Breakthrough: Multiplexing

This was not just a Mercedes-Benz problem, of course. All over the world, luxury car makers started to struggle with the explosion of ‘features’, driven by Moore’s Law (which states that the density of transistors that can be printed on an integrated circuit will double every two years) and the latest, ever-more-powerful (and yet cheaper) microprocessors and their insatiable appetite for more inputs, more outputs – more data.

At the same time car companies started to hire engineers from the burgeoning worlds of computing and telecommunications – remember, this was also the era when the mobile phone went from a brick-with-buttons to something you could just about fit in your pocket. OEMs hired kids like me, with freshly minted degrees in microelectronics and telecoms engineering. Ironically, this new wave of engineers hit on a very old solution – something called multiplexing, or ‘mux’ing’.

So, what is it? Multiplexing is any method that allows engineers to send more than one signal over the same medium. Now ’medium’, as far as we are concerned, is an electrical wire. It could be air, or light, or puffs of smoke, but let’s keep it simple here and just talk about electrickery.

The problem of sending more than one signal – more data, in other words – over one wire was not a new one for 1980s Benz ’bahnstormers. Telecommunications engineers first encountered it in the 1850s – the heyday of the electric telegraph. The first trans-Atlantic cable was completed in 1858. It had seven wire strands, so in theory up to seven people on either side of the Big Pond could send messages such as ‘Am in train STOP Home soon STOP Do we need milk? STOP’ – but seven is not very many, is it? A French engineer by the name of Emile Baudot was the first to come up with the solution, in 1891 – something called time-division-multiplexing, or TDM to its friends.

In short, a TDM system splits the data to be sent into time packets – let’s say you have one second to send ‘your’ message, then we leave a gap of half a second, then I have one second to transmit mine. Baudot came up with fast enough automatic telegraph encoding and decoding machines (think a sort of mechanical Morse-key typewriter) to send the messages fast enough to fit into these one-second time windows. For the first time, one wire could allow multiple people to ‘speak’ together – up to four people per wire, in fact.

To the present day we speak of baud-rate in honour of Emile Baudot. Run a speed test on your Internet connection and it may well refer to ‘baud’ – that’s a nod to our man Emile, long gone but not yet quite forgotten.

From the 1890s to the late 1940s, telecoms engineers built on Baudot’s ideas and cooked up various means of sending more and more telegraph, then telephone, data over the same wire. But it was really Claude Shannon who made the great breakthrough in multiplexing.

Shannon was a genius, but an oddball. Brilliant mathematician, computer scientist, heister of Las Vegas casinos (yes, really), juggler and unicyclist, he’s like a zanier and happier American version of Alan Turing. He’s such a fascinating character that we might have to come back to him in a dedicated Ti piece, but for today what’s important is that he is the originator of Shannon’s Theory.

This theory, formulated in 1948, defines the methods that allow engineers to send very complex digital data over a single wire, theoretically with little or no risk of data corruption. Critically, with enough redundancy – basically, saying the same thing multiple times – we can send multiple safety-critical signals over a single wire. We would later come to call this wire a ‘bus’ – a term that comes out of power engineering, but you can (wrongly) think of it as a bus route. The wire is the road, the ‘buses’ are the various packets of data running around it.

So, fast-forward to the ’80s again. Various technical teams now resurrected these theories from previous generations of engineers, and figured out how to put multiple data streams reliably on one piece of wire in an automotive setting.

Chrysler was first out of the gate. The world’s first multiplexed vehicle was the 1988 Dodge Dynasty – a vehicle so obscure that I had to put several extra 50p coins in the Google-o-meter to learn more about it. Never let it be said that Ti does not go the extra mile in bringing you the deepest of automotive deep cuts, folks. It looks a standard late 1980s ’Murican barge, but looking at the specs – self-levelling suspension, fancy stereo, power everything, memory seats etc – one can understand why Chrysler came up with its own in-house multiplexing protocol: CCD or Chrysler Collision Detection. The name is a bit strange, but sort of makes sense when you consider that the engineers were primarily trying to avoid data collision – bits and bytes crashing into each other on the same data line.

Quick doff of the hat in passing to the 1989 BMW 850i too – the world’s second mux’ed car and one that gets considerably more of the limelight than the Dodge Dynasty.

The Chrysler and BMW systems were the first, but far from the only multiplex solutions. The late ’80s and early ’90s saw a plethora of bus systems emerge, as car companies and their suppliers surfed a wave of creativity. Various bus systems with complicated monikers – VANbus, PCI, UART – emerged, a wealth of standards were published and (mostly) promptly forgotten. Studious chaps with short-sleeved white shirts and multiple pens stuck in their shirt pockets were in high demand. It was a good time for my peeps.

But in 1991 the Mike Tyson of multiplexing stepped into the ring. For Mercedes-Benz launched the mighty W140 S-Class – a seminal car with a ‘world’s first’ feature list longer than a first edition Guinness Book of Records. It was a mechanical and electronic tour de force – and it would not have been possible without multiplexed wiring harnesses. But M-B had not just launched yet another bus – it had joined forces with Bosch, Intel and Philips to cook up the Controller Area Network or CAN bus – to this day, the automotive multiplexing system.

Many technical treatises have been written about the CAN bus, and I strongly recommend them to all you insomniacs out there. But here is a very notion of how it works. CAN is – just like Baudot’s original – a TDM system, but one with a twist. Each ‘node’ – or ECU – on the system sends out a data packet, which contains things like a ‘start message’ flag, the 64-bit message itself (which might be temperature information, engine or road speed, or any other data that might be useful to other ECUs on the system), various ‘checksum’ data that allows the sender and receiver to check that the data has not been corrupted (techniques derived from Shannon’s Theory), and an ‘end message’ sign-off.

But here’s the clever bit – the message set also includes a unique 11-bit identifier field, which tells us which ECU is sending the message – and hence its priority. This allows a process called ‘bus arbitration’ – in short, the ECUs with highest priority get exclusive use of the bus’s bandwidth when required. Let’s take a practical example. In normal times, cruising down the road, the infotainment head unit system might give the crucial information that Mrs Twohig just requested yet more heat to the driver’s seat, while I am once again annoying her by playing Appetite for Destruction too loudly, for the nth time. The bus will happily handle all the information required to ensure Mrs T’s posterior is thermally comfortable while we have a heated argument about unnecessary guitar solos.

Let’s say though – and may the motoring Gods forbid – things take a decided turn for the worse, and the vehicle departs from controlled flight, possibly because the driver is paying more attention to winning the argument about 1990s metal than to the road conditions. Unfortunately, we are now in a half-spin and heading for a solid object. The CAN bus system will now – very reasonably – deprioritise any bus messages about heated seats and music or video choices, and will instead give priority to data coming from stability control, restraints systems and other such high-priority ECUs. The infotainment stuff is unceremoniously shoved off the bus, and all available seating, or bandwidth, is dedicated to data that will hopefully save our lives – vehicle speed, yaw-rate, brake status, torque output: boring but important stuff like that.

It’s this built-in arbitration hierarchy that has made CAN as ubiquitous as it is today. And it truly is – CAN bus is to multiplexing what Hoover used to be to vacuuming up dust. It’s not the only system out there – standards with funky names like MOST, LIN and FlexRay still exist, and increasingly, automotive Ethernet carries lots of high-speed data around cars like Teslas, but CAN is still the Heinz ketchup or Hellmann’s mayonnaise of automotive multiplexing. I have no idea what car you drive, but if it is less than 20 years old, I can confidently guarantee it uses CAN.

Having imposed nearly 2000 words on you under the pretext that multiplexing was a Breakthrough, I admit I have second thoughts. Was it really? I would argue yes, in that without multiplexing, engineers would not have been able to exploit the power of microprocessors, and we would not be enjoying the benefits that we’ve previously presented – remarkably powerful, reliable and frugal modern engines, safety systems like ABS, ESP and airbags, handy connectivity solutions like CarPlay etc. But on the other hand, it’s really more of an enabler than a technology in itself. But hey, your opinion is as good as mine – I’ll let you decide, folks.

But maybe one less obvious impact of multiplexing is its profound impact on our car culture. On the one hand, bus systems finally put the tin hat on the Sunday morning mechanic. Diagnosing electrical faults, or even basic servicing of modern vehicles now requires specialised equipment – the most common being the CANalyzer, a piece of kit beyond the reasonable reach of most home mechanics. No matter how good a wrench you are, you’re not going to be able to reflash your Tesla 3’s Car Computer with a 10mm socket. Even a SnapOn one.

As a result, my own little personal stable has nothing at all that is multiplexed because I kind of like being able to sort everything out myself with a multimeter and a 12V test bulb (well, my Ford Model T does not even have 12V).

On the other hand, multiplexing enabled a whole new generation of tinkerers and tuners to work magic with more modern cars, armed with a laptop and a CAN adaptor. When we ‘chip’ a car these days, we no longer physically replace an EEPROM chip in a plug in the ECU (gawd, those were the days, eh?). We plug in a laptop, hack through the OEMs’ – sometimes pretty notional – security systems and download a few megabytes of software and data to various ECUs – which all courses through the car’s various buses. None of that would be possible without good old Emile Baudot and Claude Shannon.