DIY Fuel injection thread.

[Yowzer]

MAP sensors are a dime a dozen from pickapart though.

[0R10N]

Ahh, of course. Got a 4A-GE one stashed away somewhere.

 

For some reason I had this idea in my head that you could only use a Link approved and calibrated MAP sensor.

[CamB]

Try and pick one that is in the list in the software. Saves aggravation.

I think the reason they pulled the map sensor was that not everyone wants a built in one especially when you only have 3 inputs (and the internal used one). Eg if using Alpha-N or an existing airflow meter.

It won't have knock control if not listed.

[CamB]

Double post

[Roman]

A 4age map sensor wont have much head room for boost, Toyota NA map sensors read atmo pressure at just under or over 4 volts from memory.

But yeah something from a turbo car at pickapart will do the trick.

Or just use what's in the car already if there's one there.

 

Saves a bit of a ball ache if you can find a common one that people know the calibration data for already.
That's where the link ones are nice, but you sure do pay a premium.


I wouldnt be too fussed about knock detection, I ended up turning mine off because it picked up too many other vibrations as a reason to pull timing.

 

Example - tune the knock thresholds in 3rd or 4th and its all fine, but then in 2nd gear it'll think there's shitloads of knock, when really it's just a screwdriver in my sump or my broken engine mounts. lel

[mjrstar]

I have an atom, they do have onboard map, its running sequential ignition and injection.

Has enough to run a 4cyl turbo no dramas. I have 1 spare analogue and one spare digital input. Was thinking maybe a switch or onboard logging.

You will need external boost control though.

[CamB]

I think the G4 Atom had internal MAP and the G4+ has external. So if buying second-hand it might have one.

[downtrail]

Yeah the older atoms have on board. Honestly it will be fine for what you need if you are still running dizzy you can use the other 3 ign channels as outputs so there are a few spare. Even if you went waste spark you'll have 2 spares.

[0R10N]

Sounds like the G4 Atom is the way to go then (Not G4+.) We'll see how things go, this is realistically at least a year away if not longer and I don't see a huge flood of those ECUs entering the 2nd hand market.
 
Currently running wasted spark with a MX-5 CAS on the G1 so no point going backwards there.

[Ghostchips]

So i was wondering about building a fuel injection/electronic ignition system if there was a way to connect sensors to USB cables?

https://thepihut.com/collections/new-products/products/explorer-phat

 

This has 4 analogue inputs and 4 outputs which i figure could be hooked up to some kind of giant MOSFET thingys that trigger the fuel injectors and it has other inputs & stuff that i don't really know how to use.  Might not need to use USB cables with this thing.

Pros : fully custom made for your car

Cons : have to program your own injection program with timing for volume and stuff.

 

Or am i way off track here?

[kpr]

big part of  what makes a usable/decent ecu  is the software/firmware .  so even if you found some hardware that could be adapted to  run an engine.   sofware is going to be the problem

 

far easier to work some overtime,  buy a link

[Roman]

I got my injector test rig up and running.

 

Interesting results from an initial muck around. Running at 12v and ~71psi

Method:

Run the injector for 6000 events at 6ms.

Then run it for 12000 events at 3ms

 

Then 24,000 events at 1.5ms spray, etc etc rinse repeat.

If the injector flowed 100% perfectly with no deadtime, it would fill the container to the same level each time.
But it doesnt because of injector dead time and non linearity at low openings.

The only thing I could find to measure with is a rainfall gauge, so my units are in mm rainfall per square meter.

Dont judge me



Looks like the injector is probably reasonably linear until down to ~3.9ms, so I'll get some more data points around this region

so then doing some conversions and blah blah, the line on this graph, where it crosses to zero shows you what the deadtime is for the linear flow section of the injector (large openings)



So just under 1ms deadtime, the value currently in my ECU was 0.79ms so 20% inaccurate.

What is interesting though is that my injector wont flow anything at all when running at 1.5ms
2ms is fine though, so at some point in between is where it just stops flowing at all.
So again, currently the settings in the ECU currently are way wrong.
Currently if my car tried to idle or crank at 1.5ms, the ECU would try to run at this setting and it wouldnt get any fuel.

My short pulse width adder table (which is supposed to compensate for this) is prepopulated with some values but they're junk too.

I promise I will buy a decent graduated cylinder so I can get some accurate measurements in some better units than mm rainfall hahaha.

There's a big difference though that's easy to see though, at 9ms the tube fills it right to the top, at 2m/s it will only be a quarter full.

Looking through some logs my car idles on the primary injectors at somewhere just under or over 2ms, so it's right in the shit spot. No wonder it's been hard to get the fuel right there.
I reckon once I pick a fuel pressure that I want to run, and run this same procedure at varying voltages. The autotune, temp compensations, idle, closed loop etc etc will all work way better.

[0R10N]

How much is everyone paying for their LSU4.2 or LSU4.9 wideband O₂ sensors, and where from?

 

I found the LSU4.2 on RockAuto, genuine Bosch part and everything, and even with shipping added it's cheaper than sourcing it from a place like Scarles or NZEFI - but surely there are other local options?

[Lith]

I only use 4.9... last one I bought was last year from NZ Performance and at the time the price was very sharp relatively speaking and I often find if I order that kind of thing from them it arrives within 24 hours. I haven't checked current pricing though

[Lith]

I only use 4.9... last one I bought was last year from NZ Performance and at the time the price was very sharp relatively speaking and I often find if I order that kind of thing from them it arrives within 24 hours. I haven't checked current pricing though

[Spencer]

You can get the 4.9 for around $90 US from my looking, maybe $70 for the other. The wideband Bosch knock sensors are also cheap on rock auto at ~$20 US.

I really want EFI on the buick, but I want millions of inputs & outputs which means spendy ECU and I don't want to stump up the cash just yet. Just got a new data logging setup and knock sensor so once that shows up I'll get into the process of carb tuning with some IRL feedback instead of just looking at the WB gauge.

[0R10N]

Hmm sweet. My LC-2 only has the 4.2 cable, looks like I have to buy a converter harness to run 4.9, but either/either is fine by me.

 

The 4.2 sensor I found on RA is US$58.79, to fit a VW Golf/Jetta.

[Spencer]

Buy it. Grab the $20 knock sensor also.

[Roman]

After spending a butt load of time testing I've got some good deadtime and short pulsewidth data in the ECU now.

 

Because my deadtimes for the outer injectors were so wrong, to get it working previously I had specified the inner/outer injector ratio as something like 3.6:1.

 

However after testing I know it's now 2.36:1. I've also got minimum pulsewdiths specified for both sets of injectors.

At 60psi the 1000cc injectors show as 1099cc, the 365cc Altezza injectors show as 460cc or something like that.
 

I wasnt expecting it to, but after getting the fuel back right again my idle is now rock solid! It used to hunt a bit and sometimes stall for reasons unknown.

 

For interests sake I've redone my fuel table using Mixture map only, and now after 3-4 iterations the whole thing is within .1 of it's target AFR. Awesome.

Previously I'd have to manually check the logs and adjust, for most areas.
 

After taking the car for a drive when cold this morning, it also seems to drive really nicely.

 

The engine feels a little gruntier, but maybe that's just a consequence of having driven just the Echo for the last few weeks haha.

I have had some Yokohama AD08Rs fitted today, looking forward to making some doort noises at Hampton on Monday!
Fingers crossed for some good lap times, finally on some non melty tires.

[Spencer]

Nice work man, actually measuring things always wins.