Anyone know how fuel injectors work?

[zahc]

I know basically how fuel injection systems work. They have oxygen sensors, mass-airflow sensors, and throttle position sensors and stuff, and a computer to predict and/or use lookup tables to decide how much fuel to inject.

So suppose I wanted to engineer a fuel injection system with off-the-shelf parts. I would need to know how the parts all work.

Normal oxygen sensors are easy. They put out 0.2 volts for lean and 0.8V for rich. No big deal.

Mass airflow sensors depend on the design, but you could always calibrate them. I think I could use an absolute pressure sensor instead. 

Throttle position sensor is just a pot.

But how to injectors themselves work? Do they take a 0-12V analog signal? Are they purely digital and you have to provide them with the appropriate pulse width (that would actually be pretty easy)? How much pressure etc? It would be great if it turns out that there is some super-common Bosch injector out there where you just give it 50PSI and switch it on and off digitally, but I can't find any datasheets.

[dogmush]

They're electrically operated solenoids.

You put 45-55psi of fuel on one side of them.  the computer feeds them 12VDC to open the valve and there's a nozzle to atomize the fuel on the way out.  That's pretty much it.

They're rated in lbs of fuel/(I forget the time actually) So a 20lbs injector, held open, would move pass 20 lbs of fuel in (time that I forget).

They're pretty simple.

ETA: Wiki has a decent animation:
http://upload.wikimedia.org/wikipedia/commons/2/29/Injector3.gif

[MechAg94]

Yeah, my understanding is they were on/off solenoids.  I don't know if they regulate the time the solenoid is open or not. 

That said, don't people sell off the shelf computers that control solenoids and all the other inputs? 

[Hawkmoon]

But how to injectors themselves work? Do they take a 0-12V analog signal? Are they purely digital and you have to provide them with the appropriate pulse width (that would actually be pretty easy)? How much pressure etc? It would be great if it turns out that there is some super-common Bosch injector out there where you just give it 50PSI and switch it on and off digitally, but I can't find any datasheets.

43.2 psi, actually, I believe is the standard.

[Tallpine]

Amazing - my old 1938 Cat worked without any electrical system at all 

[Brad Johnson]

They're rated in lbs of fuel/(I forget the time actually) So a 20lbs injector, held open, would move pass 20 lbs of fuel in (time that I forget).

Per hour.

Brad

[Nick1911]

Yeah, my understanding is they were on/off solenoids.  I don't know if they regulate the time the solenoid is open or not. 

They do.  It's basic PWM.  All digital.  Of course, it must be timed to complete it's cycle in the time the intake valve is open.  Hook an oscilloscope to a running EFI car, and you can see for yourself.  Heck, you could even get some baseline timings for that engine.

That said, don't people sell off the shelf computers that control solenoids and all the other inputs? 

Megasquirt.

[dogmush]

That said, don't people sell off the shelf computers that control solenoids and all the other inputs? 

F.A.S.T.

[drewtam]

There are many different types of injectors, obviously diesel, propane, nat gas and gasoline are different.
Among gasoline, there are the throttle body injectors, port injectors, and direct injection (starting to become popular with OEMs).

I expect you are asking about port injection. My understanding is there are 2 types, "high impedance/resistance" type and "low impedance" type.

I have only seen 2 wire solenoids for automotive injectors (the newest diesel injectors have gone to a 4 wire control). They have many different valve designs and spray patterns.

Among the low impedance, there is either voltage control or current control of the solenoid. These waveforms are built into the ECM drivers. I think the source voltage is 12v, its not ramped to a higher voltage like the spark plugs. The voltage may have a PWM form to control current.

In summary, it is a somewhat complex signal sent at the appropriate time, centered on intake opening. The wave is controlled for rush in and hold current and voltage to maintain solenoid life and to produce quick response times. 

Look here for more info...
http://www.stealth316.com/2-injectortypes.htm
http://www.stealth316.com/2-fuelinjection.htm
http://www.stealth316.com/2-calc-idc.htm

Megasquirt is a low cost, open source, open hardware platform ECM. That support group might have all the info you are looking for. Here is thier injector wave form high level description...
http://www.megamanual.com/v22manual/minj.htm

[geronimotwo]

i don't believe aerly fuel injection even used O2 sensors.  they had a cam or distrubutor sensor for timing and approximated the amount of fuel needed.  i had a '72 volvo that would run lean/rich depending on the state of charge of the battery.

also, i do not believe that modern injectors use 12v.  it is significantly less than that, maybe 5ish?

[bedlamite]

http://www.megasquirt.info/
http://www.megamanual.com/mtabcon.htm

[MechAg94]

F.A.S.T.

Thanks.  When I was finishing up college, I was involved with a project to covert a minivan to propane.  We used propane vapor injectors.  The professor knew a former student who did that sort of programming and was able to reprogram the onboard computer to use the propane injectors.

[RoadKingLarry]

Fuel injection doesn't have to rely on fancy computers to work. The 1957 Corvette had fuel injection and I'd bet it didn't use a computer to make it work.

[dogmush]

Fuel injection doesn't have to rely on fancy computers to work. The 1957 Corvette had fuel injection and I'd bet it didn't use a computer to make it work.

True, I work on Marine Diesels everyday (Most notably 71 series Detroits and KTA series Cummins) that use mechanical injection quite well.  But EFI has been standard in most small injected gas engines long enough if someone asks about "Injectors"  you can pretty much assume a computer of some kind is involved.

[zahc]

I think that airplanes had fuel injection going back to the '40s. They called them 'pressure carburetors' but it was just open-loop mechanical fuel injection.

[drewtam]

Everyone still builds mechanical diesels. We just don't sell them in the US/EU/Japan. Or its for the military.

[AmbulanceDriver]

Heck, my truck is a '96 dodge w/ the cummins diesel, and it's mechanical fuel injection - one of the last years they did that (I think '97 was the last year)

[TommyGunn]

I think that airplanes had fuel injection going back to the '40s. They called them 'pressure carburetors' but it was just open-loop mechanical fuel injection.

The German fighters had 'em -- we were a little late with them.  As a result the Nazi pilots could pull negative Gs while we could not.   Ever see film of American fighters flying wingtip to wingtip, and they each, in turn, roll and dive?  That's so they don't pull any neg Gs.  A carburated plane can't do neg Gs while with fuel injection there's no problem with any Gs.

[zahc]

What I think is interesting, is that the injectors in my dad's old Detroit Diesel Series 60 were driven off the cam, so the cam provided the pressure to force the fuel through the superfine holes in the nozzle, but somehow, the injectors were ALSO electronically controlled. I never quite figured that one out; they must have some way to regulated the dwell time of the valve even though it's mechanically driven off the cam...it's obvious they can change both the fuel dose and the injection timing, so maybe they just have a really long-dwell cam lobe and the ability to switch the output on or off at any point in the downward stroke.

[Tallpine]

What I think is interesting, is that the injectors in my dad's old Detroit Diesel Series 60 were driven off the cam, so the cam provided the pressure to force the fuel through the superfine holes in the nozzle, but somehow, the injectors were ALSO electronically controlled. I never quite figured that one out; they must have some way to regulated the dwell time of the valve even though it's mechanically driven off the cam...it's obvious they can change both the fuel dose and the injection timing, so maybe they just have a really long-dwell cam lobe and the ability to switch the output on or off at any point in the downward stroke.

Like I said before, my old Cat dozer had no electrical parts at all, except for the magneto, on-off switch, and spark plugs on the pony motor.  You quit work when it got dark, unless there was a moon.

The 2-cyl gas pony motor was started (well, sometimes ... :( ) with a hand wound rope with a knot on the end.

There were two mechanical fuel pumps, a lower transfer pump, and the four port injector pump itself.  I never had any problem with any of that except that I had to get one injector nozzle rebuilt.  You just don't want to ever let the main engine run out of fuel 

[230RN]

Ever see film of American fighters flying wingtip to wingtip, and they each, in turn, roll and dive?

Huh!  I always wondered about that "peeling off" thing.  Thanks!

Terry, 230RN

[drewtam]

What I think is interesting, is that the injectors in my dad's old Detroit Diesel Series 60 were driven off the cam, so the cam provided the pressure to force the fuel through the superfine holes in the nozzle, but somehow, the injectors were ALSO electronically controlled. I never quite figured that one out; they must have some way to regulated the dwell time of the valve even though it's mechanically driven off the cam...it's obvious they can change both the fuel dose and the injection timing, so maybe they just have a really long-dwell cam lobe and the ability to switch the output on or off at any point in the downward stroke.

We still use that for everything from unregulated to Tier 4 / 2007 On-highway.

Cat calls it MEUI - Mechanically actuated Electronically controlled Unit Injector.

The cam drives a piston/pump in the injector. The fuel is allowed to dump into the return system until the solenoid closes the exit valve/gate. Once the valve closes, the fuel pressure under the pump (inside the injector) increases quickly because its an in-compressible fluid. The pressure increases quickly and pushes open the injector tip check valve and begins injecting at high pressure. Flow rates and injection pressures are determined by the cam lobe profile + injector nozzle size + check valve setting + injection timing.

This gate valve can open and close quick enough that we can complete multiple injection events in a single compression stroke.

The newest systems have 2 solenoids for other interesting tricks that I can't talk about in detail.

[birdman]

They do.  It's basic PWM.  All digital.  Of course, it must be timed to complete it's cycle in the time the intake valve is open.  Hook an oscilloscope to a running EFI car, and you can see for yourself.  Heck, you could even get some baseline timings for that engine.

Megasquirt.

Not true.  As an entire combustion cycle at high rpm takes a very short amount of time (20ms @ 6000rpm) and the air velocity in the intake tract is 30-60 fps (for a 400hp V-8), the air/fuel mix only moves about 7-14" during the entire cycle, and only about 2-3.5" during the intake stroke  In fact, at peak rpm and full throttle, most fuel injected engines run 80-90% duty cycles (16-18ms out of 20ms total).  Only in DIRECT injection engines does the injection need to occur during a specific phase (compression cycle).

If you hook up a scope, and run at high revs/high load, you will see the PWM over the entire engine cycle is very high, meaning most of the time the injector is open, the intake valve is closed.

[230RN]

Wow, I remember reading about "brush carburetors" where a rotating bristle brush dipped into a pool of fuel and ran against an edge, spraying the fuel into the airstream.  Sort of like running your finger against a wet toothbrush.

This was in a book about maintaining automotive systems which was old enough to contain instructions on how to change out the leather pads on the clutch and maintain the carbide headlamps --pre-1920, I believe, and I think Sears and Roebuck published it.

I loaned the book to someone and he moved away and I never saw it again, which is why I never, ever, ever lend out books any more.

Ever.

Terry, 230RN

ETA:  I found a reference to brush carburetors.  For once my search engine techniques worked:

http://www.ehow.com/info_12180607_wick-carburetor.html

[rcnixon]

Huh!  I always wondered about that "peeling off" thing.  Thanks!

Terry, 230RN

Google "Miss Shilling's orifice" for a perspective on how a carburettor can be made to handle some negative G's.

Also, has anyone else worked with Hilborn mechanical fuel injection?  It was pretty simple, back in the day.  They have become pretty sophisticated now though.

Russ