--- DIY fuel injection --- possible? (For Scooter)
 

I have a GY6 (scooter) engine I want to convert to fuel injection for fun.

I was thinking, would it be possible to make it REALLY simple by simply using a O2 sensor and making it always run in closed loop? I feel like all you would need would be a simple voltage controller and as the O2 sensor reads leaner the injector would squirt more... ?

Or is that WAY to simplified?

That sounds pretty simple to me, but afraid it wouldn't be very effective. It would probably work on a steady running engine with a constant load and no fluctuations in the weather. You could do it using Mega Squirt, probably cost you around a couple hundred bucks to do it though.

It may be too slow, i.e. here they say a code will throw if the o2 sensor takes 1 second or more to respond at idle:
4x4Wire - Tech: Time to Replace the O2 Sensor on your 3rd Gen Toyota?

if you were idling, cracked the throttle open, and had to wait a whole second for the fuel to catch up, your engine would probably die of leanness.

So you would HAVE TO base the injector on both an O2 sensor AND the throttle?... That gets a little more complicated :)

Yeah, you need something to calculate a basic load map. Most small engines don't need timing advance, etc... so no need for all that jazz, but you at least need a pre and post reading to keep response times, and to have a little bit of protection in case one or the other fails.

TPS and O2 might cut it together, with not much else. In the fuel mapping, assume a constant VE (Small rev range, calculate it yourself.), determine the proper fuel/air ratio for the intake volume, and set it up so the TPS alters the fuel map, and the O2 is used as a secondary reference.

You could probably befriend a tuner geek that can write a simple fuel management system for you.

Yeah I need something VERY basic (i.e. low cost).

I was looking at the engine and since its so small (the piston, the cylinder head, everything!) I don't know how much LAG there would be from just running it off the O2 sensor? I can't think that it would really be THAT much since the piston is so small?

Or maybe I could combine a really small jet (like in a carburated setup) to run parallel with the injector. That way, in extremely fast throttle openings, the engine will still run at a normal air/fuel ratio and not completely go lean...

http://ecomodder.com/forum/attachmen...1&d=1277956160


??

It's a very small, air cooled engine... it's going to be pretty susceptible to leaning out.

You could give it a try with an accelerator jet/pump, but that's going to add some complexity to the system overall, won't it?

Even with the idea that I had, you're still going to be running weird most of the time, because load still changes with throttle angle and situation, and you're not compensating for that with any sort of manifold pressure or airflow readings.

Well I can run a Tee off the injector which already needs a pump in the first place so I wouldnt really need anything else...

I was thinking any change in throttle is going to be directly reflected by the engine running leaner or richer... so the O2 sensor would detect that and tell the injector to act accordingly ... or no?

Maybe im thinking WAY to simple lol

Re: The pump - you'd need a power valve, then, because you don't want your accelerator jet always leaking fuel. More complexity.

Yeah, the injector would reflect changes in air mass. Safest bet would be to nix the accelerator jet, setup the injector control to run rich at all times, and allow the O2 to adjust the mixture back to stoich as necessary. That would protect the engine, but waste fuel for throttle dynamics. It probably wouldn't be so bad, though. Again, we're talking about a very small engine.

Why not build something like this, and tweak it after you get it working? It's easier to ask/answer questions once you've actually identified issues. It's nice to think hypothetically, but that's all it'll ever be unless you force yourself past the obvious objections and "just do it", then fix the problems once they're physically present.

you can get enough sensors off ebay/junkyard for cheap to do a much better job than a carburetor, especially with an o2 sensor. But it's a notable amount of work to get it to a point where it is doing a better job.

lemme peek at the obduino code since it has a similiar problem to solve re: consumption from sensor readings..

Basically you want (going with iat and map instead of maf for availability/ simpler driver setup):
rpm signal
intake air temp
manifold pressure

you also need some constants
displacement
volumetric efficiency
and a specific gas constant

from there you can make a pretty good guess as to how much air mass is going into the engine, and how much fuel to add.

And yah, you are gonna need a microcontroller + injector(s) + pumps/relays + other support circuits

the megasquirt might be overkill, there was another atmega based fuel injection system too, lemme see...
http://ecomodder.com/forum/showthrea...html#post61268 but that does timing and has turbo maps and stuff

so, yah it is doable certainly, but not simple by most measures, if you want something that is better (or even equivalent) than a carburetor.

Here's some details on megasquirt, not simple.
How MegaSquirt^® Works

What if you used a MAF sensor (nichrome wire suspended in the airstream) to get an initial setting, then the O2 sensor tweaks the setting tables over time?

Or if you operated the throttle with a servo motor so the controller knows exactly what to expect?

This actually sounds like a great idea. I could use two injectors, one that's controlled by the throttle (as a basic setting, always calibrated to run a tad bit lean) and then a second one that gets it's info from the O2 sensor (to fine tune)

What I'm trying to get around is a complicated computer thing (like megasquirt)

I read at some point about a DIY solution for small engines that was being called "MicroSquirt"... not sure what ever happened to it.

The problem with controlling an injector by the throttle, is that the throttle isn't necessarily going to match the RPM/Load, hence the reason you should have some sort of flow reference. You can only assume that you have the injector set to run slightly lean, for the RPM that the throttle would normally maintain under a no-load situation (where you'd be setting up the map). As soon as you changed RPM, loaded the engine, etc, or changed throttle, you'd be dynamically changing the fuel curve based on throttle only... while the engine is still loaded or at a different RPM than the throttle can support.

Hopefully, that makes sense to you. You'd still be stuck with a slow responding O2 sensor-backed system that runs worse than a carb'd setup.

I'm trying to be constructive as possible here within my scope of knowledge and thoughts. I don't mean to seem condescending to you, if I'm coming off that way.

Now, if you operated the throttle so that it could only ever be X over 0 (where 0 is "neutral", or the throttle required to maintain RPM under zero load), you could use your latest proposal to build the system effectively, however, RPM changes would be slow and arduous, at best. Sudden changes in load and/or engine speed could cause serious lag problems, and you'd be best to still have an RPM reference so the throttle control could reset itself to 0+X.