Murphy’s Law tells us that if anything can go wrong, then it will – and, as MacGillicuddy's Corollary reminds us, at the worst possible time.
These days I use my first experience of Murphy in action as a “lessons learnt” in the engineering course that I teach. Here’s how I tell what happened to us.
The AWA Research Laboratory group that I was in earned our living by designing and building transmitting installations for radio broadcast stations. On this particular job, there was pressure to keep the tall masts as short as we could. Normally this is done to save money, because the masts are a big part of the cost; but on this occasion there was no choice. The site was on the approach path to the local light aircraft aerodrome – so, no tall masts for us, and we had to install obstruction warning lights (the red lamps) on top as well.
When you have to make masts behave as though they are taller than they really are, there are two ways to do it. These days you’d normally use a “top hat”, which is simply bits of wire that attach to the top of the mast. It’s called capacitative loading, because it works as a tiny capacitor. Obviously you can’t make those wires go upwards, but they still have some effect if they go sideways or even downwards a little bit. But in the early 1970’s, the industry still normally used inductive loading.
This method places an electrical gap in the mast, a bit below half way up, where there is a big fat insulator to separate the top and bottom parts of the mast. Across this gap there is a large coil to carry the current. In electronics terminology this is called an inductor, but the old name for an inductor is a coil, which is exactly what it looks like and how it is made. This coil is big – it’s made out of at least 25mm (1”) copper tube and is at least 300mm (1ft) diameter, so you can see it on the side of the mast from ground level.
The design of these coils calls for lots of experience. You never know exactly how big they need to be to do their job correctly until it’s too late, because you have to make measurements of the mast to work it out, and usually the masts haven’t yet been built when you’re doing the design. So you make them a little bit bigger than needed and you don’t use all of them. The hard bit is the problem of those aircraft warning lights which have to be powered, because the only way to get the electricity up there is to put the power cable inside the tube that the coil is made from. This means that the unused part of the coil has to be shorted out, which wrecks its performance. So the coil has to be only a tiny bit bigger than what you will need, and judging that is where all the experience comes in.
So we built the radio station, put it to air, and it seemed to work. But some months later, we received a gentlemanly “please explain” letter from the network’s chief engineer. He’d been out measuring coverage and the signal was only half as strong as it should have been. Could we please assist him in understanding his measurements?
In the broadcasting business, this is panic material. To put it in perspective, sales managers live and die by 10% shifts in ratings. Go up 10% and you’re a hero with bonuses, go down 10% and you’ll probably have to find another job. And while it isn’t true that losing 10% of signal strength means that 10% less people listen to you (the rationale being that for any given level of signal, 10% less people receive that level of signal) – that’s pretty much how it’s perceived.
So because I already told you about how making that coil too big wrecks it, you already know that’s what had gone wrong. We took a hacksaw to the coil and cut off exactly half of it, and the problem went away.
How could we have got it so wrong? Of course, Murphy was the explanation.
My boss was the experienced designer. He worked out it should be about ten and a half turns, so being careful about these things, he specified it as 12.
He gave it to the project engineer, but only mentioned the number 12. So the project engineer, knowing all about making sure you have enough to adjust it, did a technical drawing with 14 turns.
You can see how this is going - everybody else had enough technical knowledge to know about making sure there were enough turns, but was only told the total number, and hence added a few more. The project manager took it up to 16, and the manufacturer to 22.
And lastly, we didn’t send anyone who knew how this worked to do the installation. I was a fairly new graduate, being dropped in at the deep end to learn the trade. My experienced co-worker had done lots of jobs but all of them were for television, so neither of us knew that adjusting the coil so that more than half its turns were shorted out was a problem.
So that’s Murphy in action – a daisy-chain of events; but there’s another aspect that we can learn from – why didn’t anyone pick up that there was a problem at the time?
Obviously, we had never really worked out how to check that the transmitter was working properly. In engineering, this is called Requirements Engineering – you need to figure out what your product has to do, then design how it’s going to do it, and finally you have to figure out how to test whether it does what it should do. Once you’ve done all that, then you give the results to the customer and ask for your money.
So what we’d messed up was that we’d only measured the signal strength in places where it had to be kept low (in the directions towards other stations using the same channel) and never in the directions where it was supposed to broadcast and where most of the listeners lived.
As someone else in the industry once told me, there are two kinds of contract documents that cause grief – the ones that are too long, and the ones that are too short. What he meant was that the long ones leave you with something that you can’t get quite right but which doesn’t really matter, and the short ones end up in arguments about exactly how much work you were and were not supposed to do for the money. So this was also a too-short contract problem - the customer assumed that the biggest company in the business would know how to make sure that it worked properly. Oops.