WEBVTT Kind: captions Language: en 00:00:00.080 --> 00:00:03.929 In my last video, we explored how analog television works. 00:00:03.929 --> 00:00:07.430 You can check out the whole video either through this card or through the link in the description, 00:00:07.430 --> 00:00:09.550 but here’s a brief overview. 00:00:09.550 --> 00:00:14.550 At its core, analog television is just an amplitude modulated radio transmission where 00:00:14.550 --> 00:00:18.820 the strength of the signal dictates brightness of a light source, with a strong signal producing 00:00:18.820 --> 00:00:22.060 a bright light, and a weak signal producing little to no light. 00:00:22.060 --> 00:00:26.160 The television set uses an electromagnet to deflect the source of the light, an electron 00:00:26.160 --> 00:00:31.050 beam, in a pattern called a raster, which is really just a ton of horizontal lines. 00:00:31.050 --> 00:00:35.470 This has the effect of producing a glowing series of lines on the face of a picture tube. 00:00:35.470 --> 00:00:39.410 The electronics of the TV set line up the incoming signal with the movement of the beam 00:00:39.410 --> 00:00:43.640 to create an image, with each part of the imaging being drawn brightly or darkly along 00:00:43.640 --> 00:00:46.090 with the signal’s instantaneous strength. 00:00:46.090 --> 00:00:48.540 With everything in alignment, you get an image. 00:00:48.540 --> 00:00:52.920 By far the most complicated part of making an image appear on the screen is making that 00:00:52.920 --> 00:00:54.480 raster pattern. 00:00:54.480 --> 00:00:58.210 The electronic components and other crucial parts such as the picture tube and deflection 00:00:58.210 --> 00:01:02.630 yoke are primitive by today’s standards, but still pretty complicated. 00:01:02.630 --> 00:01:06.350 We really just need a way make a repeating pattern of lines from a light source, there’s 00:01:06.350 --> 00:01:08.830 got to be a simpler way to do it! 00:01:08.830 --> 00:01:10.740 Enter: Mechanical Television. 00:01:10.740 --> 00:01:15.720 The earliest televisions actually got some of their inspiration from fax machines, really?, 00:01:15.720 --> 00:01:18.060 and relied on a couple of important developments. 00:01:18.060 --> 00:01:21.180 So first a bit about the fax machines. 00:01:21.180 --> 00:01:24.710 Facsimile transmission actually predates the telephone? 00:01:24.710 --> 00:01:29.350 (what?), with images such as signatures being commercially reproduced over telegraph wires 00:01:29.350 --> 00:01:34.930 as far back as 1865, and the earliest fax-like device being invented by Scottish inventor 00:01:34.930 --> 00:01:36.910 Alexander Bain in 1846. 00:01:36.910 --> 00:01:41.540 Now, I’m simplifying a great deal here, but the theory was that if you could synchronize 00:01:41.540 --> 00:01:46.200 the movement of a scanning device with a drawing device, you could replicate an image. 00:01:46.200 --> 00:01:50.250 If you scanned a piece of paper line by line and sent a signal over a wire to match the 00:01:50.250 --> 00:01:54.590 darkness of the ink, you could reproduce the image by syncing up a drawing mechanism with 00:01:54.590 --> 00:01:56.100 the scanning one. 00:01:56.100 --> 00:01:59.400 These early fax machines worked, but they were very slow. 00:01:59.400 --> 00:02:03.950 Nevertheless, they showed us that you could, via electro-mechanical means, reproduce an 00:02:03.950 --> 00:02:04.950 image. 00:02:04.950 --> 00:02:10.690 Fast forward to 1884, and 23-year old Paul Julius Gottlieb Nipkow created the Nipkow 00:02:10.690 --> 00:02:11.690 disc. 00:02:11.690 --> 00:02:14.610 This is the core of most mechanical television systems. 00:02:14.610 --> 00:02:19.550 Nipkow realized that a spinning disc could methodically scan an image line-by-line simply 00:02:19.550 --> 00:02:23.050 by placing evenly-spaced holes in a spiral pattern. 00:02:23.050 --> 00:02:24.490 This is a home-made Nipkow disc. 00:02:24.490 --> 00:02:29.590 I took a really awful vinyl record that I’d be happy to destroy, and marked 32 divisions 00:02:29.590 --> 00:02:33.730 around the circumference, like 32 very skinny pie slices. 00:02:33.730 --> 00:02:38.020 Then I methodically drilled a hole along these lines, with each hole being drilled a 32nd 00:02:38.020 --> 00:02:42.440 of an inch (roughly .8 milimeters) closer to the center than the last. 00:02:42.440 --> 00:02:44.940 The result is a spiral pattern of holes. 00:02:44.940 --> 00:02:48.450 This might not seem like much, but it’s actually extremely clever. 00:02:48.450 --> 00:02:53.110 If you put a square-ish shaped mask in front of the holes, its height being slightly less 00:02:53.110 --> 00:02:57.110 than the distance between the holes, you’ve made a device which mechanically creates a 00:02:57.110 --> 00:02:59.940 raster scan using these physical holes. 00:02:59.940 --> 00:03:04.901 John Logie Baird realized that with this Nipkow disc, you could in theory focus an image with 00:03:04.901 --> 00:03:09.340 a lens onto the disc, and you could use a light sensor to give an instantaneous reading 00:03:09.340 --> 00:03:13.510 of how bright each part of the image was, with the holes in the nipkow disc serving 00:03:13.510 --> 00:03:16.770 as a way to divide the image into transmittable pieces. 00:03:16.770 --> 00:03:21.870 Back in 1873, Willoughby Smith discovered the photoconductivity of selenium, and with 00:03:21.870 --> 00:03:26.500 this knowledge Baird used some selenium to create the light sensor for his televisor. 00:03:26.500 --> 00:03:31.599 I’ve mounted this Nipkow disc to an AC motor which will spin it at 1,800 RPM, giving a 00:03:31.599 --> 00:03:34.640 complete revolution 30 times per second. 00:03:34.640 --> 00:03:36.480 Before I turn it on, look through the mask. 00:03:36.480 --> 00:03:41.020 I’ve put an extremely bright LED behind the disc so you can see the holes. 00:03:41.020 --> 00:03:45.209 As I slowly turn the disc, you’ll see that only one hole is visible at a time, and each 00:03:45.209 --> 00:03:48.110 hole gets closer to the left than the next one. 00:03:48.110 --> 00:03:52.460 When I switch the motor on, the holes blend into a moving line, and as it gets faster, 00:03:52.460 --> 00:03:55.069 the line seems to widen into a square. 00:03:55.069 --> 00:03:59.099 This square is very uneven because my homemade Nipkow disc was made hastily and with poor 00:03:59.099 --> 00:04:00.099 precision. 00:04:00.099 --> 00:04:01.099 But here’s the key. 00:04:01.099 --> 00:04:04.230 Only one of the holes is actually visible through the mask at any given moment. 00:04:04.230 --> 00:04:06.660 It’s just moving too fast to see. 00:04:06.660 --> 00:04:11.180 Baird used the selenium light sensor to create a signal from an image being scanned by the 00:04:11.180 --> 00:04:15.950 disc, and on the receiving end, another identical disc would spin at precisely the same speed, 00:04:15.950 --> 00:04:20.280 and a light source such as a neon lamp would vary its brightness along with the signal 00:04:20.280 --> 00:04:24.599 strength presented by the light sensor, and thus, you’d get an image. 00:04:24.599 --> 00:04:27.289 I shall now attempt to show you how this worked. 00:04:27.289 --> 00:04:30.990 Now before you get too excited, I’ll admit that my mechanical television doesn’t work 00:04:30.990 --> 00:04:32.059 as well as I had hoped. 00:04:32.059 --> 00:04:35.599 And that’s all on my insistence in using crap I had laying around, rather than going 00:04:35.599 --> 00:04:37.779 through the process to make a proper LED driver. 00:04:37.779 --> 00:04:41.900 However, I hope you’ll get an understanding of what’s going on. 00:04:41.900 --> 00:04:45.099 This 10W LED chip is what we’ll use as a light source. 00:04:45.099 --> 00:04:49.680 It’s really bright and, importantly, it can react very quickly to changes in the voltage 00:04:49.680 --> 00:04:50.680 it receives. 00:04:50.680 --> 00:04:54.050 First, I’ll simply power the LED continuously. 00:04:54.050 --> 00:04:57.780 As the disc spins up, the lines start to blend into each other, and eventually the whole 00:04:57.780 --> 00:04:59.580 “screen” is illuminated. 00:04:59.580 --> 00:05:04.249 Now, I’m going to switch the LED on and off at a higher and higher frequency. 00:05:04.249 --> 00:05:06.110 First, 5 hz. 00:05:06.110 --> 00:05:09.879 The screen appears to just be flashing, nothing too extraordinary, but you might be able to 00:05:09.879 --> 00:05:12.909 see some odd stuff happening as the light switches states. 00:05:12.909 --> 00:05:15.779 Now I’ll switch it on and off at 60 hz. 00:05:15.779 --> 00:05:17.639 Something odd starts to be visible here. 00:05:17.639 --> 00:05:21.809 See, the disc makes a complete revolution 30 times per second, and with the light flashing 00:05:21.809 --> 00:05:26.059 at twice that frequency, only some of the holes are lit up as the disc passes over the 00:05:26.059 --> 00:05:27.219 LED. 00:05:27.219 --> 00:05:29.719 Now let’s move to 1,800 hz. 00:05:29.719 --> 00:05:33.469 Frequencies that are a multiple of 30 will appear stable as an even number of pulses 00:05:33.469 --> 00:05:35.289 fit within each revolution. 00:05:35.289 --> 00:05:38.110 If you mess with that, though, things get weird. 00:05:38.110 --> 00:05:42.530 Bumping the frequency up just a tad makes the pattern move in relation to the disc. 00:05:42.530 --> 00:05:46.809 The holes in the disc are directly responsible for creating the patterns you see. 00:05:46.809 --> 00:05:50.189 Without the disc, the LED appears to just be continuously illuminated. 00:05:50.189 --> 00:05:51.189 But, it’s not. 00:05:51.189 --> 00:05:53.469 It’s flashing really quickly. 00:05:53.469 --> 00:05:57.819 The disc allows for that flashing to be visible because it physically obscures different parts 00:05:57.819 --> 00:05:59.889 of the light source over time. 00:05:59.889 --> 00:06:04.059 This is just like the electron beam in the CRT television, except instead of electromagnets 00:06:04.059 --> 00:06:08.479 moving a beam across the surface of a picture tube, the light source is physically moved 00:06:08.479 --> 00:06:10.180 via the location of these holes. 00:06:10.180 --> 00:06:14.860 It’s a pretty crafty way of producing a raster scan, and it actually works. 00:06:14.860 --> 00:06:18.280 This is the best imagery I could get my televisor to produce. 00:06:18.280 --> 00:06:22.099 This pattern was generated through manipulating audio samples in Audacity. 00:06:22.099 --> 00:06:26.969 To give you an idea of how poorly this mechanical TV works, well the image I intended to make 00:06:26.969 --> 00:06:31.219 was not a map of the world as this vaguely suggests, but that of a circle. 00:06:31.219 --> 00:06:33.710 Here’s a look at true video. 00:06:33.710 --> 00:06:38.259 What you’re seeing here is a very low contrast, very low resolution image of Seth Meyers. 00:06:38.259 --> 00:06:42.130 I mean obviously, how could you not recognize him? 00:06:42.130 --> 00:06:46.289 Yeah OK, it’s garbage, but you can see that there is certainly something there and it’s 00:06:46.289 --> 00:06:47.430 moving slightly. 00:06:47.430 --> 00:06:51.080 Like a talk show host’s head might when said talk show host is talking. 00:06:51.080 --> 00:06:52.180 On his show. 00:06:52.180 --> 00:06:56.599 To make this image, I simply placed my phone behind the televisor with the screen brightness 00:06:56.599 --> 00:07:01.650 all the way up, and I placed this solar panel with an audio cord patched into it into one 00:07:01.650 --> 00:07:05.689 of my trusty Tascam DR-05 audio recorders, which I use all the time. 00:07:05.689 --> 00:07:08.229 In fact there’s one in my pocket right now. 00:07:08.229 --> 00:07:11.020 And yes, that’s directly from a solar garden light. 00:07:11.020 --> 00:07:14.919 The solar panel would produce a high current whenever it saw bright light, and it would 00:07:14.919 --> 00:07:17.409 produce low current with less light. 00:07:17.409 --> 00:07:18.429 Duh. 00:07:18.429 --> 00:07:22.369 As the disc spun, it would only allow the tiniest bit of the image through to the solar 00:07:22.369 --> 00:07:24.639 panel at any given time. 00:07:24.639 --> 00:07:29.050 This would produce a quickly varying signal with amplitude corresponding to image brightness. 00:07:29.050 --> 00:07:33.559 The TASCAM would just encode these relative brightnesses as sound samples, at a sample 00:07:33.559 --> 00:07:38.759 rate of 48 kilohertz, and then because I’m really lazy, I just hooked my LED into an 00:07:38.759 --> 00:07:41.770 audio amplifier and played that sound back. 00:07:41.770 --> 00:07:45.779 The LED would become brighter with a stronger signal from the amplifier, though as it’s 00:07:45.779 --> 00:07:49.449 a diode it would filter out any AC components of the signal. 00:07:49.449 --> 00:07:52.679 Quite honestly I’m amazed it produces anything at all. 00:07:52.679 --> 00:07:58.339 I opened the file in Audacity just to see what it looked like, and it’s pretty intriguing. 00:07:58.339 --> 00:08:15.089 Here’s what it sounds like, for those interested. 00:08:15.089 --> 00:08:26.750 Now, in case it’s not obvious, let’s go over the reasons mechanical television didn’t 00:08:26.750 --> 00:08:27.750 catch on. 00:08:27.750 --> 00:08:31.300 First, up until now, I’ve not let you hear what this sounds like. 00:08:31.300 --> 00:08:35.880 Here’s what a 12 inch vinyl disc sounds like at 1,800 RPM. 00:08:47.000 --> 00:08:49.120 I’m sure that would never get old. 00:08:49.150 --> 00:08:54.280 But aside from that, there are just so many practical concerns with mechanical TV. 00:08:54.280 --> 00:08:55.490 First of all, the image is tiny. 00:08:55.490 --> 00:08:59.160 And it’s a horribly low resolution--only 32 lines. 00:08:59.160 --> 00:09:03.180 That’s the only reason a signal can be recorded as an audio file. 00:09:03.180 --> 00:09:05.150 Not a lot of bandwidth is needed. 00:09:05.150 --> 00:09:09.840 Because the disc obscures almost all of the light source, hardly any light gets through. 00:09:09.840 --> 00:09:14.900 This LED is fricken bright, it’s painful to look at directly, and yet through the Nipkow 00:09:14.900 --> 00:09:18.900 disc, nearly all of the light is blocked, and it produces a dull image. 00:09:18.900 --> 00:09:23.200 When these devices were first in development, the light source would often be a neon lamp, 00:09:23.200 --> 00:09:25.630 like the orange light in a powerstrip’s switch. 00:09:25.630 --> 00:09:29.090 Imagine how dark the image would be with only that for a light source. 00:09:29.090 --> 00:09:33.340 One of the biggest troubles with mechanical television is image synchronization. 00:09:33.340 --> 00:09:37.021 Because we’re using a big spinning thing to divide the light into chunks, the disc 00:09:37.021 --> 00:09:42.060 has to be in precisely the right place if you want the image to land where it should. 00:09:42.060 --> 00:09:46.730 If we take the mask away, you can see that the image just repeats itself over and over. 00:09:46.730 --> 00:09:50.930 But each adjacent image is actually shifted one line up or down. 00:09:50.930 --> 00:09:55.110 The most critical part of synchronization was ensuring the disc is spinning at the exact 00:09:55.110 --> 00:09:59.630 same speed as the scanning disc of a camera, but it would also be necessary to slow down 00:09:59.630 --> 00:10:03.630 or speed up the disc in slight increments to get the image aligned with the viewing 00:10:03.630 --> 00:10:06.930 mask, and with the top and bottom in the right place. 00:10:06.930 --> 00:10:10.700 But the most damning problem is that of geometry. 00:10:10.700 --> 00:10:15.830 Imagine we wanted to make a display with the resolution and size of this small CRT television. 00:10:15.830 --> 00:10:19.250 Well, the face of the tube is about 15 cm wide. 00:10:19.250 --> 00:10:24.250 With 480 lines of resolution, there would need to be 480 holes in the nipkow disc. 00:10:24.250 --> 00:10:28.060 Remember, only one hole can be seen through the mask at once for this to work, so the 00:10:28.060 --> 00:10:31.850 holes have to be at a minimum 15 centimeters apart. 00:10:31.850 --> 00:10:35.930 So the disc’s circumference would have to be 72 meters, with a diameter of roughly 23 00:10:35.930 --> 00:10:38.290 meters, or about 75 feet. 00:10:38.290 --> 00:10:40.460 I live in a building that’s 6 stories tall. 00:10:40.460 --> 00:10:44.180 A mechanical television to rival this TV would be taller than my building! 00:10:44.180 --> 00:10:49.090 And, it would have to spin at 1,800 RPM just like this one to make 30 frames per second 00:10:49.090 --> 00:10:50.130 possible. 00:10:50.130 --> 00:10:51.790 This things scares me spinning this fast. 00:10:51.790 --> 00:10:54.940 I’m pretty sure a 75 foot disc would just explode. 00:10:54.940 --> 00:10:56.380 In fact, let’s do the math. 00:10:56.380 --> 00:11:01.950 A 72 meter circumference means that the edge of the disc would travel 2.160 kilometers 00:11:01.950 --> 00:11:05.790 per second, or well above mach 6. 00:11:05.790 --> 00:11:06.980 Yeah. 00:11:06.980 --> 00:11:10.740 If the disc were rolling, it would make it from New York to Los Angeles in about 35 minutes--not 00:11:10.740 --> 00:11:14.610 in a straight line, mind you, but by traveling along actual roads. 00:11:14.610 --> 00:11:17.900 So, the Baird television system didn’t get too far. 00:11:17.900 --> 00:11:22.110 It was certainly genious and is an important part of the history of television. 00:11:22.110 --> 00:11:26.740 But is was far too limited, clunky, and, to be honest, it had crappy image quality. 00:11:26.740 --> 00:11:31.490 I’ve added some links in the description to videos of mechanical televisions that actually 00:11:31.490 --> 00:11:34.820 work, and I think you’ll agree that’s they’re pretty cool, but it’s a damn good 00:11:34.820 --> 00:11:37.220 thing they didn’t become mainstream. 00:11:37.220 --> 00:11:39.750 As always, thank you so much for watching. 00:11:39.750 --> 00:11:42.610 If you liked this video, a thumbs up would be most appreciated. 00:11:42.610 --> 00:11:47.680 I’m absolutely thrilled that this channel has over 21,000 subscribers now! 00:11:47.680 --> 00:11:49.680 I never thought that would be possibl. 00:11:49.680 --> 00:11:53.360 If you’re not one of the people in that number, and you liked this video, I humbly 00:11:53.360 --> 00:11:56.430 ask that you become one of them by pressing that subscribe button. 00:11:56.430 --> 00:11:59.950 I’m doing my best to keep videos like this headed your way, and I’ll see you next time!