How The BBC (Still) Sends Audio To Transmitter Sites

Running a radio station is, on the face of it, a straightforward technical challenge. Build a studio, hook it up to a transmitter, and you’re good to go. But what happens when your station is not a single Rebel Radio-style hilltop installation, but a national chain of transmitter sites fed from a variety of city-based studios? This is the problem facing the BBC with their national UK FM transmitter chain, and since the 1980s it has been fed by a series of NICAM digital data streams. We mentioned back in 2016 how the ageing equipment had been replaced with a modern FPGA-based implementation without any listeners noticing, and now thanks to [Matt Millman], we have a chance to see a teardown of the original 1980s units. The tech is relatively easy to understand from a 2020s perspective, but it still contains a few surprises.

In each studio or transmitter site would have been a 19″ rack containing one of these units — a card frame with a collection of encoder or decoder cards. These are all custom-made by the BBC’s engineering department to a very high standard, and use period parts such as the familiar Z80 microprocessor and some Philips digital audio chips, which followers of high-end consumer audio may recognize. As you’d expect for a mission critical device, many of the functions are duplicated for redundancy, with their outputs compared to give warning of failures.

The surprise comes in the NICAM encoder and decoder — it’s a custom LSI chip made exclusively for the BBC. This indicates the budget available to the national broadcaster, and given that these units have in some cases been working for over 35 years, we’re guessing that the license payers got their money’s worth.

You can read about the original switch-over in 2016, and a little more about NICAM, too.

25 thoughts on “How The BBC (Still) Sends Audio To Transmitter Sites

          1. Not to the moon, maybe, but how about regions close to the coast?
            Salty water/air quickly dissolves all kinds of electronics not using gold.
            Considering that the UK is a rather shabby/worn little island with lots of humidity and no sunshine, a higher-end design wouldn’t have had been totally out of place, maybe. ;)

        1. Golden chips existed, but mainly in the form of ceramic versions.
          Which often were precious and used in prototype designs.

          The former Soviet Union had a soft spot for ceramic/gold chips, too, for example.

          “Royal” Z80s with white ceramic and gold pins existed, too.

          One is even shown on the Wikipedia page for the Z8o.

          https://commons.m.wikimedia.org/wiki/File:Zilog_Z80.jpg#mw-jump-to-license

          Ironically, though, gold isn’t the best conductor, even. Silver is a tad bit better, but it corrodes/oxidates.
          Gratefully, the level of conductivity isn’t so important for computer chips.

          1. Gold would only have been a microns thin plate covering of the pins/socket contacts, so its bulk resistance it not important. Having a non-corroding surface makes the contact resistance lower, especially over time, which could otherwise dominate the bulk resistance.

          2. Soldering gold plated leads isn’t good for long-term reliability. The gold dissolves in the solder to some extent and over time crystalises out forming a series of differing metal junctions which can behave like random network of diodes and appear open circuit at low voltages. Its not a practice permitted in many critical applications.
            Socketed gold leads in gold-flashed contacts not quite as much a problem.
            If you’re troubleshooting an old radio or low voltage audio/analog circuit with gold plated diode or transistor leads then re-doing those joints is the first thing.

    1. Some reasons to socket a component:

      1. Component prone to failure, enable quick field repairs.
      2. Component under ongoing development, enable quick field upgrades.
      3. Component not available at time of PCB manufacture.
      4. Component too valuable.
      5. Component at risk of damage during soldering.

      I’m going to guess that for the custom LSI it’s likely to a combination of 2,3 and 4. For the other socketed non-programmable ICs (such as the Phillips filters and Z80 CPU), it’s likely to be 3,4.

      The value of components is interesting. It was common to socket valuable ICs. There were many more components to fail in older designs. Given the cost and time of debugging and fixing a board, it could be cheaper to simply replace it and swap the valuable ICs in. That makes less sense today as there is so much more integration.

      Why tin and not gold? Because it was the ’80s and gold cost money. Gold plated parts just weren’t as common then. Even today you’ll still commonly find tin socket. A 28P Mill-Max DIP tin socket on Mouser is $1.64. The gold contact version is $3.94. The Arduino R3 uses the tin plated sockets. Gold sockets are typically only used in critical situations such as measurement devices, devices that have life-or-death implications, or those expected to survive harsh environments.

      Given the number of sockets that the BBC used in their designs, the use of tin sockets is not at all surprising to me. Got to be careful with the telly-licence money.

      1. Thank you for responding, I could under having the smaller chip with stick on label socketed, but it escaped me that the main chip was too valuable to solder in.

  1. That’s quite an assortment of DIP chips from not just Philips. Motorola;Texas Instruments; Hitachi.And one I don’t recognize, number MJ140 9044A. Now I’m gonna be up all night figuring out what brand that one is. And I thought I’d never see TTL again. BOY don’t that take ya back!

  2. A lot of chips apparently datecoded in the 1990s for an original 1980s unit… though the NICAM ASIC appears to be ’86, do we suppose they had a huge stockpile made, or maybe recovered and redeployed the ASIC from earlier, more 74ey and maybe failure prone or power hungry boards.

  3. Does anyone know how the DAB transmitters are fed? I ask since in my area (East Suffolk) I hear a short gap during studio switchover (typically a few seconds before the presenter of the next show comes on air) which is not present on FM… oddness.

    1. Don’t know for BBC specifically, but many DAB sites are fed with IP data over fiber from a central multiplexer. Some ‘less professional’ installation just send the muxed stream over the commercial internet. The data is locally re-synchronized using a GNSS timing receiver

  4. The BBC definitely got their money’s worth! I engineered several semi-custom systems for them in the 80s and getting things through their commissioning was no joke. The good side was that they were extremely technically across things, and so “tough but fair” would be a good description. They played a part in a lot of advances in audio, back even to the Blumlein patents, some seminal loudspeaker designs which they licensed out, and I believe that the first digital desks were a BBC-Neve collaboration.

  5. There is a group in the UK that do tours around the country with old BBC outside TV broadcast units. They’ve got two trucks now, one is late 70’s all analogue, and the other is 90’s with VTR’s, title machines, widescreen and stereo. ooooh.
    Often at the steam fairs, well worth a trek to see and give them a donation to keep the stuff running (the generator bill is immense for starters).
    They welcome a good chat with anyone but relish talking shop and really explaining it all.
    It’s quite awesome to see just how complicated it was to do this stuff back in the “early” days.

Leave a Reply to DroneCancel 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.