CCP question

[jamie_s]

Hi all.

Im working on a tachometer project, that uses an input from a trigger wheel on an engine, to measure rpm.

The pulses from the trigger wheel are fed into an LM1815 IC, which basically is a zero crossing detector and outputs a 50-70uS positive pulse at each zero cross.
The trigger wheel has 36 teeth, therefore 36 pulses per engine revolution.

My picbasic code for 16F628A uses the CCP1, with a 1:1 prescale and counting 16 rising edges, pic is running at 4mhz.

Basically im measuring the time for 16 teeth on the wheel (which is aprox 0.4 revolutions of the engine) and working out the RPM based on this time.

This works fine up until around 4000 rpm, then it gets unstable - as in the results i get from the pic output fluctuate quite randomly.

I've basically worked out that at 4000 rpm there is approx 2400 pulses per second on the ccp pin and at 6000 rpm there is approx 3600 pulses per sec (0.27 msec between them?).

This means at 4000 rpm, the ccp event fires 150 times / sec and at 6000rpm around 225 times / sec (every 4.4msec)

Would the unstability be caused by a limitation of the CCP part of the pic or would it be due to the code taking too long to execute, causing discrepancies in the timer?

below is a snippet of my code, minus the rpm calculations and minus the various outputs that are triggered at various rpm. See the last few lines where the %%%%% are.... i think the execution time of these lines may contribute to the cause?

TRISA = %00100000			' Port A all output but MCLR
TRISB = %00001010			' Port B all output but 1,3 (Si and CCP)

T1CON = %00000001			' TMR1 1:1 prescale , and turn it on
CCP1CON = %00000111			' Enable the CCP1 capture, 16th rising edge



startloop:

IF (capture = 0) Then loop	' Wait here until captured 

period.lowbyte = CCPR1L		' Store the captured value in
period.highbyte = CCPR1H	' period variable

IF overflow = 1 then		' Skip the output if the timer overflowed
capture = 0			' Clear capture flag
overflow = 0			' Clear overflow flag
goto reset
EndIF


***** RPM calcs get done here (i've left them out) *******



SerOut So, N9600, [#rpm,13,10]	
EndIF

capture = 0			'Reset capture flag to wait for next period
period = 0
rpm = 0

Reset: 
IF (capture = 0) Then reset	 	' Wait here until next period captured 
TMR1H = 0				' Clear Timer1 high register   %%%%%	
TMR1L = 0				' Clear Timer1 low register    %%%%%
capture = 0				' Clear capture flag           %%%%%
overflow = 0				' Clear overflow flag          %%%%%

GoTo startloop				' Do it forever

[G8RPI]

Hi Jamie,
There should be no problem with the CCP hardware, it can run up to a few hundred KHz. Your code looks OK too. Have you put 'scope on the output of the LM1815? I'd suspect this first. You could also try connecting a signal generator to the PIC in place of the engine/sensor/LM1815/ combination.
It is possible that you are getting interference from the engine at higher speeds that is not connected to your RPM signal.
You don't say how you have configured the LM1815, this could also be the problem.
Get your circuit running on the bench with a signal generator (another PIC will do the job) first. It's safer than by the engine, and you can hear yourself think!

Robert G8RPI.