Remote Vehicle Starter Help

[Art]

Hi Guys, I have used a 16F877A to control a couple of relays so it can start my car. The sequence to start the car begins when the Unlock button on my remote keyfob is held down for approx two seconds. I have a working demo: The unit turns the ignition on, and then waits for the heater coils (diesel), and then cranks the starter motor for a period of time. Problem is, there is no ideal period to crank the starter since, depending on temperature, etc. the motor will be more difficult to start sometimes. Looks like I need some feedback from the engine to determine when it is running off it's own steam so the starter motor can be disengaged asap. The obvious feedback to use is the tachometer (Less than 100 RPM while cranking - approx 800RPM while idling). The turbo pressure sensor may also be useful (0 - 1.9 Volts while idling - 1.9 - 4.5 Volts while racing). The current program is cycling all the time, even the power LED is a POV display. Good news is while the starter motor is cranking I can stop everything and wait for any PBP commands that delay program execution such as COUNT or ADCIN. I'm thinking the tach signal would be the most reliable indication, so any programming ideas without having to make a full on digital tachometer(which I have never done)? I do still have RBO interrupt pin free in hardware. I'd like to consider all options and have my head clear before I go playing with the ECU in my vehicle. Cheers Art.

[Art]

Well I did press return after my sentences.

[Charles Linquis]

It is my belief that anytime you are controlling machinery of any sort, you need to use interrupts. People can wait, machines can't. It sounds like you aren't using DT-INTs, on a timer which would make pauses unnecessary, and which would allow you to read the RPM (assuming you have access to a signal - like the crankshaft position sensor) simultaneously with everything else. Search the forum for my tachometer example. Since uses a timer to poll pins, you can use any pin at all for your tachometer.
If you have a belt pulley on the engine, you can use that as well. Years ago, I glued two magnets on a pully (using JB Weld - a good epoxy), and read an engine's RPM with
a Hall-effect sensor mounted on a bracket that was about 6mm away.

[amgen]

Some alternators have a tach or "R" output that has different voltages during cranking and running. Some diesel or gas generators use that output to stop cranking.
Don

[Art]

Charles, I did see your post with that code before posting this. Some of it's description seems ambiguous to me, but that's likely my own fault. PRELOAD value for example... I don't know what my shortest tach half cycle is, and then in code comments it says "depends on clock speed", but I don't know what to change if using a 4MHz Xtal. Seems a lot to work out when I only need to measure the time between two pulses. The current program is cycling all the time as said in my first post, or the power LED wouldn't work... as I also said, it's a POV display. I'm not using any pause commands at all. Timing is done by counting program cycles so there is opportunity to multitask. The alternator does sound like a good idea, even if it doesn't have the right output as described, the output of the regulator wouldn't reach it's full voltage until idling. With the right resistor divider, I assume I could produce a logic high or low state on a pin so there would still be no halting program execution.

[Charles Linquis]

I think voltage out of the alternator is probably not an accurate measure. What if the battery is discharged?

PRELOAD is simply a var I load into TMR0 on every interrupt to get the interrupt period to be what I want. If you are running 7000 RPM and your tach signal is a square wave that has one cycle per revolution, then the period is 8.5 mSec, and a half-cycle is 4.25mSec. In order to make certain that you don't miss a pulse, your interrupt rate (period) would need to be 4mSec. At 4Mhz, PRELOAD would be 0xF067 (PRELOADH = 0xF0, PRELOADL = 0x67). Directly from MisterE's PIC MultiCalc.

Measuring the time between two pulses is OK if your signal is not "noisy". But the signals I had to measure were very noisy. Sometimes the time between two pulses was 1.5 mSec, sometimes it was 5mSec. That really screwed up my program. In order to get an accurate reading, I had to average a large number of pulses. Then my "quick and dirty" routine wasn't so quick anymore. And the slower the input signal, the longer my routine took.

Timer interrupts to the rescue. I run an interrupt at 500uSec. Now, I can measure any number of signals simultaneously - on any pin. I count transitions over a time interval - say half a second, and I automatically get the average number of pulses over that time interval. The best part is, it all takes place entirely in the background. I can read a variable anywhere in my program, and it has the latest RPM value. The accuracy is limited only by the sampling time.