FUEL INJECTION (GENERAL & 4-STROKE) - page 1 of 8
All books and articles which are an idiot's guide to the
functioning of carburettors start with a description of how a
venturi (or "choke") works - by lowering the static pressure of
the incoming airstream so as to suck up some fuel. This process
is not efficient, it is known in thermodynamic terminology as
"irreversible" - that is, you cannot get out of the process
anything like the amount of energy which went in. My girlfriend
is irreversible, for example.
In effect, the engine is strangled purely to provide some power
to drag the fuel in. Atomizing the fuel nicely (making a fine
spray) so as to make it evaporate better takes more power, a more
restrictive choke, which adds to the strangling effect. In
addition, the chokes and spray bars play havoc with shock waves
travelling up and down the inlet port, meaning that little
advantage can be taken of their ramming effect.
Worse, when a highly-tuned engine is "off the cam" and is
suffering from blowback, the incoming air gets fuelled three
times - once on the way in, again when spat out, and again when
dragged back in. This accounts for the "stand-off" phenomenon -
the cloud of petrol spray seen by carb bellmouths, totally
messing up the settings.
All of this doom and gloom begs the question, if chokes are so
bad, why not pump the fuel in under high pressure, so as to get
rid of them? The answer is that motorcycle design, even at the
highest racing superstar level, is heavily influenced by stupid
greasy oik stick-in-the-mud has-been old pillock luddite twats
who much preferred life in the days of an old tobacco tin full
of assorted jets, needles, and slides with different cut-aways.
Kenny Roberts once said that carburettors are bound to supply the
most suitable fuel mix for an engine as it then receives the fuel
supply which its suction demands. Many journalists believed this
and dutifully reported it, whereupon he must have gone back to
the bar for a good laugh with the Yamaha engine men.
When I was a development engineer working in the F1 car world, we
also used to have competitions to see who could get a journalist
to print the most ridiculous horseshit. My technique was to look
slightly worried whilst describing some complete nonsense which I
made up as I went along: it was amazing what the average hack
would believe then, and we could read all about it in the next
issue of Furry Dice Monthly.
Another aspect of Mr Roberts' management technique, which he
strangely forgot to mention in his superb book, is that when he
finds something particularly good, he will tell everyone else
that it is complete crap, thereby keeping the advantage to
himself. Expect to see a fuel injected Yamaha GP bike soon.
Page 2 of 8
The story which King Kenny put about is quite obviously false
even now: modern carbs (2- and 4-stroke) have emulsion tubes,
progression jets, power jets, air bleeds, and all sorts of other
complex contrivances to fool the engine and provide anything but
the mixture which it demands.
By far the most blatant form of this meddling with the "natural"
demands of the engine is the accelerator pump, which is actually
a transient form of mechanical fuel injection. If you don't
believe this, try opening the throttle with the engine switched
off - there is no venturi effect, but the fuel still squirts out.
Mechanical fuel injection was used in the car world on racing 4-
stroke engines for decades with moderate success. Typically, the
system would be not unlike an HGV diesel metering unit: a kind of
pump, driven directly by the engine, which therefore delivered
more fuel with increasing revs. The stroke of the pump would be
controlled by a shuttle linked to the throttle system, normally
with some sort of cam which allowed the part throttle fuelling to
be controlled quite well.
These systems never really translated into the bike world as they
only worked well over a narrow rev range and the pickup could be
rather abrupt. This is not a big deal on a machine weighing 500kg
with a pair of 15 inch rear slicks, but is a lot more dicey on a
The Norton-Cosworth twin was one of several bike engines to ditch
a mechanical fuel injection system in favour of a pair of trusty
Amal carbs. What a bag of shite that thing was. Conventional
wisdom has it that poor management caused the death of the Great
British Motorcycle Industry: in fact, that Norton engine proved
that had the (highly esteemed) design engineers of the GBMI been
Japanese, they would have been ritually beheaded by their
shareholders. That even Cosworth couldn't sort that engine out
goes to show just how bad it was in the first place.
I worked in the Development Department at Cosworths for a while
during the pimply stage of extreme youth. In those days I lived
on beer and curry, my face looked like a pepperoni pizza and my
ringpiece like a sliced blood orange. I was even more short-arsed
then than I am now, and I used a Norton-Cosworth engine as a step
to allow me to reach things on the top shelf in the stores - this
is significant, as it is the only known case of one of these
engine units being anything other than a total waste of space.
The mechanical fuel injection system reached its height of
development in the early eighties, the last days of the Cosworth
DFV Formula 1 engine (Lucas mechanical injection) when the dog's
dick road car to have was the Audi Quattro, which had a Bosch
system, again totally mechanical.
Page 3 of 8
It was then that fuel injection really came of age, with the
increasingly popular use of microprocessor-based electronic
control units (ECUs). These allowed for huge flexibility in
controlling the fuel delivery, making the systems much more
adaptable than the older mechanical ones.
There are many concepts like this, which have been on the shelf
as Good Ideas for years, they are just waiting for the technology
to become available so that they can be made to work properly.
An example is the Wankel engine, which has been around as an
unreliable oily shitbox for ages, but has now been made viable
with recent advances in tip seal materials and lubricants. The
irony is that no sooner had it become workable than it was
immediately banished to the outer darkness again by the
environment people - a Wankel will never get through any half-
reasonable emissions test, and will surely be banned for everyone
apart from the military before very long.
I wonder if a certain member of the Great British Motorcycle
Industry has worked that one out yet? Probably not, they still do
not know how to install a carburettor after eighty years of
The coming of the ECU allowed fuel flow to be controlled
precisely by switching electrical shuttle valves (for that is all
that a fuel injector is) for an easily controlled amount of time,
thereby controlling the fuel flow. Most systems have a high-
pressure fuel pump which keeps the supply side of the injector
(known as the fuel rail) under pressure. A small regulator known
as an FPRV (fuel pressure relief valve) dumps excess fuel from
the rail back to the tank to keep the pressure constant, normally
about 3 bar or 50psi, though there are a few 5 bar systems
High pressure pumps are usually electrically powered, and made by
a very rich kraut called Robert Bosch. Some racing engines have
mechanically powered pumps, but these need to be spun over
electrically to pressurize the fuel rail before the engine can be
started, of course. Lucas make a superb one of these.
As a safety feature, on road going vehicles the pump is located
close to the engine to reduce the amount of high-pressure fuel in
the system, and the supply is designed to be cut off by a relay
in the event of an accident. Most cars have a low pressure or
"lift" pump just to get the fuel out of the tank and up to the
Page 4 of 8
The first really good electronic injection systems were made by
Lucas. The ECU frequently controlled ignition timing as well, and
was then known as an Engine Management System, or EMS. At the
time, Lucas were world leaders in EMS technology, a reminder of
the great days not so long ago when an engineer knew that if he
couldn't find what he wanted in Birmingham, it would be a waste
of time looking around the rest of the world.
Some engineers at Lucas showed their warm appreciation of the
progressive and sympathetic management style there by buggering
off with all of the secrets to start their own companies, which
flourished. To this day, outside the group of big multinationals
who make their own systems (Lucas, Bosch, Marelli, Ford, Honda,
etc) there is a plethora of small specialists who mainly
concentrate on racing in a shifty sort of way because much of
their stuff contravenes Lucas-held patents. Lucas seem not to
mind too much as it all goes to help the sales of Lucas
injectors, pumps, FPRVs, etc.
The basic set-up of high-pressure fuel rail, injectors, and EMS
(normally known as the black box or brain) has changed only in
detail since the early Lucas systems. The main area of
improvement has been in the brain itself, with the availability
of more powerful and user friendly programming systems
(software), and faster, more sophisticated chips (known loosely
as "firmware", being neither hardware nor software. Really deep
people, electronics engineers).
The brain contains a series of instructions which governs fuel
delivery and spark timing; these are normally worked out
beforehand on a small computer and then "uploaded" to the ECU,
where they are stored in a chip for use. The first chips, known
as EPROMs (eraseable programmable read-only memory) were a bugger
to use as the only way to alter the program was by pulling the
chip out and putting a new one in. These could take anything up
to half an hour to prepare (the process known as "burning in").
It was then normal to see F1 engineers with little foam-lined
boxes full of chips, each with slightly different programs,
which they could swap with whatever was in the car when required.
My own experience of this system was that no matter how many
chips you had, none of them was quite right, so that meant half
an hour in the back of the transporter to burn another one in.
The other approach (especially with Italian racing teams) was to
stand around and argue passionately about which of the present
chips represented the best compromise, but this normally took
more than half an hour anyway.
Page 5 of 8
The Great Leap Forward was then the EEPROM, or electrically
eraseable PROM. This could be reprogrammed via a cable from a
portable computer, meaning that a new chip program could be
uploaded without any fiddly work in about 30 seconds. The cable
(or comms link in computer jargon) also meant that an ECU could
be controlled directly from the computer keyboard.
This is especially useful in dyno testing, when all engine
functions can be monitored, changed, and recorded on the
computer, and the chip only programmed right at the end of the
session. It is also possible to re-map car ECUs by driving up and
down with a lap-top computer on your, er, lap. I have done this
at 150mph in a racing Jaguar sports car whilst trying to cure a
misfire, but have not tried it on a bike yet - I might be crazy,
but I'm not stupid.
The writing of software for an EMS and the subsequent programming
of the ROM is a very highly skilled job demanding rare talent and
experience: my guess is that only about one per cent of those
people who think that they know how to do it really do, perhaps
no more than 20 in the world at the highest racing level.
What doesn't help is that there are only about five dynamometer
systems in the world good enough to keep up with the latest EMS,
so that tends to mean a lot of very expensive track tests.
Money must then also be thrown at telemetry systems, because most
test drivers and riders who are fast enough to be suitable seem
only to get their IQ levels into double figures when going
downhill with a following wind, and their feedback cannot be
This is a pity, because modern EMS systems offer truly awesome
control over engine functions, more than most engineers can use
at present. As is often the case, adaptability also brings
headaches, because with so many things to play with, so many
things that can be altered at the touch of a button, setups are
often way out and people just go around in ever decreasing
circles until either they disappear up their own arseholes or
they go back to carbs out of sheer frustration. Confucius said
that a man with a clock knows the time, but a man with two is
All modern software for programming ECUs is based on what is
called the "map". A two-dimensional map is the easiest system,
often used for ignitions. A sensor on the engine tells the ECU
what the speed is, and the ECU will select the required spark
timing from that point on its map. Thus, the programmer must
enter a specific timing for every speed. To make things simpler,
the program will require to know the timing required at (say)
250rpm intervals, and the software will join the dots up to
decide what the timing should be between these intervals.
Page 6 of 8
The next step is the 3-dimensional map, where the spark timing
depends on another parameter as well as the rpm, normally the
throttle angle, which the ECU can again find out by means of a
sensor. Therefore, with a speed and throttle sensitive 3-d mapped
system, the EMS is constantly checking these two inputs to find
out what the spark advance should be by looking at the required
point on the map - say 11,000rpm up and 30% open across.
This is a very powerful tool; the advance for best power on full
throttle at any and every particular speed can be found by dyno
test and entered, the part-throttle timing can be advanced to
provide good response and mid-corner pickup, the low-speed low-
power setting can be retarded for easy starting, and a huge
amount of advance wound on for high-speed closed throttle (ie
overrun) conditions to stop popping and banging.
Normally, modern EMS systems have two three-dimensional maps
based on speed and throttle inputs, one map for ignition, the
other for fuel flow.
A typical 3-d fuel map will have high fuel flow for low-speed
wide throttle (starting mixture), smallish flow on half throttle
(to give a lean mixture for good economy and response) and a well
rich mixture for best power on wide open throttle. Entering the
fuel-flow values for every single point on a 3-d map takes ages,
and so normally only a few points are checked on the dyno and the
rest filled in by guesswork.
This is why many racing engines run like ruptured ducks at low
speed; the engineer simply couldn't be bothered to spend time
testing an engine costing L75,000 with a service life of about
400 miles under conditions which the driver would never use. As
maps for engines of the same type normally look pretty similar,
there is usually a standard map which is fine-tuned to each
engine in the important areas. "Important areas" on racing
engines are wide-open throttle in the powerband; for a production
engine "important areas" are idling and low-speed pickup, which
are normally adjustable by screwdriver to keep the luddites
You would think that, with the ability to fix fuelling and spark
advance at any possible combination of speed and throttle at the
touch of a button, all engine management problems would be
solved. The very first EMS systems did that, and were crap. There
is so much more to engine functions than a pair of 3-d maps, and
the surface has only been scratched. Again, the availability of
advanced electronic systems only makes life for the engineer
harder; it is no wonder that many just reach for the tin of jets.
Page 7 of 8
Perhaps Karl Marx had this sort of thing in mind when he wrote in
his Address to the Working Classes "....no improvement of
machinery, no application of science to production, no
contrivance of communication, ......will do away with the
miseries of the industrial masses...". Perhaps not.
The most obvious (but not the most important) problem with the
map system is that there are many more factors to be considered
in the fuelling and spark advance than just engine speed and
throttle position: the temperature and pressure of the intake air
are two factors which leap to mind. If the air is hotter and less
dense, it will need less fuel to provide the correct mixture.
This can be programmed in to be adjusted automatically, in the
same way that a top racing team will put in smaller main jets and
drop the needles a notch or two when going to Kyalami or
expecting a heatwave.
To save the complication of having a 5-d map to account for these
extra inputs, a function is normally provided in the software to
jack the entire fuel map up (for a richer mixture) or down. Of
course, the provision of an inlet air pressure sensor means that
the best possible mixture can be obtained in conjunction with a
ram-air system, which increases inlet pressure with increasing
Other inputs which can be monitored and used to warp and stretch
the maps (for just moving them is too easy, and therefore a bit
passe' these days) are coolant temperature (an engine running hot
cannot stand so much ignition advance), fuel temperature (if it's
hot, it expands, so you need to keep the injector open longer to
get the right amount in), detonation sensors, and exhaust gas
If one of these sensors fails, or a connector in the loom comes
apart, it can pull the maps all over the place, so the ECU is
normally programmed to accept sensor readings over a certain
range only: if a sensor goes out of the range then the ECU
assumes it's knackered and uses a default setting instead. Of
course, all of these sensors and systems have to be powered, and
many modern EMS consume as much as 10 amps from the battery -
which carburettors never did. Oops, starting to get into
miserable old bugger mode.
So now we have maps which will respond to any engine parameter
that anyone can think of; the modern engine has more sensors on
it than you can shake a stick at, surely that is adequate? No.
Page 8 of 8
The trick which took a long time for people to latch on to (and
which some still haven't caught up with) is that it is not just
the value of any particular input at any time which counts, it is
also the rate of change of that value. For example, the fuelling
and ignition advance at, say, 30% throttle and 11,000rpm must be
completely different if the throttle is being snapped open (for
example blipping between down-changes), wound open gradually
(coming out of a corner), or held steady (cresting a rise when
there is no grip to transmit any more power).
Anyone without L2m or so to spend on a state-of-the-art dyno will
end up trying to set these transient parameters by the time-
honoured system of going up the road, changing something a bit,
then going up the road again.
Other little tricks which the smart EMS can undertake are timed
cold startup enrichment (like a choke), the provision of
different ignition advance curves in each gear (less advance in
the low gears makes response worse and therefore improves drive
out of slow corners by making it easier to avoid high-siding and
to keep the front down); different fuelling (richer mixture in
high gears to stave off detonation); a "stutter" system
(artificially low rev-limiter with the clutch pulled in to allow
full throttle starts), etc.
They can also control waste-gates on turbo systems, provide semi-
automatic gearchanges (in the true sense, not just ignition cuts)
and traction control through cutting the sparks to one or more
cylinders or by "fly-by-wire" throttle control.
The list is limited only by the imagination these days, and the
lack of suitably intelligent engineers. It should be emphasized,
however, that the single most important factor governing the pace
of race engine development is....MONEY. Excellent state-of-the-
art EMS black boxes can be had for well under L1,000, and the
injectors, pump, loom, etc will bump the price for a full system
up to about a gorilla and a half.
The catch is that any EMS performs only as well as the man who
programs it, and there are an awful lot of wankers around who
think they know all about doing the job, but really haven't a
clue and will blow your engine sky-high as soon as look at it.
Mind you, some people don't need EMS to blow engines, they can
manage that easily on carbs anyway. The select band of
programmers who are "in the know" charge nearly a third as much
as barristers for their services (gasp). A classic case of supply
and demand, which puts Karl Marx's "from each according to his
abilities, to each according to his needs" right out of the
window. There is no doubt that racing, which exists to sort out
one winner and lots of losers, is the ultimate capitalist system,
and the technology involved is merely a reflection of that.