Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Joseph Davis]

Aaaaand, http://j2534-3.sourceforge.net/

[DSMR]

Thank you for the wealth of knowledge! I have not had a chance to do any further testing because I am finishing working on a DSM (97 with 98 ECU with Ceddy mods). The car came in with plethora of problems so I have been swamped!

I really have no intention of selling the EBC I make. I wanted to do something for the DIY guys for cheap! The only thing that I might sell is a custom PCB for everything but that is optional. I would like this all to be open source. All my code and methods will be posted for all to see. This is the main reason I chose the Arduino controller

I have dabbled in cell phone apps before and can stumble my way through it if I have to. When I get that far I will seek help from people who can actually code! Getting the app to talk with the Arduino is fairly simple to do though!   

[Joseph Davis]

Been through a couple cars on CeddyMods, works nicely.  I have a spare 98 Mirage ECU (same motherboard) I flashed with CeddyMods if you need one I currently have a 2WD dyno so most AWD cars go elsewhere, and the 2WD DSM community are super rare.

[DSMR]

Been through a couple cars on CeddyMods, works nicely.  I have a spare 98 Mirage ECU (same motherboard) I flashed with CeddyMods if you need one I currently have a 2WD dyno so most AWD cars go elsewhere, and the 2WD DSM community are super rare.

I like ceddy mods better than link for some reason. I always thought that link was trying to force your hand to tune it "their" way rather than the way you would tune anything else. Line up your injectors, remove boost cut, check everything and your set. Had it done within 4 or 5 flashes and it is OBD2 compliant. I had to ghetto rig a pin in the ECU for the 97 but luckily no bricked ECU. CEL on knock works great! Closed loop on the car really helped fill all the gaps and the knock sensor safety nets the stock engine pretty well. Car did not make the power that it should but I really think it was bottlenecked somewhere. Car was a 97 GST spyder. Made 276hp/302tq at 20ish psi. Owner wants 300.

Off topic but I wanted to pick your brain on the 02 Mirage LS ECU's. Rumor has it that they can be reflashed with tephra's latest blend. If so, those ECU's are much cheaper than the EVO's. I would love to just switch DSM's over to that for the enhanced logging and real time tuning (not to mention it would be cheaper than link with faster datalogging and better processor). What say you? 

[Joseph Davis]

What cars make and what their owners want are sometimes three different things.


Not a fan of Sliderbarzzzz.  But I don't care about real time tuning - the more I tune the more it presents itself as a crutch for the incompetent to finger jab until the problem goes away.  This often results in a poorer outcome than understanding what tou are doing, measuring, calculating, and making a change off of what the motor actually wants.

[DSMR]

You make a good point about real time tuning. Technology has spoiled us. Back in the day it was moving around values in hex to get the desired result. This meant that you had to have a full understanding of the disassembly of the ROM along with how your changes would affect the car. Now all a tuner has to do is beat on a keyboard while watching a wideband.

Back on topic:

PID loop is in full motion. I have found a wonderful PID library that also includes an app to help tune in your values. The app opens up a window that displays real time values of all your I/o's, setpoint, and the controllers Ki, Kd, and Kp values in real time. I will be making a rig to test this on before implementation.

Mr. Davis has inspired me not to finger jab my tuning of the PID so I will use Ziegler-Nichols method of PID control tuning. It is still experimental but it seems to be a crowd favorite in the control theory world. 

[Joseph Davis]

Link me to PID library plox?

[DSMR]

Link me to PID library plox?

http://www.arduino.cc/playground/Code/PIDLibrary

Is what I will be using. He did a GREAT job on this. Here is his blog: http://brettbeauregard.com/blog/2011/04/improving-the-beginners-pid-introduction/

Here is the "Front End" which is the process I will be using to tune the K parameters with: http://brettbeauregard.com/blog/category/pid/front-end/

As of now I have everything working and just need to sit down and test. MUCH credit needs to be given to Mr. Beaurgard for his work on this! So far by looking at his code, this should rival any industrial PID system.

[92CXyD]

What about this?

http://autospeed.com/cms/title_The-Independent-Electronic-Boost-Control-Part-2/A_2542/article.html

[DSMR]

What about this?

http://autospeed.com/cms/title_The-Independent-Electronic-Boost-Control-Part-2/A_2542/article.html

Thanks for the link. Very interesting how they used IJPW as the control element to control boost. My train of thought on this would be to use a map sensor with actual closed loop to control this though. Are there any arguments you might have that would suggest using injector PW would be a better method to use rather than using target PSI and Ki Kp and Kd? I am always open to suggestions, thanks.   

Edit: Their kit is $190! We can make one for  $$

I plan on this being $70 MAX with all the bells and whistles. So far my parts are into the $50 range.

[DSMR]

Got the PID test code to work:



This is just provided test code that I verified with my Oscope. Looks to be working great! Next is to port all of this into my code for some closed loop goodness.

I am going to build a test rig to test my solenoid using shop air. Should be interesting. Stay tuned.

[nock]

Well count me in. I am a massive ardino fan. You should check out "holset he351 know were its at" over on the forced induction section. Yes I am far to lazy to cut and paste on this fucking iPad. I recommend you use the pid lines from my code for the holset controller. As I recall the library was written for much slower processes like temp or levels in a large tank. The one I used started with a program that was written for servo control and then stripped down even more from there. It should be more then fast enough and use much less overhead.

Also, back in day a little bit, I remember the old ms1 boost controller. It was just the GM solenoid a fet and a map of rpm vs throttle position. The cells in the map were just filled with the solenoid pulse widths. So you had to guess to make it work. it was cool having it based on tps because you could have that control back instead of just spooling up and peeling out all over the place. At the time there was a pid pressure map capability, but most people had trouble making it work on 59000 different engines. Ie not a 350. I think you can still look at the ms1extra code for "inspiration"

The phone thing has me intrigued. Maybe using the accelerometer in the phone as a from of launch control? There are already some apps out there that get a whole lot done by using the GPS/accelerometer. Most of its for road racing but I'm sure there are other possibilities. Gear detection is one that comes to mind, so having a boost by gear set up and using the g-meter/GPS to figure witch gear it's in. If you could make the can interface work you would not even need the gps. If there was a way to somehow work a 2-step, and a nice looking delay box into that (I have now idea how) then it would be something you couldn't go to the drag strip without.

Before i go to far off into game changing app land I should point out that I don't know jack shit about phones, most of my knowledge is limited to electronics and embedded programming. I would like to lend a hand just let me know. My holset project is pretty much dead now. I think I should go update my thread.


Ps. wholly thread jack batman

[DSMR]

Well count me in. I am a massive ardino fan. You should check out "holset he351 know were its at" over on the forced induction section. Yes I am far to lazy to cut and paste on this fucking iPad. I recommend you use the pid lines from my code for the holset controller. As I recall the library was written for much slower processes like temp or levels in a large tank. The one I used started with a program that was written for servo control and then stripped down even more from there. It should be more then fast enough and use much less overhead.

Also, back in day a little bit, I remember the old ms1 boost controller. It was just the GM solenoid a fet and a map of rpm vs throttle position. The cells in the map were just filled with the solenoid pulse widths. So you had to guess to make it work. it was cool having it based on tps because you could have that control back instead of just spooling up and peeling out all over the place. At the time there was a pid pressure map capability, but most people had trouble making it work on 59000 different engines. Ie not a 350. I think you can still look at the ms1extra code for "inspiration"

The phone thing has me intrigued. Maybe using the accelerometer in the phone as a from of launch control? There are already some apps out there that get a whole lot done by using the GPS/accelerometer. Most of its for road racing but I'm sure there are other possibilities. Gear detection is one that comes to mind, so having a boost by gear set up and using the g-meter/GPS to figure witch gear it's in. If you could make the can interface work you would not even need the gps. If there was a way to somehow work a 2-step, and a nice looking delay box into that (I have now idea how) then it would be something you couldn't go to the drag strip without.

Before i go to far off into game changing app land I should point out that I don't know jack shit about phones, most of my knowledge is limited to electronics and embedded programming. I would like to lend a hand just let me know. My holset project is pretty much dead now. I think I should go update my thread.


Ps. wholly thread jack batman

Thanks guy. I read your thread ages ago and it looked like solid stuff! The Closed loop code I am tweaking with looks like it will be sufficient. You can adjust sampling times and whatnot inside the library so I think it will be adequate. This of course will be tested though. You may have hit the nail on the head when it comes to overhead. Eventually I will want to port this over to an uno so people wanting to use this code can use the Arduino to burn the chip and place it into a PCB. The mega is a bit more beefier so it will have to be tested with an Uno. 

If you have an Uno and an Oscilliscope, I would like to test some of the code to get the frequency correct.

When I get a chance I will go into porting all of this code. I have been kind of busy lately so I have not had a chance to piddle.

I went overkill on the thread jack. My next post on this will be in a new thread. haha

[rawr]

Ive got an uno and a gps controler nist traceable frequency standard, bunch of counters and scopes if you wwnt me to check it.

[runsfromdacops]

any chance this thing with do boost by gear based on rpm vs vss just like most of the honda ecus? with 6 gears plz

[Speed Phreak]

Here is the boost controller I use for the JRSC setup on my daily driver ... No really.