George Lendich wrote:
Paul,
Leaving it as is so as you know the origin.
A patent has been established but it's free to
use by AC XA builders
Subject: AirSIG: KEFI Redux in Tux - Part 1
I mentioned earlier that some further eXploration on XA FI happened
during the long dark winter; here's a rundown on that for those
interested.
You remember perhaps where KEFI was left, before all the hub-bub about
going to Castle, getting our own building, convincing the IRS we were a
non-profit, blahDblah, that we had built a KEFI prototype that sensed
MAP and RPM, and gave us what's referred to on the automotive EFI side
as a "straight-line fueling curve". This all came about in an attempt to
get a modern equivalent of the mechanical Continental FI system, revered
for it's "simplicity ergo reliability".
All fineNdandy, except we ended up with a couple of notSoNice attributes
along the way: one, it was hard to see how to make an *actively*
redundant KEFI setup, and two, we were still left with that dang
"recirculating fuel back to the tank(s)" plumbing that came to be
expected of any modern EFI system.
I'd like to report on a couple of advancements in the above two areas I
think it's safe to say represent real "breakthru's". As I also mentioned
earlier, the setting for this activity was that Henry wanted to use FI
on his D'Fly, but with the way his tanks are setup, he badly wanted to
avoid having to return excess fuel back to the tank(s). I won't bore you
with all the circuitous twistsNturns these brainstorming sessions took,
but suffice it to say that oddly enough, as a result of combining some
better understanding of the Continental FI system and me compulsively
longing for a truely *active* redundant solution (which wasn't a
showstopper for Henry; he'd have been perfectly satisfied with the level
of reliability that conventional EFI could provide, AS LONG AS he could
avoid the fuel return), we ended up serendipitously with a solution to
BOTH issues.
RETURNLESS FUEL RAILS
First the "returnless fuel distribution" problem. In the last couple
years, both Chrysler and Ford have fielded returnless EFI in some of
their cars, and rumor had it that they did this by PWM-controlling the
in-tank fuel pump over a very WIDE range. What we didn't know was if
there was anything special about the pumps they were using, and of
course since the ECU was controlling the PWM operation internally, we
didn't have much insight into how the control scheme worked on these
cars either.
We were also hoping that we might be able to get enough pressure range
>from the controller to directly use a constant fuel injection scheme,
where we just had nozzles (with fixed orifices) at the end of the
individual cylinder fuel lines, instead of having a pulsed injector like
conventional EFI. This endeavor led us to an interesting discovery,
which we'll discuss in Part 2. But back to the fuel pressure
regulation...
Soooo, we built a PWM-controller and selected a nice external gerotor
pump and started doing some testing. We found we were indeed able to
control the pressure with decent finesse, but that the range of duty
cycle (and ergo pressure range) on the pump didn't look anywheres near
sufficient to control the fueling continuously with nozzles, directly
with the PWM. This was OK, since the first goal WAS to eliminate the
requirement for fuel returns, and it looked like that was definitely
licked. Here's how we finished off that round of eXperiments...
We hired a new Ford vehicle for a day, a small Escort with about the
same displacement as an EA-81, and proceeded to study it with scope and
EGOR AFR-meter, and everything else we could throw at it. We found that
Ford was indeed controlling their pump just as we had already built with
our controller, AND using about the same range of duty cycle on the PWM.
Good so far. The KEY element in their system was a trick sensor that sat
on the fuel rail measuring fuel pressure AND was connected to the intake
manifold also, so that the signal back to the ECU is a differential of
rail pressure AND manifold pressure, which is exactly what you want to
end up with, but here they had it packaged in one nice neat
special-purpose sensor. I had heard of this sensor during my travels in
auto-EFI land, and Henry plunked down the big bucks at the Ford
dealership to buy one for the cause (not actually big $$; the sensor's
around a modest $50). But we still needed the rental car to give us some
hints on how to wire the sensor, and how it worked in situ.
We also noticed that with this kinda electronic pressure regulation
scheme (the conventional return-type fuel rail regulators are pnuematic
and only have one mechanical setpoint), Ford was using their
control-ability of the rail pressure to not JUST keep it at some fixed
differential pressure as would be done in a conventional system, but was
actually raising and lowering the pressure at the injectors to provide
them with a greater range of min2max fuel delivery. Altho interesting
use of the technology, we didn't see how this would be a desired
complexity increase for an XA application, but it DID give us more
confidence that the control of the fuel pressure via PWM could be made
fast and flexible.
OK, so now we have a nice inexpensive sensor and PWM controller, and the
external pump we selected (a Walbro that cost about $100 new) seems to
live perfectly contended on from 40-100% duty cycle, which gives us way
more than enough control range to keep a fixed differential rail
pressure on normal EFI injectors, as demonstrated on the Ford
guinea-pig.
So now Henry is happy, we've eliminated the fuel returns for his D'Fly
system, but I'm still pine'ing for the holy grail of active redundancy.
He's happy if at most only one cylinder is lost with any single failure,
but I'm still yearning for FI nirvana.
How we reached EFI Shangri LaLa land is for Part 2. :) Stay tuned.
Gar
__ AirSIG DropBox: ftp://airftp@interstice.com/pub/DropBox/AirSIG __
The Aircraft Rotary Engine Newsletter. Powered by Linux.
http://home.earthlink.net/~rotaryeng/ http://www.linux.org