Recreating An Analog TV Test Pattern

While most countries have switched to digital broadcasting, and most broadcasts themselves have programming on 24/7 now, it’s hard to remember the ancient times of analog broadcasts that would eventually stop sometime late at night, displaying a test pattern instead of infomercials or reruns of an old sitcom. They were useful for various technical reasons including calibrating the analog signals. Some test patterns were simply camera feeds of physical cards, but if you wanted the most accurate and reliable test patterns you’d need a Philips pattern generator which created the pattern with hardware instead, and you can build your own now because the designs for these devices were recently open-sourced.

[Matt] from [Matt’s TV Barn], as the name implies, has an interest in various television technologies, and this is such an iconic piece of machinery that can’t easily be purchased, so he set about recreating the signal generator on his own. All of the source for these cards can be found at this GitHub page for anyone interested, and most of the parts can still be found today. The main exception being the adjustable inductors, which [Matt] needed to build from scratch. Soldering everything to the board was a bit nerve-wracking but with the help of a robotic lifter (and a lot of patience) the board was soldered without any damage.

The final step of the build is to get everything up and running by following the original testing and commissioning procedures that were developed and released by the original company that built these devices. This was a feat in its own right, as there are a large number of complex tools needed to perform these operations that [Matt] plans to talk about in future videos, but with everything completed he has a piece of TV history that’s about as authentic as anyone could possibly get anymore. You don’t have to go to these lengths to get the aesthetic of analog TV, though. Take a look at this restored analog TV that uses a Raspberry Pi to generate its analog signals.

24 thoughts on “Recreating An Analog TV Test Pattern

  1. “While most… it’s hard to remember…”

    That… isn’t the right statement.
    Do you mean EASY to remember?
    Easy to forget?
    Hard to forget?

    Those all make different statements, but I can’t tell which one you actually mean.

  2. I wonder, a large enough memory to store all the pixels, add a counter chip set to loop through the memory at the right speed, a little bit of discreet component glue and plug it straight into the antenna or component video. Low parts count, provided the clock was accurate and you could adjust the phase, it might be an easier way to make a test pattern, it might even fit into a mint tin.

    1. That’s true for analog NTSC/PAL color TVs, maybe.
      Higher end black/white TVs are held in high regards, though.
      They make nice pieces of furniture or decoration items.
      Also, they’re more reliable and can easily be restored (by comparison).

  3. UK satellite service Sky TV has a test card for 4k-UHD – HDR. The purpose of the test card is to demonstrate the tv panel and the connected Sky Q digibox is set up to view HDR. If the tv panel or Sky Q digibox is not set up properly, the 4k-UHD – HDR test card will not be displayed.

  4. Lovely old test patterns (or test cards in the UK). I liked the tweaks they made to Test Card F when they made Test Card J, like re-centering the image so the X was exactly in the middle of the screen, and you could see that the clown actually had chalk!

    And then there’s Ceefax In Vision, which was how you could pretend you had a Teletext TV if you couldn’t afford one. 😃

  5. There’s more to it than that…it’s the analog part of the TV that needs to be tested as well…there’s a filter, IIRC, that separates the color burst and subcarrier from high frequency dots. Part of the test pattern is how well those high frequency dots are resolved and what the breakpoint is between the dots and the color subcarrier. There’s also a test for color carrier phase accuracy and the time it takes to transition over the greatest subcarrier phase difference…there are two bars in the color test that, if you look closely at where they join, there will be a small gap…that’s the place.

    All the above is from memory, quite a while ago I investigated the test pattern. An old HP or Philips catalog might have the data on what is testable with these generators.

  6. All old hardwares should be open sourced like this! Including MCU firmwares, PAL/GAL/FPGA programming…
    I have for example on ald Sony EV-S9000E Hi8 VTR that is completely unusable only because a small MCU, mainly in charge of power management and LCD display, is dead. It could be replaced by some other one, except that this MCU was with integrated OTP memory, with a specific “firmware” now impossible to find anywhere…
    And being able to restore/clone old hardware is something important both for archive preservation (audio, video, data…) and technology history.

  7. I found at NV back in the early 2000’s that our analog / digital Philips test equipment (VM200?) and a newly designed National Instruments ADC box we built for lab qualification, had a color bar error in excess of 2.2 degrees in almost opposite directions. I had to use a scope (10GHz), Excel, and some convolutions to measure the phase and amplitude, and finally tweaked the analog generator parameters down to 0.3% (best we could do with limited digital knobs). Just another day at the job…

    1. Oh, and the IP block we bought came with a DVD, which we found out, was recorded by a pocket VCR from a printed colorbar, and relabelled by their marketing team as a Calibration disk.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.