[Diy_efi] injecting 5 port head

[Adam Wade]

--- Jean Belanger <jean at jeanbelanger.net> wrote:

> I was not forgetting that, I was just simplifying
> things for a general explanation.

It seems you "simplified out" the main problem, unless
I missed something.

> As for making it easier with multiple injectors,
> Rover did not seem to think so when they made the
> MPi Mini.

Unless you have at your disposal an engine dyno with
at least 4-gas analysis and full engineering access to
your ECU, and some long-term professional injection
system designers and programmers, I doubt that Rover's
solution would be "easier" to develop.  I was looking
for the simplest way that could be managed with a
fairly low-tech solution that would be reasonably
cost-effective for a hobbyist, and I think the
solution I offered is one of the easier/cheaper
approaches for a hobbyist with limited funds and
tuning tools.

> You should have a look at the Rover patent because
> they describe 2 modes of operation. One mode is the
> obvious pulse on an open valve and the second is
> a single pulse which starts when the number 2 (or 3)
> cylinder valve is open and finishes when the number
> 1 (or 4) cylinder valve is open covering the
> overlap region. I don't know if the MPi engine uses
> them both but it seems the best way to go to high
> RPM and still have a reasonable size injector
> without worrying about the time needed to open and
> close the injector during the intake overlap.

That's a reasonable assessment.  However, I'd wager
that one could produce more power using a more
"discerning" setup, even at high rpm; unless I'm not
clear on something, it would seem that there would
still be mixture distribution issues with that
technology, albeit of a much lesser nature than at
lower engine speeds.

> I assume that this mode also requires a cam which
> does not have an excessive amount of overlap between
> intake and exhaust otherwise you'd be injecting fuel
> out the exhaust.

I suppose it would depend on what sort of fuel economy
and emissions profile were required by the designer;
if you're not running a cat, unless you're concerned
about emissions or fuel economy, there's no reason to
worry much about unburned fuel passing into the
exhaust.

>> In particular, as I understand it, you cannot
>> inject fuel prior to the closing of the "inside"
>> cylinder's intake valve and have any of that fuel
>> end up in the "outside" cylinder -- for cylinders 2
>> and 3, you are limited to the time between the
>> closing of the #2/3 intake valve and the closing
>> of the #1/4 intake valve, which is a pretty short
>> period, especially at higher rpms.

> This is true for discrete injection pulse per
> cylinder but with the mode described above it is
> no longer true if you time the pulse correctly
> (which has to be done anyway). 

I'd assume that the dynamics of intake flow, and where
it goes when, at what velocity, is a very complicated
thing indeed in siamesed-port intakes like the one
described, and would probably vary radically with
changes in throttle and engine speed.  This is one of
the reasons why I lean toward a solution that pretty
much confines a certain injetor's contribution to a
particular cylinder; it would be far easier to tune
without needing CFD or experimental flow rate and
direction data to insure reasonable mixture
distribution, IMO.

> During the intake overlap, some fuel will go to both
> cylinders but the amount to each should be similar
> from cycle to cycle.

In a steady-state condition, I'd say "more or less". 
In a dynamic condition, I'd be willing to bet the
numbers fluctuate pretty heavily, and if someone were
to try that, I'd suspect their four-gas exhaust data
would support that theory.  I don't have access to
anything like that, but if anyone on list does, I'd be
VERY interested in seeing it.

> So you just need to time the injection pulse to get
> a constant AFR between the cylinder pair.

Which, as I noted, is likely a lot tricker than it
sounds when looking at the entire operating range of
the engine, and which would be either very
time-consuming or very difficult (or both!) to achieve
without pretty solid knowledge of the true dynamics
within the manifold, under a given set of operating
conditions.

>> I think a staged injector system might be in order
>> here; in fact, it's possible that the optimal
>> combination would be a single sat. drive injector
>> for cylinders 2 and 3, and staged dual injectors
>> for cylinders 1 and 4, which would allow a honkin'
>> big injector for higher-load running and a small
>> injector to handle idle control.

> And this is simpler than one injector per port?

In terms of achieving even and correct fueling for
each cylinder individually?  Yes, much simpler indeed.
 With that setup, you can deliver fuel for #2/3 in
much of the before the cylinder's intake valve opens
(and before the #1/4 intake valve opens), and can
deliver enough fuel for cylinder #1/4 ONLY during the
time period where its intake valve is open and #2/3
intake valve is closed.  Once you have the code set up
so that for a given value in a map cell each of the
two different styles of injector setup deliver the
same amount of fuel, it then becomes very easy to test
and tune for best possible power without having to do
any fancy trickery or computations (or measurement
while on the dyno) to keep fueling consistent between
all cylinders.

> It seems quite complex for both hardware and
> software.

I wouldn't think it would be terribly difficult for
software.  You'd be running 2 cylinders staged and 2
cylinders not staged.  If you were concerned about the
code, you could use 8 staged injectors and not have to
worry about code balancing the two different types of
injection; this would cost a little more in hardware,
but would make programming a no-brainer using any
system that can handle staged injection, as long as
there is real estate available for mounting all 8
injectors.

> But for this staging you only need 2 injectors per
> port not the 3 you're suggesting

It sounds like you're relying on your claim above that
it is easily "do-able" to use either a single injector
or a single pair of staged injectors to fuel the
adjacent cylinders through the single port, and I
don't think that is within the abilities and means of
most people on this list.  I suppose it would be
possible with proper resources, but even if it were,
it would likely be a much simpler and quicker
implementation to just insure that there is one
injection event per cylinder delivered by a dedicated
injector (or staged pair), since you don't have to
computer intake flow throughout the operating range,
or test via trial and error and four-gas data to tweak
the injection timing throughout the full operating
range (and then you have to do the computation as well
to determine if the optimal injector opening times
have changed, and to insure they are triggered on
schedule).

> (or is it 5 total?).

6.  I'm not sure how 5 would be workable in any
meaningful sense.

>> Timing the injection on the
>> center pair of cylinders will be critical.

> I agree that timing is critical but it's for the
> outside cylinders.

Sorry about that.  I wrote the reply and had
everything backwards regarding the cylinder pairs,
then went through and corrected them all.  Looks like
I missed that one in my editing.

| Kawasaki Zephyr 615 (Daphne)       Kawasaki Zephyr 550 (Velma)|
| "It was like an emergency ward after a great catastrophe; it  |
|   didn't matter what race or class the victims belonged to.   |
|  They were all given the same miracle drug, which was coffee. |
|   The catastrophe in this case, of course, was that the sun   |
|     had come up again."                    -Kurt Vonnegut     |
| M/C Fuel Inj. Hndbk. @ Amazon.com -  http://tinyurl.com/6o3ze |


 
____________________________________________________________________________________
Cheap talk?
Check out Yahoo! Messenger's low PC-to-Phone call rates.
http://voice.yahoo.com