THE PROBLEM:

This has been bugging me for the last year, and so I feel like I should do something about it. I've been looking at the options one has today for modifying Subaru engine management, which is essential to getting the most out of your car, and I see some shortcomings:

Piggyback (UTEC, XEDE)
* expensive
* limited functionality and engine management quality

Aftermarket ECUs (TEC, Motec, Link, etc.)
* expensive
* sometimes not as powerful / functional as the stock Subaru ECU for the daily driver

Reflashed stock ECUs (ECUTEK, Cobb)
* expensive, particularly when considering that no new hardware is required beyond the ODBII interface
* limited/no access to user modification of maps/parameters (especially Cobb)


To me, these shortcomings lead to:

* fewer people with modified EM
* tuning and ECU knowledge guarded by a few individuals / companies
* reduced access to information
* a smaller community of people tuning and sharing tuning ideas
* slower and lesser innovation for everyone

THE PROPOSAL

What I'm proposing is coming up with an open platform for modifying code and data on the stock Subaru ECU much as the ECUTEK and Cobb commercial solutions do. The "platform" would consist of low cost hardware (say $100 for a USB version) which interfaces to the ODBII port and free, open source software running on a PC (and perhaps ported to other platforms). Some of the features I'm envisioning are:

* Datalogging to PC at improved rates by modifying Subaru ECU code
* Ability to back up your original ECU code before modification
* Ability to change maps / tune maps on the fly
* Ability to support non-OEM sensors through OEM sensor connections
* Ability to share everything (maps, code changes, etc) with others and form a tuning community

At this point you are probably thinking one of the following

(a) Many of the companies listed above have invested significant time and money in developing their products and technologies - what you are proposing would put them out of business, to the detriment of the Subaru community.

I don't believe so. Cobb for example, has a reputation for conservative, proven tuning solutions which appeal in part to a demographic that wants things to "just work" with minimal risk or tinkering. Using an open source EM solution would be unappealing to these people in a way that is analogous to how some companies might choose a Microsoft or a preconfigured commercial UNIX operating system for a critical server over a homebrewed linux solution. Does linux cut into Microsoft's market share? Of course it does - but it also helps promote innovation in the industry and make powerful software and tools accessible to everyone for free.

(b) That sounds great, but where are we going to get the knowledge to do this, and who is going to do the work?

The only way this will fly is if there are many people involved in the software development, the tuning community, and testing and using the platform. We are already part way there. I've been doing some reverse engineering of flashing / remapping the Subaru ECU over the last couple months, and I already have code written and hardware designed which allows me to reflash 2002 and 2004 WRXs from my PC. There's no glitzy UI or fancy features yet, but it works. What's sorely missing at this point is:

1. exact details about the processors used on the Denso ECUs so the code can be reverse engineered
2. details about how all the maps work exactly (having (1) will help solve this)
3. my very limited tuning knowledge / dyno access to develop new maps
4. lots of code development to make this a powerful and easy to use tool for everyone

(c) Why does everything you say have to be in outline form?

It's just easier that way.

YOUR OPINION?

Tell me what you think of this idea, good or bad...