Electronic Fuel Injection (EFI)
 
I'm planning to convert my Buick straight-eight from carbs to sequential EFI. Why? First of all, I'm fed up with ancient carbs and their limited tunability. I'm also going to install a Procharger centrifugal supercharger, and EFI is the way to go in a blow-thru application. And I want to precisely control the spark, too.
In addition, there's a lot of fascinating fabrication included, and it involves pioneering, as 
so far I don't know anyone who would have succeeded in making a sequential EFI system work with siamesed intake ports... in a Buick straight-8. There are eight cylinders but only four intake ports.
I know it has been done on a Morris Mini, though, with 4 cylinders and 2 ports.  I don't know how they did it, but this is how I'm going to do it.


The factory specs for
a stock 1946 Roadmaster cam @ .004" valve lift:
Intake Opens:
14º  Before Top Dead Center
Intake Closes:
71º After Bottom Dead Center
14
º+180º+71º = 265º total  

Now remember a full cycle in a four-stroke engine means 2 revolutions = 720 crank degrees. The firing order 1-6-2-5-8-3-7-4 dictates that the cylinders sharing the same intake port (same color), fire 180/540 degress apart from each other. I plotted the cam timing events as seen by intake port  #1, on a circle comprising two crankshaft revolutions instead of one.
valve timingWe can see that even with a stock cam, the intake valves overlap by 265º-180º = 85º.  Let's assume we have port injection, squirting 100% duty cycle (open all time). For the 85º that both valves are open, the cylinders share the fuel squirted into the port. Then for a period of 180º, cylinder #2 will get all the fuel (valve #1 being closed). Then both valves are closed for the next 275º, and all that fuel will remain in the port. But then the valve #1 opens, and will suck up all the fuel waiting in the port, plus all the fuel squirted over the next 180º, when only the #1 valve is open.  This means that cylinder #1 will get 275+180 = 455 degrees fuel, plus half of that overlapping 85 degrees = 455 + 42.5 = 497.5 degrees. While cylinder #2 only gets 180 + 42.5 =  222.5 degrees fuel.

So cylinder #1 steals 69% (497.5
º / 720º) of the fuel (like cylinders 6,8 and 3, respectively). So the engine would actually run on four cylinders or burn a hole in piston 2, 5, 7 or 4 (running lean while getting only 31% of fuel (221.5º / 720º). The more radical the cam, the more intake valve overlap and less difference in mixtures => better but not good

In reality, the injectors are typically selected to operate at 85% max duty cycle. The injector must have a chance to close between openings.
With a batch fire port injection, where an injector fires once per revolution or once per two revolutions, the actual injection timing is random, and varies from one cylinder to another.  With a normal non-siamesed port engine it doesn't matter. Most of the fuel gets squirted on a closed valve (which is nice as it gets a chance to vaporize and atomize in the warm port), and some of it gets squirted on an open valve. Anyway, all cylinders will get an equal amount of fuel.  It's only when the emissions come into play when you get interested in precise timing.
 But for a siamesed-port engine, a batch-fire system would be disastrous, as demonstrated above. Now there are three options:
1) Build a TBI-like system, in which the injectors are placed upstream, just below the throttle blades, squirting to a common plenum. The cylinders will see this as "constant flow", much like a carburetor.  You must build the intake for "wet flow". All the flow paths must be "downhill", as there must not be any places where the fuel might drop off and puddle up.
2) Build a sequential port injection.
3) Build a mixed injection with one set of injectors at the port and another at the plenum.

Let's look at option 2.  In this case, a single injector will serve to both cylinder #1 and #2.
You can't start squirting to cylinder #2 until #1 intake valve has closed. And then you have 180 degrees to squirt until the #2 valve closes. The valves in adjacent cylinders are always 180 degrees apart, no matter what cam specs. This means a maximum 25% duty cycle (180/720) if we had 8 injectors (2 per port). But we decide to have 4 injectors which are serving double time at 50% duty cycle (180/360) Assume you are going to rev the engine to 6000 rpm = 100 revs/sec = 0.01 sec/rev = 10  ms (milliseconds) pulse width at 100% duty cycle. Remember we can only squirt half-revolution = 180 degrees. That means all fuel must be in at 5 ms.  That spells like four 70 lb/hr injectors for a naturally aspirated 320 engine. With 15 psi of boost, we need around 140 lb/hr injectors, even more considering supercharged engines typicallly operate at richer mixtures to control heat and detonation.  You will have a hard time controlling such a huge injector at  low loads like idle and cruise - the pulse width becomes so short. This is when you might want t look at option 3. Make a system with 4 + 4 staged injectors - the port injectors operating at low loads, and the plenum injectors stepping up at high loads (boost). This way you can keep the injector size within reason. Unfortunately the ECU  that I had already purchased , the ViPEC V88, cannot drive 4+4 staged injectors in a sequential mode, but only in batch-fire mode. So I'm stuck with option 2.

Now let's still look at option 2.  The injection start point  for cylinder #2 is the closing point of intake valve #1. The injection start point for cylinder #1 isn't  critical, as long as it's all there before valve #2 opens. But it's convenient to put it 360 degrees apart from cylinder #2, so the injection pulses come at equal intervals, and the injectors get proper off-time. So cylinder #2 will get all of its fuel squirted to a open valve, while cylinder  #1 will get all of the fuel  squirted to a closed valve at low loads, and approximately half  of it to an open valve at max load. 

injector sequenceinjector output












The illustration above shows the cylinders that need precice timing (2, 5, 7, 4), and the corresponding cylinder that is 360 degrees apart. Each injector has it's own Output in the ECU, so 8 cylinders = 8 injector outputs.
The injector outputs (as well as ignition outputs) will fire at 90 degree intervals, based on the firing order. You might be tempted to fool the ECU and give it a false firing order, but if you are going to control direct ignition, too, you must keep the actual firing order.

So, cylinder #2 injection needs to be spot-on, while you want to inject to cylinder #1 at 360 degrees later. That means we're going to use only 4 outputs and double-wire the injectors. Each output can handle two high-impedance injectors with no problem at all. You could wire the Outputs 2 & 7 to Injectors 1 & 4, Outputs 5 & 4 to Injectors 3&4 . However, in the software, the injection point is given as reference to #1 cylinder TDC. So, let's wire Outputs 1 & 8 to Injectors 1 & 4, and Outputs 3 & 6 to Injectors 3 & 2. It means that four outputs remain unconnected.

Enough theory for now, huh?  ;-)
 
[Frontpage]

Copyright © Jyrki Pykäri