poslathian
LeafLabs
you can also fire interrupts in software, if thats useful. Say you want a timer interrupt to fire whenever the timer runs out OR when an external event occurs.
We dont have library support for that yet, but its very doable and supported directly by the stm32's NVIC (interrupt controller)
gbulmer
Moderator
hmm, this is a very interesting feature request. The simplest thing I can think would just be to configure your external interrupt to trigger on whatever event pulse you are measuring and then in the interrupt routine, configure the timer.
More precisely, configure the timer at startup, and in the interrupt routine either pause, reset, change the overflow, or whatever more minor operation you need.
Yes, could do that, but the timer hardware has input capture support, and it avoids possible problems of being blocked by a higher priotity interrupt.
Silntknight
Moderator
Silntknight - I am thinking about it, but it's past 1:30am here, so I may need some sleep.
Yea, I get that feeling too. I'm usually up until 2AM.
I think I am just going to mess around with creating a timer based on the intake valve sensor. I'll figure out implementation later, but right now I want to use an interrupt to switch a variable state, like I showed earlier. I'm going to port this over to Java for some testing too.
Would something as simple as i++ take one clock cycle? If so, then I can make an accurate counter based on clock cycles. Because the spark plug fires very quickly, I guess I could use a timer-based interrupt to signal it. Just another idea.
I'll just play around with it for a while and see what I can make of it.
poslathian
LeafLabs
ohh, I think i misunderstood the usage here. If youd like to count the number of ticks in between pulses, for instance to determine the frequency of your sparkplug firings, then INPUT_CAPTURE is definitely what you want. Ill see If I can drum up some sample code for you.
Silntknight
Moderator
Poslathian, my usage is actually to control an engine. Basically, when a sensor is triggered (falling), I want the fuel injector to fire. Then, after a calculated delay, the spark plug should fire. I used an external interrupt to complete this process, but Gbulmer thought it was a very faulty idea, so we've been working on improving the code. I would like to make a PWM that uses the sensor (again, the falling edge) to set the pulse. Then, there is the pulse length, which I have the equation for. Then, after a short delay, the spark plug fires.
Ideally, I'd like to be able to (now that I think of it) adjust the prescaler for a PWM on-the-fly so that it matches the period of the engine. The problem is that I'm not sure if I can use the prescaler to set periods continuously over the domain 15ms<t<200ms. If I can set it in 1ms increments over that domain, then I guess I have a (partial) solution. The other part, about the spark plug, is still pretty difficult, but setting an interrupt based on a timer (TIMER_OUTPUTCOMPARE- I need sample code here too, thanks!) would solve that problem. That's the ideal solution.
More realistically, the solutions I listed might have to come into play. The next best thing would be to use the intake valve sensor to directly generate a PWM pulse (actually, this may be the ideal solution). I would like to take an input that will pulse on and off and make a PWM that is based on the falling edge of this signal. The signal would vary from 5 to 60Hz. Using this solution, I can better time the fuel injector (which is more critical) and still use a timer interrupt for the spark plug (which is a little less critical for timing). Anyway, everything *should* be explained in the previous posts. Hope this helps you understand!
EDIT: I just noticed that there is a uint16 Timer1.setPeriod(val) feature! Some example code here might be nice because that would be (nearly) perfect! Thanks everyone for helping me get this far. This might be the solution!
Silntknight
Moderator
Ideas! First, a qualifier. I am assuming I can adjust uint16 Timer1.setPeriod(val) and Timer1.setCompare1(val) on-the-fly. If that is true, then my code is basically:
`void setup() {
Timer3.setChannelMode(TIMER_OUTCOMPARE);
Timer3.attachCompareInterrupt(sparkPulse);
pinMode(pin, PWM);
}
void loop() {
//...//
Timer3.setPeriod(delta_time); //Precalculated period in us
Timer3.setCompare(?); //I honestly have no idea what to put here- see below
pwmWrite(pin1, pulseDelay);
}
void sparkPulse () {
digitalWrite(pin2, HIGH);
delayMicroseconds(5); //Can it just be switched on and off like that? I think a delay would help subsequent circuitry.
digitalWrite(pin2, LOW);
}
From the example on the Timers page, TimerN.setCompare(1) means 1 full period. I can figure out exactly how many microseconds must elapse from the start of the period before the interrupt must occur. Do I have to make some fractions? If the period is 15000us and I need the spark to go off at 13500us, how do I do that? I'm sure there is a way.
gbulmer
Moderator
Silntknight - here is an attempt to draw the sequence diagram I inferred from your definition of the problem.
Engine Management Sequence Diagram
I hope it makes sense!
I am still unclear about how some of the delays are calculated.
The injector starts injecting as it closes. It injects for an amount of time specified by a stoichiometric calculation (done elsewhere).
and
After a delay that lets the flywheel move to 18 degrees before the very top of the compression stroke, the spark plug fires. That delay is also calculated elsewhere and is based on the fuel injector pulse length.
How do those values get to the running Mape program?
Also, at one point (but I can't find it right now) a potentiometer was mentioned, but I don't see where that plays a role in the problem.
Once it is clear what needs to be done, it should be doable.
Timer4.setPrescaleFactor(22);
Timer4.setChannel1Mode(TIMER_OUTPUTCOMPARE);
Timer4.setOverflow(49053); //Set peroid to 15000us
Timer4.setCompare1(44148); //Trigger at 13500us
Timer4.attachCompare1Interrupt(sparkPulse);
That should do what you need.
EDIT: Prescale factor should have been 22, not 2
Silntknight
Moderator
That was actually a great representation of the control unit.
The delays are calculated in the loop. I use a pressure and temperature sensor to get all the data I need. The rest is (relatively) simple math.
a = (analogRead(15)*2949)/4096+1966; //Servo PWM duty cycle
pwmWrite(14, a); //Servo control from 0 to 90
p = ((analogRead(17)-164)*230)/3850+20; //MAP actual (kPa)
t = (analogRead(19)*500)/4096; //MAT actual (K)
pulseDelay = 1000*(((.79*2*14.01+.21*2*16)*7920*.148*.6)/(.0821*14.9*101.325*20))*(MAP/MAT); //Stoichiometric calculation for fuel injection timing- requires C, D, and F- simplify the number once found (allow for lean mix)
delta_t = (1.121*(timeNew-timeOld))/2; //Adjusted delta_t
sparkDelay = 215.94214021463*((timeNew-timeOld)/2)-pulseDelay+corr; //Delay before spark fires (should the pulse calculations be at the beginning or end?)The final numbers (pulseDelay and sparkDelay) control the delays. pulseDelay sets how long the injector pulse is. sparkDelay sets how long to wait AFTER the the end of the pulseDelay. The sparkDelay code might be modified if I am able to use the preexisting PWM controls (which I think I can do pretty easily at this point). Maple, in fact, is calculating these in it's loop.
The potentiometer is actually the throttle. It controls a servo which then controls the air valve. Then, the temperature and pressure sensors go to work. Pretty elegant, at least in my opinion.
Silntknight
Moderator
That should do what you need.
For that one number, yes. I would need these numbers to vary. I thought of another idea, which involves resetting the counter timer using the intake valve (I'm sure someone already said this). I would set the period to a slightly-larger-than-usual value and use an interrupt to Timer3.setCount(0). If I understand this correctly (and the setCount function works), then I will be able to adjust the period continuously. A PWM is triggered when the count is 0, right? So resetting it in this way should work for me. Anyway, I digress. Crenn, what calculation did you use for Timer4.setCompare1(44148); //Trigger at 13500us? I need to know how 44148 is related to 13500us. Thanks!
gbulmer
Moderator
It is looking good.
If poslathian can find some code to make the timer get triggered by the external sensor events, then it should be rock solid, even with USB and the millis timer interrupts.
You might consider sketching something like that sequence diagram, with added annotation. I'd be tempted to show which timers are used, which pins are triggers and outputs and the delay calculations. Then you'll have a useful way to look at all key parts of the whole system in one picture. But I do like pictures :-)
Silntknight
Moderator
I am at "work" now, but when I get home I'll draw some diagrams and post my draft of the code. There are two ways I am approaching it, but they are very similar and are pretty easy to implement. I don't even have to modify and libraries or core files, which is nice considering that's what I expected to have to do. Ironically, asking for help with this turned into me being prodded (in a good way) until I looked at tools already available and figured out how to use them. That seems to be the only way I've ever actually learned anything. I guess that's a little strange, but again, I digress...
You must log in to post.