Yeah, the sad reality is that most of the aftermarket boost controllers throw away function for flashy lights and such. None of them that I have seen do true closed loop control. They are all based on wastegate duty cycle which to me always seemed stupid. You should be able to set "8 PSI" and always get 8 PSI, assuming the wastegate allows it.

My arrangement with the MAP sensor and PWM circuit is probably the simplest form of closed loop you can do. As the voltage from the MAP sensor rises, the PWM duty cycle goes up and the wastegate opens more. As it falls, the opposite happens. A voltage divider made of a single resistor and pot between the two set the boost level....The more I talk about it now, the more I want to pull that stuff of the shelf and work on the design. I could have a $75 closed-loop EBC. By switching out various resistors based on gear or TPS (use the spare output on the Microtech...) it could be duel, triple, quadruple, etc. stage.

True enough. I also hear that the Haltech control kind of sucks, though I have not used it myself.

This is a very difficult thing to do. Even if you use a solenoid that vents when unpowered, that will not account for the driver circuit failing full on. It's a less common failure for sure, but it does happen. Using a normally "closed" solenoid has the exact opposite failure mode. I guess you could use some kind of crowbar circuit so that if the duty cycle out to the solenoid is ever 100% it switches a big SCR across the output to blow the fuse. Actually this kind of circuit is very effective...

Yep. You can set boost based on throttle position and RPM as well.

 

some info on the MS boost control:
http://megasquirt.sourceforge.net/extra/boost.html

 

this makes me wonder why no major aftermarket has done this. Perhaps there is something you are missing?

 

Because a dial and a power LED don't sell a product. Flashing lights and fancy screens do. I think I may make this one of my winter projects...

 

Aaron,

I've been thinking about this... mainly because I don't want to study for my final tomorrow haha. Anyway, how do you plan to keep the thing from overshooting? It's the same basic problem in any closed loops system--keeping the error band and # of oscillations from the target point as small as possible.

The reason I say this is that lately I've been looking through the instructions for the AEM EMS. I'm going to be helping a friend of mine try to at least get his turbo Integra at least driveable on it. I flipped to the section about their onboard boost control system, and it is indeed a fully closed loop system:

http://forum.aempower.com/downloads/...ide%20V2.0.zip

Jump to page 121. Notice that it uses a P + I control system, like a closed loop O2 sensor (I guess it doesn't use a derivative... or it doesn't mention it at least). See http://en.wikipedia.org/wiki/PID_controller for reference on this.

Would your boost controller use integral control? If I understand this correctly, the AEM uses a standard tuning practice in industrial control mechanisms. It finds the error (deviation from set point), applies a basic multiplier to it (the proportional gain), then calculates an integral of the error over time as another correction factor to decrease the overall error band.

I'm not sure I understand where the self-correction will be in your system... you still have to correct the pulsewidth based on how close the boost gets to its targer.

 

Has anyone looked at the Independent Electronic Boost Controller that AutoSpeed created? http://autospeed.drive.com.au/cms/A_2541/article.html

It can map boost according to injector duty cycle.
It can be built for under $30.
I am going to build one this spring when i buy a TII.

 

Back when I had my Turbo DSM a lot of guys used a fish tank valve as a poor man's boost controller. IIRC, the valve was placed in line with the boost solenoid that would detect an excess of boost and open the WG.
There were a couple of designs that used the same principle but brought the control inside the car. One guy made one that had 3 modes, 2 that you could preset the boost, and the third was stock setting for just regular driving. Do a search in the DSM forums and you can probably find them. I don't know how similar the turbo system on the RX is compared to a DSM, but I would imagine the same principles would apply.

 

If boost overshoots the point set by the voltage divider, PWM time goes way up and the wastegate opens more...

 

ohh you mean make something like this??

i needed a project for one of my microcontrollor classes! so i made a simple design as well. with lcd and key pad entry of boost limit, worked in the lab. been to lazy to test it yet though...
used an hc11 microcontrollor ( was hella over kill), a turbo II boost sensor, and a solinoide off the solinoid rack. then it was like 500 lines of assembly code bleh..... , tested it in lab with a hand pump haha





but hey i got a good grade and it was fun! ill test it one of these days

 

Nice design! A lot more involved then mine. Have you considered releasing it as an open-source project? A lot of people here (including myself) would be all over it...

 

i may if it works, when i ever get aruond to testing it, been so busy with other cars latly haha...

 

very nice!
i like the use of the hc11

i got really annoyed and frustrated in my microcontroller class (that was using the hc12). but looks like it can have some fun real world applications

 

haha first i learned using the intel v25 board, which was quite confusing indeed,

then the second time around we learned on a hc11 board and that was alot easier

i feel the hc11 is probably over kill for this kind of application, i was considering useing a PIC chip to run things ( they are ussually real cheap and compact)

 

yup, i guess you could say overkill, but hey if it's what you know how to do then it works

honestly i have learned more stuff on FPGAs than microcontrollers, so i would probably find it easier using those. imagine using an FPGA for this. i haven't touched assembly coding since that microcontroller class, but HDL code on FPGA i have done a lot of. but THAT would be overkill! haha.

 

so does it work?!?