So as I lay awake with a bout of insomnia last night I was working out the basic control concept I wanted for my prototype. Simple is the key at this point and should be easily done with a PICaxe microcontroller, and inputs from the Throttle Position Sensor, and a MAP sensor.

Fun with flow charts!

that's a great idea. my only argument would be that, being that megasquirt isn't very useful as an FI tuning tool on these cars (wish it was, but with a subaru the stock ecu can be reflashed, rendering it unnecessary), it doesn't really serve any purpose that couldn't be accomplished much simpler and cheaper with a bespoke atmega board and some code.

i would suggest that we prototype the whole thing on an arduino then move to having bespoke boards printed. that will result in the lowest possible cost, can be completely open source (megasquirt is not, very important detail), and it will allow people with an arduino at home to wire their own circuit and use it as is.

with an open source project it is extremely important that there are no proprietary dependencies. every link in the chain must be something like GPL or MIT licensed.

The one part of this I would alter is the very last about setting the motor speed (pressure).
Cross reference throttle position and engine rpm to find boost target.
This way you have absolute control over what rpm and/or throttle position sets what boost level. The level of control is far greater.

What would be wrong with having a tiny off-the-shelf arduino (or similar)?

http://dangerousprototypes.com/wp-co...duino-W490.jpg

With an SDCard reader, the user could load a .CSV with boost targets

I did consider using boost level and I think that may be a good option down the road, but for my prototype I need a more simplistic system. Since I don't even know what sort of boost it will actually produce I need something more straight forward and throttle position will give a nice liner response starting at zero, and ending at full RPM's at WOT.

I think arduino's are a good option and very capiable, but for my prototype it's really more then I need.

I favor the PICAxe for a few reasons:

1. I'm used to working with them
2. They use BASIC so they are stupid simple to program even for non-programmers like me.
3. They are dirt cheap. I mean like $3 cheap.

What about power source? 24v like the Phantom, or more? I had some old industrial UPS wet cell batteries I was thinking would work, although they are heavy.

I'm sure most people will just go with SLA's. I'm going to try Ultracapacitors both because they are excelent for this sort of high drain setup, and because I've been wanting to play around with them. The down side is they aren't cheap.

Voltage really depends on what motor you decide to use. I'm looking at around 36V but I'm still researching that side of things.

I'm not sure I was clear. What I was trying to suggest is that you could monitor the engine rpm and the throttle position. Using those two values, you reference a table that provides the target boost pressure, or motor rpm if you prefer. The result is the same: you regulate your motor based off two input variables instead of one and have a far more powerful system with minimal extra effort.

mainly just that it costs more than it has to. you're paying for a uart-serial chip that you won't need, and a few other things. with an atmega on a bespoke board we could get the cost down below $10 probably.

i would argue against using a picaxe for the same reason you argue for it: BASIC lol.

i agree with most of that. my main point of contention would be throttling down when the system detects it has reached target. it would be best to just return early and let it loop over, no need for unnecessary motor control (one of the most computationally expensive operations in the system), and it will result in oscillation in all cases.

of course when the system is armed a simple scaling of throttle percentage to motor speed is less than ideal as well, as we're not concerned with motor speed, we're concerned with boost. it should read map, then compare to a boost target referenced from rpm and throttle position. this is a far more flexible design, imho.

Funny, would said thing for 2 option. A PIC uc program in Assembler, way more efficient and simple, but that because I am use to program in assembler, I think best is to go with what you know best.