficulties. Increasing the resolution automatically increases the disk diameter, because you cannot make the holes as small as you want. But the speed at the outer edge of the disk at 1m is already almost 100 m/s at 30 revolutions per second! So this quickly becomes very challenging for the material and the motor.
Можно, только диаметр диска вероятно будет исчисляться десятками метров. А вместо электромотора понадобится что-то типа газотурбинного двигателя от М1 Абрамс.
The brightness of the Nipkow display is not sufficient. Usually only a few percent of the light from the LED can be used, the rest is simply absorbed by the disk. This is a fundamental problem and makes it virtually impossible to use as a projector.
Nice, however a brushless DC motor would be more in keeping with the original Baird design. They had no position feedback as you are using, just relying on the AC mains to hold the motor at the right speed & angle - The AC waveform could be easily generated by your Arduino.
Is it that the entire diffused sheet is illuminated a single colour and that the hole over the sheet them displays that single pixel in that colour as it passes? I was expecting a row of LEDs in a straight line behind the disc.
Hi Mike, thanks for the question - correct, the entire LED diffusor screen lights up in only one single colour at any given moment (see LED unit assembly at 4:27). Each passing hole then forms a pixel with this colour. Not a very efficient use of light by the way ... ;-)
@@makermac70 then would it be possible to have multiple of the rgb element in a row for higher resolution and bigger screen? Could you make a 1080p version?
A 1080p version (1920x1080 pixels) would be 60 times bigger compared to the 32x32pixel version (32*60=1920) assuming the same hole size (approx.1mm^2 which seemed a good value for 3d printing). The current display size (expressed as a simple bounding box) is approx.. 22 x 18 mm, so a 1080p version picture would be approx. 1.3 x 1m (!) which needs a huge disk. Alternatively, the hole size can also not be reduced significantly due to the small amount of light which passed through. These are the limitations of mechanical television. The biggest Nipkow disks I have heard of could get no more than about 240 lines of resolution and were really difficult to handle. And unfortunately a parallel string of RGB LEDs does also not help, I'm afraid...
@@makermac70 The disk size is where someone else's idea of using a belt would come in handy, and the screen would be a perfect rectangle. Also using a much brighter (and probably cheaper) white LED in conjunction with a color wheel could compensate for using smaller holes.
this was not intended as a receiver for NBTV signals. The NBTV club standard supports 12.5fps based on analog signals (1Vpp with embedded sync ). I am using higher framerates (25...30fps) with 18bit/pixel color-depth
Wow, this is a great project! We loved it so much that we decided to feature it in this week's episode of The Electromaker Show!: ruclips.net/video/qm7wvYJCvWI/видео.html
With each rotation being one frame, the disc has to spin at 30 revolutions per second to get 30 frames per second. 1800 revolutions per minute. What I'd try is laser cut the disc to have the holes with closer radial spacing, and make two sets of them, each going halfway around the disc, and one set offset halfway between the others. Same 1800 RPM now makes a higher resolution, 60 fields per second, interlaced image. A bigger image requires a larger diameter disc, so drop to a PAL style 25 frame, 50 field per second picture at 1500 RPM. Why do all of these put the image to the side and use "vertical" lines? I'd put it at the top to use "horizontal" lines.
Hi, thanks for your comments. I did use horizontal scanning before, but the aspect size will become 9:16 then (which is quite odd - I wanted to have some kind of "normal" landscape 16:9 format) Yes, lasercutting is indeed a possibility, but I wanted to use more "common" techniques such as 3D printing for this project for everyone being able to reproduce it.
@@electricpaisy6045 As I explained, the size and RPM would remain the same, and the two sets of holes would make an interlaced image for higher resolution at double the *field* rate, just like an interlaced NTSC or PAL TV had a *frame* rate of 30 or 25 but a *field* rate of 60 or 50. Why can't people read and understand what's perfectly and plainly written?
@@greggv8 then i don't understand how you mean to place the holes. I thought you meant both sets being the same. If they should be placed in different positions, what would be the difference to one big set other than the sequence?
@@electricpaisy6045 the offset so that the second set traces paths between the first to create an interlaced image. That increases the apparent resolution without having to spin the disk faster. Field 1 is on half the disk and Field 2 is on the other half.
Hi, thanks for the question. The reason for the Mega was: I needed a lot of IO pins for the 3 parallel DACs and enough internal RAM for the double (frame) buffer. For more details see my project description on hackster (link above).
@@makermac70 Thanks. I've seen your article on IEEE and images were so neat so I couldn't resist my curiousity! I thought it was Fusion 360. Thanks anyway! Good job mate!
I was thinking about that just now lol! It also made a lot of noise and shake violently when turned on, something a unbalanced disk would do when it started spinning I imagine. hahah
Is this compatible with the NBTVA 32 line standard?, I'm thinking of making a display myself as I have the mindsets online televisor kit but I'd like to have something more reliable and of a higher quality
Thanks for the question - No, this was not intended as a receiver for NBTV signals. Mainly because I wanted a color display (NBTV is black and white only) and the NBTV club standard supports 12.5fps based on analog signals (1Vpp with embedded sync ). I am using higher framerates (25...30fps) with 18bit/pixel color-depth.
What an amazing device with awesome results. My mind is blown.
I don’t understand fully how this works, all I know is that it’s really really cool lol.
I love that you like watching small channels! 👍🏼
@@adriangaleron3293 I agree fully, @Stuff Made Here , you should totally shout some channels you see producing cool content in a video some time, to give a boost to people who do cool stuff.
Wow Stuff Made Here is here
I tried building a Nipkow camera and display once, it did not work well. Yours is very satisfying to see.
wow I didn't expect to see you here
It would be nice to see you revisit the homemade record with he man and laser
bees
Superb !
I built my first mechanical TV camera/monitor in the early 1980s.
It was monochrome, used red LEDs as a light source and I recorded the images on cassette tape !
Man, Nipkow would be proud to see his ideas are still studied and replicated. Really makes you appreciate the leaps in technology moving image projection has one thru.
WOW. This is the best version I ever saw. Almost no camera artifacts, high res, very synchronised and in colour! you deserve at least 25k Subs!
Very cool, ever since I learned about these mechanical TVs I had been hoping to find a good quality video of one. I certainly didn't expect it to be a color version!
3D Printing is a total game-changer in my honest opinion, there is so much you can do with it.
Super cool! We use Nipkow disks in one kind of confocal microscopy. Really neat to see this application.
Absolutely Amazing Build!
This is an absolutely beautiful project! Great job dude
I'm no expert, but I imagine that if you could get enough of blackout paper or plastic, you could replace the large disk with a tape loop that encircles the LED panel. In that way, you could make the display as wide as you like without drastically increasing the width of the device. As a bonus, you'd lose the distortion you get from right to left. The only caveat is the length of the tape is going to be a function of the operational frequency of your DAC and the amount of columns in your display. If it was too long, you'd have to use a tape loop mechanism like those used in older telephone equipment.
You might think about making an analog source for your device. You'd need several light sensors and colored film to filter them with, but then you could have fun building a frequency division duplexing encoder/decoder pair to transmit the signal with. As long as you use shielded wire to transmit the signal, you shouldn't have to worry about interfering with electrical equipment in your area. The encoding would be a simple amplitude modulated multi-stream with a frequency domain for each color of light, possibly with a clocking signal if you have need of it.
I don't remember for sure, but I think something similar was used in early television sets, except without the FDD, as you only needed one amplitude signal, as there was only one color of lamp. Tune your set into the proper frequency, and the light brightened and dimmed in time with the spinning of the disk.
With greater width you will have less fps. Limit is in how fast you can change color from "pixel" to "pixel".
But yes, loop tape design would take less space and eliminates distortion.
Yes, that was exactly earlier tv setup. Camera records analog signal and transmits via radio where this device receives it and plays analog light signal in sync. You just would need to catch both speed and phase of rotation to properly display an image.
This approach was later evolved in ray tube tv's, they do basically the same, but signal already contains blank parts to sync receiver phase with transmitter. But same approach: direction of electronic ray changes from pixel to pixel line by line, and analog brightness value is taken from the signal.
I can totally see a hacked up beltgrinder here. A beefy audio setup will be needed but hey.
i just ran through the numbers, and a 16dpi 2x2" display really wants about 64 inches of belt, running at ... 55 mph. each scan is 2 inches. 16 scans per inch, for 32 scans per frame, for 64 inches a frame. 16 times a second, 1024 inches... 93km/h. ~55mph.
scan with two projectors at an inch high, halve that.
its doable.
as for scanning... um, i just ordered a bunch of the photodiodes... bw61 or something? "spectral response equivalent to human eye". available with RGB filters. for extra.... sigh. use tinted glass.
took me years to understand the "vertical" aspect of them, that its just an analog signal for the duration of a "scan". too much stuck in the "traditional" electronic method, distinct elements, pixels..."quanta"
so. RGB photodiodes poked where one would be for B&W, 3 signals of intensity, and three RGB leds masked by dot appropriately. pretty simple. getting levels right is tricky?
with digital stuff you can easily bust your scan up into "pixels" as you simply sample each diode in sequence... mhz!... the response time of the diode becomes the limiting factor.
probably a way to hack a CCD module to just output RGB with no "pixels". you arent interested in that, just the value of what the entire sensor is being hit with at any given time.
Шикарная реализация. Браво!
Cool build, the image quality is very nice, well done!
Amazing, thanks for sharing it. great memories of old mechanical TVs !
very vivid colours
Absolutely great 👍 congratulations
beautifully done. and the picture quality is amazing for what it is
Strip old smartphone for nice diffuser or old led tv.
You could use a diffuser and edge-lit back plane meant for LCD, have LEDs drive it, would give better light distribution.
Amazing work! Subscribed!
Beautiful build.
😀Thanks! I assembled this TV and thus realized my ten-year dream!! Unfortunately, it is impossible to add sound, since the frame rate is variable and the sound will lag or rush.
I want one ! Definitely!
Have you tried a cylinder instead of a disc? Imagine you add a wall to the outer edge of disc about 1.5" wide, You put the holes on the wall and you get a perfectly rectangular frame, The disc is placed horizontally so you get the holes spinning horizontally for horizontal scanning like CRT tube, I would like to see a working model of this.
Thanks for the comment - yes, a rotating drum is a possibility to achieve a more rectangular picture. I did some tests a while ago, but the disk was easier to handle in the end.
@@makermac70 You can make the edge small enough like 20mm and have even bigger resolution like 108 lines (1/10 of HD), You would have to use a magnifying lens projecting the small frame and a more powerful RGB light source, the holes will be like 0.5mm each. I can help you with the design of the disc if you take care of the electronics.
I have been waiting for someone to try building a display using two hexagonal spinning prisms (one vertical and one horizontal) like are used for laser displays…
thanks for detailed instructions and files.
I know a lot of people say the view port is way too small but I think it's beautiful.... Contrary to belief size doesn't matter...
I really think this has potential!
I have been thinking about the use of a Nipkow display for use in a pseudo-steampunk (very) post-apocalypse text mode terminal. It's essentially a bit of worldbuilding where civilization collapsed, and is in the process of recovering. Some group, expecting a collapse, put a bunch of information and interactive educational stuff into very very durable and long-lasting computing devices. The eggs are meant for the owner to bootstrap the computing peripherals with each new part adding capabilities and access to more data.
The Nipkow disk display is meant to be a Victorian-accessible text-mode display that could start out very simple with a hand-cranked disc and eventually be improved upon to the point it could act as a very simple remote terminal display. Your display is very encouraging.
thanks... sounds like an interesting project, please post a link when it is finished. I look forward to seeing it in action.
Need to put that disc motor and LED timing together in a servo loop. Then add more lines and enlarge the view field. Can't wait for the full HD version.
Probably it is, the jitter might be because the motor is just a simple DC motor and the angular velocity is not stable.
It can be improved by putting sync holes for every position on the disc so any jitter would not affect the whole frame, just one scanline.
You'll need some means for homing, another IR, or double the slits at the home point then you are at the start of frame when "two slits came way faster than usual". Can be synced in init stage to avoid wasting time during displaying, just ignore the extra slit there.
Although it might not be in the spirit of the original invention, but with a faster board, you could have used PWM to control RGB output. That would mean less pins, less heat as you are switching, and most importantly a more linear color and brightness reproduction. With LEDs, doubling the current will not necessarily produce double light output. While doubling a PWM duty cycle will actually and exactly double the light output. Not using an R2R network DAC also means you avoid non-linearity introduced by resistor tolerances. You can also improve contrast by applying a gamma curve, which is an easy operation on more powerful boards.
Edit: when doing the math, it turns out that the ATmega clock at 16 MHz is actually more than fast enough for PWM. For the original 6 bit per pixel with 1024 (32x32) pixels per frame and 30 frames per second, 2 MHz are already sufficient. So you could even push for 8-bit PWM, which would require 8 MHz.
Hi, thanks for your interesting comment on the DACs. I fully agree with the non-linerarity and power consumption issues R2R DACs bring. But I had thought through this and my conclusion was that PWM-DACs would not work so well since the max PWM frequency is only 62.5 kHz at 8 bits on the Mega (and other Aruidnos) by setting the clock prescaler to 1. So the trade-off between a necessary high PWM frequency (in the MHz range), the required AC output signals frequency (around 30kHz) and the LPF characteristics were the blocking point for me. See my own analysis of this on hackster in the comments section. But I may be wrong ! Let me know if I made a wrong assumption...
@@makermac70 Because of the low pixel rate, it should at least theoretically be doable even with the Mega, but I do agree that a bit more headroom would go a long way. You'd basically have to sync the rotation speed to the internal clock of the Mega.
My calculation was 1024 pixels * 30 fps * 64 steps (6 bit) = 1.966 MHz. But then you'd have to manually reset the timers with every pixel, and the picture would probably still pump because every cycle the exact duty cycles would vary.
With a CPU running at for example 200 MHz, you could easily do 8 bit or more, and still have a lot of headroom, as each pixel would have multiple on-off cycles.
While only running at 84 MHz, I have particularly good experiences with Arduino Due. In addition to being a lot faster clock-wise, you also get 32-bits, so calculations like gamma can be a lot faster. Maybe for V2...
Anyway, and unbelievably as it is, but modern DLP projectors also use hard on-off PWM, as the mirrors only have two stable positions. As modern color wheels have 6 segments, and each segment requires 8+ bits, and they typically display up to 120 frames per second, you get millions of mirror flips per second.
@@graealex the pi pico might work, it can go up to 133 MHz and defaults to 125 Mhz
@@DouglasWalrath MicroPython is usually not the best solution for timing-critical applications, however that is the main language for the Pi Pico. It's of course possible to program in C++ for the Pi Pico, but it's rather complicated to setup the cross-compilation toolchain and a project, including having to interact with make files. I'd rather stay inside the Arduino universe, where you just download the Arduino IDE and be good to go.
Generally, ESP32 would be the best solution, as it'd also make it easy to transfer pictures and videos to it via Wifi. It also runs at 240 MHz, has two cores, so you can block one completely for the timing-critical stuff, the ESP32-WROVER module comes with a whopping 8 MB RAM and 4 MB flash, and it provides 16 independent PWM channels with up to 16 bit resolution, 8 channels of which are called "high speed" and 4 high speed clock modules - although like most peripherals in the ESP32, it runs at "only" 80 MHz, which gives a PWM modulation frequency of 160 kHz at 8 bits resolution. You also have the ability to do DMA, which means you can modulate the PWM channels with only little CPU utilization, although DMA is a bit more involved to setup.
The first question is whether the LED is able to switch at 8MHz or not.
To be complete, one could also implement the nipkov camera. Just add a lens and photo diode. The needed ADC could be harder than the resistor-based DAC, because the arduino adc is too slow. One would need to use a dedicated adc chip.
Great video and project.
Amazing work, thanks for sharing!
You could use mirrors and lenses to make the picture bigger, reposition it, and correct for distortion.
I see how it 'scans' as the holes pass the light...
How hard would it be to adjust timings for increasing the number of holes, or even using an vey narrow slit?
The pixel timing is no big issue. Currently with 30fps, each rotation (one complete picture frame) takes 33.3ms
with a 32-hole disk we'll get 32x32=1024 pixels per frame so each pixel duration is 32.55us. This could be improved easily for higher resultions.
Pixel hole size is a different story - it can not be reduced significantly due to the small amount of light which can pass through. In this design in use approx.1mm^2 which seemed a good value for 3d printing.
So if the resolution would need to be increased, the disk needs to get - a lot - bigger...
Would interlacing be possible on something like this? The only reasons I can think of is that the flicker would be unbearable at 30fps, and you wouldn't have the phosphor decay that CRT's have.
Now you just need to make a version with only Vacuum Tube Logic, and Non-LED Lights!
Karl Nipkow would be proud!
Just wow! So cooooool !!!
Wow! The great thing about 3D printing is you can be very precise with hole placement and size. How about a flying spot camera?
thanks for the comment. A flying-spot camera would surely be the perfect addition, but here I was more interested in the display side of things... but a nice idea for a future project
that looks superb!
Some ideas: (1) Add reflective tape to the back of the disk as well to minimize lost light (to a blocked aperture) (2) Decrease pinhole spacing such that it's just enough for the LEDs to color transition (increases brightness) (3) If you revert this idea, you can use a fast thermal sensor to build a thermal camera.
this music bops
you have to try the mirror screw technology !
a thing of beauty
The Nipkow design was a great contender up until CRTs. Of course today we can rapidly modulate light intensity but it would have been much harder in the 1930's. Have you researched the light modulation methods from the era? What would they have used, mirrors on galvanometers to switch light toward and away from the spinning holes?
Yes, the "television light bulb" was a real challenge in the beginning. In early mechanical television sytems a neon glow lamp was typically used which produced a dim orange image at 7.5 fps. The signal came from a radio AM receiver unit and was amplified using triodes. Later on, Kerr Cells were used which could produce brighter images in combination with more advanced amplifier systems.
They also had more advanced scanners, for the sets anyway. Starting with mirror drums and eventually getting to the Mihaly-Traub system. Mirror screws are also an interesting design. If you had a small image with sufficient brightness you could project onto a screen, either a large one or something closer to tv set size
Wooow impresionant job, well done 10/10*
Amazing
Creative video, thanks :)
I have an idea, instead of using a disk make an conveyor belt like mechanism around led, and instead of round holes make square holes of size of pixel. With conveyor belt mechanism you can make display with square shape and less distorted .
Ever thought of making a barrel version of the nipkow disk, like instead of a plate what if you used a spinning barrel that covers over the whole assembly. I'm wondering if that would make a bigger screen based how wide the ring was and how fast it traveled
Thanks for the suggestion. I had actually thought of this before and even built a prototype. However, the construction was quite bulky, so I discarded it. Balancing the drum also caused problems at high speeds. But the advantage is that there is less distortion in the image.
So cool!
Any reason a belt with holes couldn't replace the disk with holes for a more compact and rectangular viewing surface? Should be easy enough to test with a belt sander.
The closest thing from the time was mirror drums but those don't sound too similar, and they had mirrors for some reason.
if you can speed up a belt enough that should work as well. I never saw this variation, perhaps not that easy.
A while ago I did some tests with a rotating drum with holes in the outside which provided a rectangular view, but the disk was easier to handle in the end.
Wonder if the triple DAC could be replaced by an addressable RGB led these days. Great project. Cheers
I'll be back when it's in 4K
😂😂😂 brilliant comment 👍🏻
For 4K the dude needs to print 3999 more spinning discs 😱
Good lord, that's gorgeous color, man! How do you do it?
How much could one scale this up?
Cool. Отличный проект!!!
How is the viewing angle? It could probably be increased by countersinking the holes. Nice project!
Thanks ! The horizontal viewing angle is around 70°, so similar to an old LCD display. This mostly depends on the thickness of the disk and the hole size.
@@makermac70 This is a great project. I've been wanting to do something similar for a while but you've went all out having all 3 colour channels. Props to you!
Could you post a slowmo of the projection by any chance please?
Trop fort
That’s worth a sub. So I did.
very good
I'm curious if there's any reason you didn't try rotating the display window 90° to the top (or bottom) to get more area for the same disc (and in turn, pinhole) coverage. That was my first thought when I saw the original design.
I tried a top aligned configuration in the beginning, but I did not like the aspect ratio, so I moved it to the side to get closer to a 4:3 picture. The area remains the same on either side.
For real any plans for adding a video input? Hdmi or composite or whatever. That would be so cool
Excellent. Maybe add a fresnel lens ala the film Brazil.
Thanks for the comment Neil, in fact the final image size in one of the issues with this application (esp. at higher resolutions). A lens for magnification was commonly used, but I wanted to preserve the original character of the picture.
"Ay what's your screen's refresh rate?"
"1600 RPM."
"What?"
This is an amazing build, if I were wanting to convert it to work on a 18cm x18cm bed, do you have any suggestions on where to begin?
Hi, thanks for the question. The disk and the motor frame are both 20cm, so you could perhaps try to scale everything down to 90% . However then the dimensions for the screw holes no longer fit (maybe try to drill them open to original size after printing). Good luck and let me know alter if it worked..
pretty damn cool
Just found you. I'm obsessed with prewar mechanica tvs. Did you show this to the guys at early television foundation? They'd love it
Thanks for the recommendation, I hadn't heard of the Early Television Foundation. The museum is not exactly around the corner, but it would certainly be interesting to have a look.
But I'm not sure if my setup is really interesting, though, since it's not an authentic replica of an old TV.
I remember seeing a mono one playing from an audio cassette tape in the 80s
The viewing hole can be placed at the bottom or top of the disc. This would allow a horizontal pattern to form properly as an image.
Yes, I even used that configuration in the beginning (at the top). But I did not like the aspect ratio, so I moved it to the side to get closer to a 4:3 picture. BTW it is impossible to see the vertical scanning direction when the disk is spinning.
Neat!
Holy shit, this is sick!
john logi would have been so pleased
I just wonder, if you made a really large disc, and you put like many holes in it, would you then be able to reach 500 lines, for instance?
The problem with this is that you can't make the disk as big as you want it to be without running into significant mechanical difficulties. Increasing the resolution automatically increases the disk diameter, because you cannot make the holes as small as you want. But the speed at the outer edge of the disk at 1m is already almost 100 m/s at 30 revolutions per second! So this quickly becomes very challenging for the material and the motor.
@@makermac70, I see, so the mechanical TV would have been always doomed to be replaced by something else.
Any way to connect a hdmi cable to it so I can play my competitive fps games like a chad?
Прекрасная работа! Возможно ли сделать что-то подобное, но с разрешением Full HD?
Можно, только диаметр диска вероятно будет исчисляться десятками метров. А вместо электромотора понадобится что-то типа газотурбинного двигателя от М1 Абрамс.
can you use a lens on this to make a projector?
The brightness of the Nipkow display is not sufficient. Usually only a few percent of the light from the LED can be used, the rest is simply absorbed by the disk. This is a fundamental problem and makes it virtually impossible to use as a projector.
Nice, however a brushless DC motor would be more in keeping with the original Baird design. They had no position feedback as you are using, just relying on the AC mains to hold the motor at the right speed & angle - The AC waveform could be easily generated by your Arduino.
Is it that the entire diffused sheet is illuminated a single colour and that the hole over the sheet them displays that single pixel in that colour as it passes? I was expecting a row of LEDs in a straight line behind the disc.
Hi Mike, thanks for the question - correct, the entire LED diffusor screen lights up in only one single colour at any given moment (see LED unit assembly at 4:27). Each passing hole then forms a pixel with this colour. Not a very efficient use of light by the way ... ;-)
@@makermac70 then would it be possible to have multiple of the rgb element in a row for higher resolution and bigger screen? Could you make a 1080p version?
A 1080p version (1920x1080 pixels) would be 60 times bigger compared to the 32x32pixel version (32*60=1920) assuming the same hole size (approx.1mm^2 which seemed a good value for 3d printing). The current display size (expressed as a simple bounding box) is approx.. 22 x 18 mm, so a 1080p version picture would be approx. 1.3 x 1m (!) which needs a huge disk.
Alternatively, the hole size can also not be reduced significantly due to the small amount of light which passed through.
These are the limitations of mechanical television. The biggest Nipkow disks I have heard of could get no more than about 240 lines of resolution and were really difficult to handle. And unfortunately a parallel string of RGB LEDs does also not help, I'm afraid...
@@makermac70 The disk size is where someone else's idea of using a belt would come in handy, and the screen would be a perfect rectangle. Also using a much brighter (and probably cheaper) white LED in conjunction with a color wheel could compensate for using smaller holes.
BRO I LIKE THE MUSIC ITS LIKE SOVJETWAVE
You should sell kits :)
Thanks for the suggestion, Kara... however this was rather meant as a DIY project description. But I'll think about it :-)
@@makermac70 mainly just the pcb for those of us who are not good with breadboards or electronics in general
@@makermac70 :D
Will this display standard NBTV monochrome images as well as color?
this was not intended as a receiver for NBTV signals. The NBTV club standard supports 12.5fps based on analog signals (1Vpp with embedded sync ). I am using higher framerates (25...30fps) with 18bit/pixel color-depth
Cool
Wow, this is a great project! We loved it so much that we decided to feature it in this week's episode of The Electromaker Show!: ruclips.net/video/qm7wvYJCvWI/видео.html
Thanks so much for sharing my work in your wonderful channel - it's an honor for me !
Great!
With each rotation being one frame, the disc has to spin at 30 revolutions per second to get 30 frames per second. 1800 revolutions per minute.
What I'd try is laser cut the disc to have the holes with closer radial spacing, and make two sets of them, each going halfway around the disc, and one set offset halfway between the others.
Same 1800 RPM now makes a higher resolution, 60 fields per second, interlaced image.
A bigger image requires a larger diameter disc, so drop to a PAL style 25 frame, 50 field per second picture at 1500 RPM.
Why do all of these put the image to the side and use "vertical" lines? I'd put it at the top to use "horizontal" lines.
Hi, thanks for your comments. I did use horizontal scanning before, but the aspect size will become 9:16 then (which is quite odd - I wanted to have some kind of "normal" landscape 16:9 format)
Yes, lasercutting is indeed a possibility, but I wanted to use more "common" techniques such as 3D printing for this project for everyone being able to reproduce it.
Why two sets of holes? The rpm would go down but the edge of the disc would have the same speed. Also the disc has to be double the size.
@@electricpaisy6045 As I explained, the size and RPM would remain the same, and the two sets of holes would make an interlaced image for higher resolution at double the *field* rate, just like an interlaced NTSC or PAL TV had a *frame* rate of 30 or 25 but a *field* rate of 60 or 50.
Why can't people read and understand what's perfectly and plainly written?
@@greggv8 then i don't understand how you mean to place the holes. I thought you meant both sets being the same. If they should be placed in different positions, what would be the difference to one big set other than the sequence?
@@electricpaisy6045 the offset so that the second set traces paths between the first to create an interlaced image. That increases the apparent resolution without having to spin the disk faster. Field 1 is on half the disk and Field 2 is on the other half.
What is the reason for choosing the Arduino Mega for this project vs an Uno or some other smaller board like a Teensy?
Hi, thanks for the question. The reason for the Mega was: I needed a lot of IO pins for the 3 parallel DACs and enough internal RAM for the double (frame) buffer. For more details see my project description on hackster (link above).
Cool shit. What CAD have you used to develop 3D models?
Thanks Alex - I have used a custom made tool to design the Nipkow disk itself, the rest was simply designed in TinkerCAD.
@@makermac70 Thanks. I've seen your article on IEEE and images were so neat so I couldn't resist my curiousity! I thought it was Fusion 360. Thanks anyway! Good job mate!
But does it run Crysis?
Also make a camera with the same principle the you have a closed loop
Corre doom?
If we take this and a circuit printer, then we can have an entirely 3d printed TV!
Bruh imagine playing mw3 on that thing
i was today years old when i realized Rolf's TV in Ed Edd n Eddy was likely a mechanical TV based purely on display port size vs size of the unit
I was thinking about that just now lol!
It also made a lot of noise and shake violently when turned on, something a unbalanced disk would do when it started spinning I imagine. hahah
Is this compatible with the NBTVA 32 line standard?, I'm thinking of making a display myself as I have the mindsets online televisor kit but I'd like to have something more reliable and of a higher quality
Thanks for the question - No, this was not intended as a receiver for NBTV signals. Mainly because I wanted a color display (NBTV is black and white only) and the NBTV club standard supports 12.5fps based on analog signals (1Vpp with embedded sync ). I am using higher framerates (25...30fps) with 18bit/pixel color-depth.
@@makermac70 ah ok, would the disc and parts be compatible with an arduinovisor setup?, I would imagine it wouldn't take much adjustment
sorry, I don't know about the arduinovisor setup
Coulda became a crazy hit if he just ran doom on it
I want to play Max Headroom on it. And Video Killed the Radio Star.
LOL arduino mega - what is this 15 years ago?
I am disappointed you didn't show Bad Apple!! on it.
I had some scenes from GOT converted originally (looked not too bad), but didn't include them in the video in the end...
William osman should watch this video