When we left off last time, the turbo was back in, every fixing checked and checked again, and I was standing over the engine bay with that particular mix of optimism and quiet dread that comes just before you turn the key. So I turned it.
Morrison started without complaint. I left him ticking over for a few minutes, idling gently, partly to let everything warm through and partly to be sure the oil and the other fluids were circulating properly before I asked anything of him. So far, so good.
Then the first test. I gave the throttle a gentle push, expecting the revs to gather quickly from around 1,800 rpm (the point where a turbo like Morrison’s typically starts piling on the forced induction in earnest), and expecting, too, to hear something of the turbo’s character: a faint whistle as it spooled up, perhaps a soft whoosh as I lifted off. Instead, nothing. The revs climbed in a straight, flat line, the way a tired engine with no turbo at all would, and not a single turbo-ish noise to be heard. Odd.
Ruling things out
Working alone, with nobody to watch the engine while I worked the throttle, I pressed my phone into service as a second pair of eyes. I propped it up with the camera pointed squarely at the turbo, started the engine, blipped the throttle, shut it down, and reviewed the footage. The actuator, the vacuum-driven unit that swings the vanes and the very part I’d spent the last post adjusting, hadn’t moved at all. Not a flicker.
So began the process of elimination.
Had I simply forgotten to reconnect the vacuum hose? It would not have been the most embarrassing thing I’d ever done, but the engine bay is a tight place and stranger things have slipped my mind. I went underneath and looked. The hose was there, connected, seated properly.
Was it blocked, then? I pulled the hose off at both ends, the actuator and the transducer, and blew through it. Clear as anything, no obstruction.
Had I damaged the actuator itself when I wrestled the turbo back into place? I pushed the vane control arm by hand. It moved through its travel with a firm, reassuring stiffness, exactly as it should. Nothing wrong there either.
Each suspect, questioned and released. Which left one obvious candidate: the transducer, the little electronic valve that meters the vacuum to the actuator on the engine’s orders.
Getting at the transducer
Reaching the transducer is not, it turns out, a five-minute job. To get to it I first had to remove the airbox, and then the headlight, and the headlight in turn was blocked by the light guard, the protective metal cage that sits over the lamp on the bull bar.

A word on the bull bar, for anyone who hasn’t met one. It’s the sturdy steel frame across the front of the van, the sort you’ll see on vehicles built for remote or rough country. Its job is to protect the front of the vehicle, the radiator, the lights and the vital bits behind them, from the things you might meet a long way from a workshop: an animal in the road, a low branch, a rock thrown up off a bad track. The light guards are part of the same idea, little cages bolted over the lamps to stop a flying stone from putting a headlight out. Excellent when you’re in the middle of nowhere, and mildly infuriating when all you want to do is get a headlight out on your own driveway.
So: light guard off, headlight out, airbox out. And there, finally exposed, was the transducer, and a surprise sitting alongside it.
A previous mechanic’s handiwork
The wires running to the transducer were not standard. The original wiring had been cut away, and a length of non-standard wire spliced in to take its place. I traced this replacement wire back to see where it went, and it led me all the way to the ECU.
A quick word on the ECU, since it’s about to become the centre of the story. ECU stands for Engine Control Unit, and it is the engine’s computer, the brain I described last time as sitting at the head of the chain: brain, to valve, to vacuum, to rod, to fins. It’s a sealed box of electronics that reads a host of sensors and decides, many times a second, exactly how the engine should behave, including how much boost the turbo ought to be making and therefore what to tell the transducer.

The ECU itself, out and in hand. Morrison’s brain. Everything the engine does begins here. Note the rows of multi-pin connectors along the top, each wire landing on its own numbered terminal.
I followed the replacement wires to the ECU, and right there beside them I found the cut ends of the two original wires, snipped off and abandoned. Someone, at some point in Morrison’s past, had bypassed the factory wiring and run their own.
That raised an immediate and rather worrying question. Had whoever did this actually connected the replacement wires correctly? Wires don’t simply bolt onto a computer; they terminate in a connector, a plug, and within that plug each wire sits in its own numbered slot so that it lands on the correct terminal of the ECU. Getting a wire seated in the right slot is called pinning it. Pin a wire into the wrong cavity and it will sit there looking perfectly connected while its signal goes nowhere useful at all. So: were these wires pinned to the right pins?
The repinned connector in hand, the spliced-in replacement wires clearly visible (blue and brown). The factory wires were cut and these run in their place, all the way back to the ECU.

Turning the problem round
How on earth do you check that, with no documentation to hand? I scoured the internet for a technical wiring diagram for the ECU, the kind that would tell me which pin ought to carry which signal. I came up empty, even among the diagrams you can usually buy for a not-unreasonable fee. Nothing.
So I turned the problem round. Rather than prove the wiring correct from the diagram inward, I would measure what was actually arriving at the far end and work backward. If the right signal was reaching the transducer, then the wires had to be connected to the right pins and sound along their whole length. The proof would be in the signal.
Here I need to explain what kind of signal we’re looking for, because it isn’t quite what you might picture. The ECU doesn’t control the transducer with a smoothly varying voltage. It uses something called PWM, or pulse-width modulation. Rather than dimming the supply up and down like a household dimmer switch, the ECU switches the full voltage fully on and fully off, very rapidly, many times a second, and varies the proportion of time it spends switched on. That proportion is the duty cycle. Fifty per cent duty cycle is on half the time; seventy-five per cent is on three-quarters of the time. A solenoid, and a multimeter, both respond to the average, so a signal that’s on 75% of the time reads as roughly 75% of the supply voltage. It’s a tidy way to control something precisely using nothing more than a fast on/off switch.
One more thing to hold in mind before the numbers: the voltage in a vehicle is not a fixed quantity. We talk loosely about a “12-volt” system, and a battery sitting at rest does read somewhere around 12.6 volts. But the moment the engine starts, the alternator wakes up and begins charging the battery, and to push charge back in it has to work at a higher pressure, typically around 14.4 volts. So the supply voltage in the van climbs from roughly 12 volts with the ignition on and the engine off, up to around 14.4 volts with the engine running and the alternator at full output. This matters, as you’ll see.
I stripped down the plug that connects to the transducer, put my multimeter across it, set the phone recording once again, and ran the engine. Here is what the video gave me.
With the ignition off, the reading sat at zero. Turn the ignition on, engine still off, and it jumped to about 9 volts. As I cranked and the engine caught, it dipped to 5.5 (the starter and everything else hauling hard on the battery for a moment), then climbed quickly back to 9.5 as it settled into idle, and crept on up to 10.5 over the next five seconds or so as the alternator got into its stride. I revved it to 3,000 rpm and it nudged up a touch more, to 10.83. Then I switched off, and it fell to 9.26 before dropping away to zero.
At first glance those wandering numbers look like the ECU busily doing its job, varying the signal as the engine’s state changes. But watch what happens when you measure each one against the supply voltage at that exact moment:
Ignition on, engine off: 9 volts against a roughly 12-volt battery is 75%.
Idle just after starting, battery recovering: 9.5 against about 12.7 is 75%.
Idle after five seconds, alternator charging properly: 10.5 against about 14.0 is 75%.
At 3,000 rpm, alternator at full chat: 10.83 against about 14.4 is 75%.
Engine just switched off, a little surface charge left on the battery: 9.26 against about 12.3 is 75%.
Every single reading comes out at almost exactly 75%. The numbers had nothing to do with the ECU modulating the turbo. They wandered only because the supply voltage beneath them was wandering, and the signal rode passively along on top of it. The duty cycle, the actual command from the ECU, never budged from 75%. Rock steady.
Two things at once
Now, that steady 75% told me two things, and the second one took a moment to sink in.
The first was the answer I’d gone looking for: a clean, coherent PWM signal was reaching the transducer, and that could only be true if the replacement wires were pinned to the correct terminals at the ECU and were sound from end to end. The previous mechanic’s handiwork, whatever else might be said about it, was electrically fine. The wiring came off the suspect list.
But the second thing was the more interesting, and I’d very nearly walked straight past it. A fixed 75% duty cycle that never moves, no matter what the engine is doing, is not what a healthy ECU produces. A healthy ECU is forever adjusting, nudging the figure up and down as the revs and the load change, chasing the boost it wants moment to moment. A flat, unchanging number is a tell-tale. It’s what an ECU falls back to when it has given up.
Engine computers are built to be cautious. When one detects that something it relies upon has stopped making sense, a sensor reading that’s drifted out of range, or a result that doesn’t add up, it doesn’t keep trying to do clever calculations on bad information. It abandons proper closed-loop control and retreats to a safe, fixed default, a sort of mechanical shrug. You’ll often hear this called limp mode. In effect the ECU was saying: I can’t work out what the turbo should be doing, so I’ll just hold this one safe value and leave it there. That value was 75%, and it would never move while the fault persisted.
So the transducer wasn’t necessarily broken at all. It might simply have been faithfully obeying a stuck command. Before I could blame it, I had to find out why the ECU had downed tools in the first place, and that meant looking at the sensors feeding it.
The sensor the ECU had lost faith in
The most likely culprit was the boost pressure sensor. This is a small sensor that measures the pressure of the air on its way into the engine, the very thing the turbo exists to raise. It sits between the intercooler and the intake manifold (which is the shared gallery that takes the incoming air and distributes it evenly to each of the cylinders; it’s the last stop for the air before it enters the engine.) That sensor is precisely how the ECU knows whether the turbo is actually delivering. If its reading goes sour, the ECU is suddenly blind to the one measurement it most needs, and limp mode is exactly the sort of response you’d expect.
Sitting right beside it was an air temperature sensor, which tells the ECU how hot that incoming air is (and therefore, as we covered on the turbo page, how much oxygen it really contains). Two sensors, side by side, both feeding the same decision. Rather than test each in isolation and risk replacing one only to find the other at fault, I decided to renew them both together. I ordered them up, waited the by-now-customary week for them to arrive, and fitted them.
Then I ran the same test again, multimeter on the transducer plug, phone recording. And this time the picture was quite different. As I worked the throttle, the duty cycle came alive, swinging across the full range from 0 to 100% as the revs rose and fell, exactly the restless, adjusting behaviour of an ECU doing its job properly. The computer had its sight back. Replacing those sensors had cleared whatever fault had blinded it.
But, and there’s always a but in this story, it didn’t last. After a few seconds of proper control the duty cycle gave up and settled back to that same flat 75%. The ECU would start out trying, then quietly retreat to limp mode all over again.
That actually made perfect sense once I thought it through. The ECU now had good sensors and was willing to command boost. But it commands boost and then watches the boost sensor to see it arrive. If it asks for boost and nothing happens, no rise in pressure, no response at all, then as far as the ECU is concerned something downstream is still broken, and back to the safe default it goes. The sensors had been one fault. There was clearly another, further down the chain.
Back to the transducer after all
So now, with the wiring cleared and the sensors replaced, I came back round to the transducer with a much better idea of what I was testing. I connected my vacuum tester to the output of the transducer, the side that feeds the actuator, set the phone recording, and ran the engine.
Nothing. No vacuum at all.
This needs a small clarification, because the transducer doesn’t actually make vacuum. It’s only a valve. The vacuum itself is generated by a pump on the engine and held in a small reservoir nearby, ready to be drawn upon, and the transducer’s job is simply to meter out as much of that stored vacuum as the ECU asks for. So no vacuum at the transducer’s output could mean a faulty transducer, or it could mean there was no vacuum arriving at its input to begin with.
So I ran the same test one step further back, on the output of the reservoir. This time the gauge showed a brisk build-up of vacuum from the moment the engine started, climbing to around 700 mmHg and holding there. The supply was perfect, plenty of vacuum, exactly where it should be. For good measure I checked the short length of pipe between the reservoir and the transducer, and that was sound too.
Which left only one possible conclusion. Good vacuum was arriving at the transducer. A live, willing PWM signal was arriving at the transducer. And still nothing came out the other side. After all the rerouting and ruling-out, the thing I’d suspected at the very start turned out to be guilty after all. The transducer was also broken.
A new part, and a held breath
I ordered a replacement. A week later it arrived. I fitted it, reconnected everything, and, with rather less optimism and rather more dread than the first time around, started the engine, gave it some revs, and…




The queue for the event coaches was already fairly long when we joined it, but far better managed than in the previous two years, when it had been a scramble with no order to it. This time there was a proper queue. The first coach came after about a quarter of an hour and we boarded the fifth, a ten-minute ride to the gates. They opened at ten; we were through at a little after a quarter past. Most people begin at the gates and work inward, so that end was crowded, and we went straight across to the far side and worked back towards the exit. We spent some five hours on the ground.
