FUEL INJECTION, Petrol

[keith5700]

Been searching on and off all day to see if anyone has ever fitted fuel injection on a small petrol engine.
I found a few references to it but nothing with any details of how to make, or where to buy, etc.
I'm just researching if it's practical for my 1/4 scale engine, before I start designing some carbs for it.

Anyone have any good info, links? Cheers

 

[milotrain]

If you haven't heard or read the Megasquirt Manual then you should. There is also microsquirt which is good for anything around one to two cylinders. Also has spark control. You would of course have to find/build a 1/4 size fuel injector but the control you get with megasquirt will let you set whatever Pulse Width you need to run the little engine.

http://www.bgsoflex.com/megasquirt.html

 

[Lakc]

http://www.homemodelenginemachinist.com/index.php?topic=8897.0
Around page 3 things began to turn towards fuel injection.

 

[keith5700]

Thanks chaps.
Well, I've decided, rightly or wrongly, that injection is the way to go.

Doesn't look like there's too much on the market at the moment for small engines.
I have installed a few stand-alone injection systems in my time, which is partly why I'm reluctant to go back to carbs.
I made a plenum for a Chevy on one of my cars,to keep within the bonnet line, and think I'll do the 1/4 scale V8 along the same lines, with a twin throttle body at the front.
The easiest ECU I've used is the Holley Commander 950, so this might have to be the backup plan if no-one invents something smaller/cheaper, in the next year or so. I'd really rather not have to use a full sized engine ECU if poss.

Megasquirt is worth a look although I'm not familiar with it at present.

I know I could get the engine to run with some simple carbs, but I want something a bit different.

 

[milotrain]

Megasquirt solves all your problems I think. Microsquirt is especially small and cheap although it's in a retooling stage. You can have a very functional megasquirt system for around $250 and it will control EDIS units, or drive coils itself, and control banked injectors. Also the documentation is fantastically complete.

Went back and read that previous thread on carbs and injection and I'll throw out some info I've got.

Mechanical Fuel Injection:
There are fundamentally two types of mechanical fuel injection; those which meter fuel based on air flow and those which meter fuel based on a fuel cam profile regardless of air flow. Typically the fuel cam profile system is simpler but only used in very high proformace cars (read the late 60's DFV F1 cars) due to the fact that tuning is limited to regrinding a fuel cam. This is great for systems that run WOT all the time, but not so much for anything else. The other system meters fuel by measuring air flow, accomplished by a plate interrupting the intake which gets pushed out of the way and flows more air as it's pushed more out of the way. These systems can be very reliable and easy to tune but they are plumbing nightmares as they rely fundamentally on the pressure and vac of various parts of the intake system.

Electronic Fuel Injection:
Again there are typically two systems for metering fuel; one of them is MAP sensor based and one is MAF sensor based (there is a less used early European system that uses an interrupting plate similar to the above mechanical fuel injection system but this only survived for a little while in the late 70s early 80s). MAP sensors measure the pressure and vac of the intake manifold and meter fuel based on that, while MAF sensors are large O2 sensor wire grids that sit in the path of the intake. MAFs are pretty delicate and expensive so I'd be surprised if they were a good solution for model engines. MAP sensors however are very small, can be plumed to anywhere that "sees" the vac or boost of the intake manifold and are cheap. GMs systems use MAP sensors up to the almost total integration of the MAF sensor in contemporary vehicles.

For the above reasons I think the MAP sensor EFI system is the easiest to implement in a small custom footprint and as luck would have it, that's what Megasquirt uses. One problem is that both Megasquirt and almost all other EFI systems need an O2 sensor on the exhaust to recalculate proper fuel supply in the initial tune. These O2 sensors will likely need to be self heating types because model engines won't be run long enough to properly heat them. Once the initial tune is setup however the O2 sensor can be removed.

 

[LongRat]

There's a small section in the book by Hubert W. Schillings about a mechanical fuel injection system he designed and built. His multi-cylinder engines are generally multiples of 10cc cylinders. While the information is quite sparse, the general method of operation is a gear pump driven straight off the crankshaft, pumping fuel through to injectors on each cylinder with a check valve behind each one. There is an adjustable feedback pipe from the pump outlet back to the tank (I assume like the 'calibrated leak' in the thread previously linked). The injectors are simple, passive nozzles with no valving or any form of control. So the fuel flows constantly, not just when required, and the pump runs in direct proportion to crank speed. The feedback pipe means that the fuel flow is not linear with crank speed, with a greater proportion leaking back to the tank at low RPM than at higher revs. While not a 'clever' solution, he claims 30% more power through greater efficiency. Simply removing the restriction of the carb and therefore reducing pumping losses.

 

[keith5700]

Thanks again.
I did consider coming up with some sort of mechanical injection system, using rpm and throttle positionas the variables. When I mapped the ecu's I've used, which were MAP based, I did notice that throttle position and manifold pressure followed each other almost exactly on the screen, regardless of rpms

Coincedentally someone at work has just ordered a Microsquirt, in kit form, to run his latest engine transplant, so I'm going to see how he gets on with that,and see how big it is.
Need to see if it can actually fire a coilpack without the Edis module too.

If it's all ok then I think this might be the way to go. Hopefully I can persuade him he needs to get a wideband O2 sensor to set it up!

Cheers.

 

[kcmillin]

I am kinda late to the party here, but I find this thread very interesting. This got me thinking, Can a diesel type fuel injection system be used on a gasoline engine? Where there is a pump that acts more like an ignition distributor.

Kel

 

[Lakc]

I am kinda late to the party here, but I find this thread very interesting. This got me thinking, Can a diesel type fuel injection system be used on a gasoline engine? Where there is a pump that acts more like an ignition distributor.

Kel

Yes and no.
Its been tried in the past, but fuel lubrisity becomes a problem for the pump. Gasoline direct injection is starting to appear on some new cars, but I think they are more along the lines of modern diesels with direct injection via common rail. I am sure Bosch came up with some zillion dollar coating to make the pumps and injectors live, as seems to be their business strategy, but I digress.... :-\

 

[LongRat]

Thanks again.
I did consider coming up with some sort of mechanical injection system, using rpm and throttle positionas the variables. When I mapped the ecu's I've used, which were MAP based, I did notice that throttle position and manifold pressure followed each other almost exactly on the screen, regardless of rpms

This is extremely interesting. Might make the setting up of a custom ECU somewhat simpler. I have ideas to build my own PIC-based ECU but I haven't completed my first engine yet so better not run before I can walk.
Please keep us updated on the performance of your friend's Microsquirt system!

 

[mu38&Bg#]

When I mapped the ecu's I've used, which were MAP based, I did notice that throttle position and manifold pressure followed each other almost exactly on the screen, regardless of rpms

Manifold pressure may not change significantly, but mass air flow will change with RPM. Fuel must be proportioned to the air mass the engine breathes, not manifold pressure. The ECU calculates air volume volume flow based on displacement, RPM, MAP, and temperature and starts from there. The Bosch mechanical system did essentially the same.

 

[keith5700]

The ecu's I've used have been map based. I chose this type as the engines were all supercharged.
The main fuel graph has manifold pressure on the y axis and rpm on the x axis. All the injector pulse times are then entered into this graph, so this is the datum setting for the fueling.
There are then other tables which are corrections to this main map, primarily coolant temp, accelerator pump jet, engine cranking enrichment etc.

The main map assumes the engine is fully warmed up and operating at a fairly constant speed.
If the engine is cold then the correction factor will increase all the settings in the main map by a certain percentage, say +50% if stone cold,reducing to +0% when fully warm.

The air fuel mixture was checked using a wideband O2 sensor permenantly plumbed into the exhaust.
It proved to be a very accurate and repeatable system.

I'm sure the MAF ecu's are good too, but I have never used one.

 

[Lakc]

Manifold pressure may not change significantly, but mass air flow will change with RPM. Fuel must be proportioned to the air mass the engine breathes, not manifold pressure. The ECU calculates air volume volume flow based on displacement, RPM, MAP, and temperature and starts from there. The Bosch mechanical system did essentially the same.

Once you have the displacement calculated, map certainly does change with both rpm and load, from which you can derive airflow. Having the map value actually saves you another lookup table, which you need to calculate load as a factor of rpm vs throttle opening on MAF systems. This is because the ideal fuel air mixture changes with load. They are both good systems, just different means to the same end.

I dont want to dissuade anyone from trying to build such a system, but the technical challenges are enormous to scale EFI to model engines. Its been done by YS and OS for the model airplane market, but rather crudely based on rpm, throttle opening, and cylinder temperature. It even appears the OS version is no longer on the market.

If your just looking to make something run, you can use a carburator. If you want something different, a crude EFI system can be made to work. If you want power, reliability, and a system that actually needs a 6x6 or 9x9 fuel map to work effectively, there is a lot of serious engineering to be done first.

 

[mu38&Bg#]

The big question is. What are you going to use for an injector?

 

[keith5700]

Lakc, I realise I could get it to run using just a single carb, but the thing is, my hobby is engineering, and has been all my life. I tend to get interested in something, planes, helicopters, full size cars, telescopes, whatever, and I make something in each interest, then get bored and go on to the next thing.
I can't see me making another engine ever again, so I want to make something which satisfies me, which means it's got to be beautiful, perfect, cutting edge,etc. I'm just glad I came across this forum before I got too far into it because the bar has been set a lot higher than I'd imagined when I started.

It may turn out that injection isn't really a realistic proposition for this sized engine, but all the research and the input from forum members is all part of the fun, for me anyway. It's nice to talk to people who know what I'm talking about too, most people I know in person think I'm nuts!

Dieselpilot, I've got two options floating around my head at the moment. The most ipressive loking way would be to have 8 injectors firing into the manifold. I reckon they could be 6mm dia. and still look scale. I need to see how the full sized ones work and see if they can be scaled down.
If this isn't on then I reckon 2 x injectors hidden inside a plenum chamber could work. I'd still have to make them but it'd be a bit easier making them, say, 10mm dia.

I know it is possible as I've seen this amazing site with this bloke making a 1/4 scale Ferrari GP v10 engine.He's made some working injectors which fit inside a 3mm dia tube. His work makes my stuff look like Lego.
Google 'Ferrari V10 Tipo' and it all comes up.

Not heard from him for a year or so though, so he may have realised he's looking at a 20 year project!

 

[Lakc]

Like I said, I dont want anyone to avoid a challenge. ;D I have given its considerable thought myself. Here is what I think your up against with the injector.

• Voltage doesnt scale: Small windings doing a hard job will probably require some type of capacitive discharge circuit, and likely active cooling.
• Pressures scale with orifice size: To effectively atomize tiny amounts of fuel you will need higher pressures and smaller orifices. Perhaps 100 bar pressure and laser drilled orifices.
• Tough work for an injector: Probably hardened, ground, and 6-7 generations to get the lifecycle above an hour.

 

[keith5700]

Hmmm... I think I need to go and read that Ferrari bloke's write up again.
I remember he used a 4 thou' orifice but can't remember anything about fuel pressure.
There's some clips of the injectors working on youtube.

 

[mu38&Bg#]

That's wild. I've not run into that site. The injector operation looks good. Too bad there are no details of the solenoid. Small scale "fuel injection" exists, but it's still in it's infancy. Walbro is making EFI systems for utility engines and I've seen others. They are quite simple systems as the price point for utility applications demands it. As long as it does better than a carb for emissions it meets the current needs of the market.

 

[Lakc]

He seems to have gotten by with a single .004 orifice and approximately 3 bar of pressure. Best of luck to him, as you said, it could be a 20 year project. I notice he is already becoming concerned with contaminants plugging that little bugger up. Been almost a year since any updates.

 

[milotrain]

We know that banked mechanical fuel injectors work, although not that efficiently (but more efficiently than carbs). We have systems that control high impedance PWM injectors for full sized engines. Is it possible to use the PWM control to bank fire a solenoid on the front end of a very high pressure fuel pump (like the bosch mechanical fuel injection uses) which feeds simple carb type sprayers? Then the electronics and valving don't have to be scaled, just the orifices in the mechanical injectors and feed lines.