Fuel Injection: Putting theory into practice….
Having initially planned to use Megasquirt (http://www.megamanual.com) as the ECU for my project, I had a last minute rethink about using a DIY ECU for my engine, and swapped to a commercial product from DTAfast (http://www.dtafast.co.uk), the S40pro ECU.
The megasquirt system had been best proven controlling the ignition side of the engine via a Ford EDIS module. The EDIS module is standalone from the ECU except for a pair of low voltage signals that allow the ECU to set the desired ignition advance (module defaults to 10 degrees), and inform the ECU of each ignition event. The module simply requires a crank driven toothed wheel with a corresponding VR sensor to function, connects to a Ford coil pack, and uses the wasted spark approach (fires pairs of spark plugs on both the exhaust and compression strokes).
The S40pro eliminates the need for an EDIS module, taking input signals directly from the VR sensor reading the crank driven wheel, and generating a pair of coil-pack outputs that directly drive a coil pack (again, it is wasted spark); the same Ford coil pack is compatible.
However, the S40pro (unlike the megasquirt) does not support a fuel-only mode of operation, where ignition operations are still performed by a distributor. This forced me into performing the full conversion of both fuel and ignition over to the ECU in a single step, where I had initially planned to get fuel working and then add ignition as an enhancement once fuel was fully debugged.
With the toothed wheel now a critical component, I sourced a generic blank from http://www.trigger-wheels.com as there was no off-the-shelf solution for my Type IV engine (several options exist for a Type I). I chose the mount the trigger wheel behind the crank shaft pulley as my exhaust header has minimal clearance to the front of it. An initial trial using a pillar drill to spin the trigger wheel and an oscilloscope to look at the waveform from the VR sensor verified that the sensor could be at an angle to the wheel without affecting the timing signal that it generated.
I sent a spare pulley and the trigger wheel with instructions on how they should be modified to a local machine shop who did an excellent job of removing the thickness of the trigger wheel from the rear of the pulley and replicating the original mounting hub of the pulley on the trigger wheel itself.
I fabricated a mount for the VR sensor out of two pieces of aluminum, using hts-2000 rods (sold by Frost: http://www.frost.co.uk/) to braze them together. That stuff is amazing, although trying to add a third piece to two already formed pieces tends to result in everything falling apart as the initial joint melts too. The fabricated mount bolts to the engine case at two existing stud locations.
With the trigger wheel taken care of, I could start the conversion in earnest.
A pair of Jenvey TF 40mm throttle bodies replaced the pair of IDF44 carburetors, bolting down directly onto the orignal manifolds. Injectors (I used Bosch 0280150756 30lb/hr 14ohm) mount directly into the throttle bodies. I used an air filter arrangement from RAT (http://www.aircooledtechnology.com) in which K&N air filters are mounted directly to the individual inlet air horns - the Jenvey air filter stud pattern does not match that of IDF carburetors so my original arrangement couldn't be reused. This also meant I had to change throttle linkage as the existing CB performance throttle linkage had bolted to the IDF air filter bases.
I opted to switch to a centre-pull linkage, again from RAT, which bolted up to the alternator stand. I had to move the point at which the linkage connects to the lever arm on the throttle body considerably inboard to obtain full throttle at the throttle bodies with full throttle at the pedal (you can see original hole and first attempted alternative, which wasn't enough, in the picture below). I tapped the hole to allow me to screw the linkage arm directly into the lever to avoid having a fixing nut behind the arm – there just wasn't room.
As mentioned in a previous installment, for better cold running I planned to fit an ECU controlled idle valve to let more air into the engine when it was cold. Having two throttle bodies made this complicated, necessitating a special kit from Jenvey that replaced the air bypass adjuster on each cylinder with a fitting to take hoses to connect all four cylinders to a single distribution manifold. This in turn took a 19mm hose to connect to the idle valve. For the valve itself I had a choice of Bosch and Ford units, depending on the amount of extra airflow ultimately needed by the engine when cold (Bosch unit flows more air; both are PWM controlled). The manifold and valve were fitted to the rear bulkhead. The MAP sensor for the ECU was connected at this manifold also, after a brass damper valve designed to average the resulting signal.
The air temperature sensor was mounted next to an air filter, with the coolant temperature sensor connected to an oil line rather than next to a cylinder head as the temperature range sensed by the S40pro was limited. I took the opportunity to also fit a thermostat in the cars oil lines as my oil was taking a long time to get up to temperature without one.
I opted to mount the Ford coil pack (a late zetec type) in front of the engine. Magnecor, for an extremely reasonable charge, made up custom HT leads to connect the late zetec coil to the spark plugs, complete with shroud air seals. An important note is that most ECUs assume 1&4 and 3&2 cylinders fire together (1 3 4 2 firing order) whereas our VWs fire 1&3 and 4&2 (1 4 3 2 order). So leads 3 and 4 must be swapped between coil and spark plugs on our engines.
With the location of everything decided I could make the engine loom. This was designed to be separate from the main car wiring loom. I sourced the specialized connectors from various sources (including DTA, eBay, Webcon, Jenvey) and the more generic components from http://vehiclewiringproducts.co.uk . Overall the loom took about 20 hours to make but does, in opinion, look as good as one from a factory.
The loom connects to my existing wideband EGO sensor (Innovate LC-1). A wideband EGO is absolutely essential for home tuning of EFI as it makes mapping – the hard part – easy. Define a target AFR map in the ECU and the ECU will generate a correction table that will give those AFR when applied to the current ECU map.
With loom, sensors, throttle bodies, ECU, and new linkage fitted, together with the fuel system modifications listed back in part 1, the physical conversion is complete. Next time I'll cover the initial programming of the ECU, pre-start checks, and first drive!