[Unklejoe]

Hello,

I am working on a project and I figured it would be best to collaborate with as many people as possible. This project is actually for my N55 car, but I figure these forums are more active.

The goal:

Make my own progressive PWM based meth injection controller.

Purpose:

Current meth controllers use different approaches for determining engine load. Some use straight boost (AEM), while others use a much more advanced approach of interpreting sensor values (Aquamist). However, all of these approaches are somewhat "inexact" in my opinion when you consider the goal, which is to inject a certain percentage of the total fuel amount at any given time. The Aquamist is probably the best (and likely more than precise enough for its purpose), but all of these kits still have the issue of installation complexity; I want to avoid tapping a bunch of wires under the hood. The purpose of this project is to create a meth controller that is both very precise and very simple to install. This is also partly driven by my curiosity.

Approach:

In short, I will obtain realtime engine load data from the OBD interface and use this to drive an Aquamist fast acting valve (FAV).

I will use an ELM327 or STN1110 OBD-II interface which is basically a microcontroller+CAN interface that accepts ASCI commands for OBD-II PIDs over an RS232 link. The commands will be sent from a microcontroller (probably an Arduino) which will interpret the response values and map them to a duty cycle which will drive the FAV and turn on the pump.

The PID of interest will probably be MAF g/s, since this is the measure of incoming air mass and should map pretty much directly with the actual fuel load (not 100%, as the car runs different air/fuel ratios depending on load, but very close).

Possible Issues:

Sampling rate. I know this car uses CAN for its OBD interface, so it should be pretty quick, but I have yet to benchmark the maximum sampling rate. I'm guessing a minimum rate of 10Hz is required, since it needs to be able to act pretty quickly during throttle-let-off events, and 10Hz would give it a worst case delay of 100ms plus the response time of the solenoid closing. I'm not too concerned about this though, as I feel like I can probably achieve rates much higher than 10Hz. I also thought about tapping the PTCAN bus to snoop in on the engine load value (assuming this is broadcast), but I don't know any of the message IDs and I doubt the piggy back guys will reveal this information.

Determining the flow curve. I plan to tackle this as follows: do a log of WOT in third gear and capture the maximum MAF g/s value. Then, do some logging of partial load conditions (perhaps right before boost is created) and capture the MAF g/s value. Then, I will linearly map these values to the PWM output. This means I am making two assumptions: 1. twice as much air requires twice as much fuel AND 2. 100% duty cycle flows twice as much as 50% duty cycle on the FAV.

Conflicts with Cobb datalogger. Since there will essentially be two OBD devices on the port, this could cause conflicts. However, since this car uses CAN, it shouldn't be a problem.

Any input is appreciated. What I would really like to know is if there's a more efficient way to query this information via the OBD CAN bus. I know the Cobb can log a bunch of paramaters. Does anyone know if the Cobb is using OBD PIDs or if it's sending some custom CAN message requests?

I plan on making all source code public. I will order the Aquamist HFS-3 or 4 this week regardless, just in case this turns out to be a failure.

EDIT: What would be best would be if one of those piggyback manufacturers would just make a device that does this
I know they already have the infrastructure...
It would really help those with flash tunes, although I can see why that would make them reluctant.

[spxxx]

Post this on *********** forums (BB Forums) - that's where the truly technical discussions occur and will surely get more traction.