PIC16F877A as a frequncy counter

[oodes]

Hi,
I am working on a project and using a PIC16F877A to measure various weather patterns. I need to set up the CCP module in the capture mode. My anemometer is rotating at 1m/s or 1Hz. Problem is i need to use Timer1 , when i prescale my T1clk from the 19.6MHz rate i can get it down to around 600,000 (Fosc/4/8), but with my input frequency being as low as 1Hz i would require a count of 600,000 but my timer is only 16bit and can only count to 65,536
Any help guys?

 

[ericgibbs]

Hi,
I am working on a project and using a PIC16F877A to measure various weather patterns. I need to set up the CCP module in the capture mode. My anemometer is rotating at 1m/s or 1Hz. Problem is i need to use Timer1 , when i prescale my T1clk from the 19.6MHz rate i can get it down to around 600,000 (Fosc/4/8), but with my input frequency being as low as 1Hz i would require a count of 600,000 but my timer is only 16bit and can only count to 65,536
Any help guys?

hi
Set Tmr1 to a known value, say 50,000, and count the Tmr1 roll overs into register, read the reg and Tmr1 for the Total count

EDIT:
If you are not sure how to do that, preload the Tmr1 with 50,000, set the Tmr1 interrupt enable, and increment a register every Tmr1 interrupts on over flow.
Reload the Tmr1 with 50000 and re-enable the Tmr1 interrupt at the interrupt

 

[oodes]

Hey,
Thanks for the response,i understand what ur saying about the rollovers but i think my problem is more than that, you will have to excuse me a little im quite new to the world of PICs. The spec for my anemomter is that 1 revolution = 1 Hz , so this can vary reaslitically up to about 50r/s or 50Hz max , how can i implement this on the CCP as my frequency will constantly be changing which in effect means i would have to be imputted new values each time into my ccpr1l & ccpr1h registers , obviously there is an easier method, am just not too sure on how to implement it. I found a method in "Embedded C programming and the MIcrochip PIC" . Which is determining the number of clock pulses (t1clk input, 614,400 =1.63uS) that would occur within my input frequency which say for argument is 4m/s or 4Hz which would be 0.25s frequncy , that would equate to a value of .25s/1.63uS=153374 needed to be on the ccpr1l&ccpr1h registers which again is not possible. Am i way off the mark here with this problem and am i approaching it from totally the wrong angle?

 

[ericgibbs]

hi,
Have you considered using Tmr1's external clock input as a counter for the anemometer pulses ,and using another Timer to measure a 1sec period.???

 

[Ian Rogers]

I have an anemometer on the market.... It has four pulses per revolution... All I do is set up an interrupt on the INT pin... Wait for 1 second.. count the pulses received.. job done.
If you need to process anything else.. use any timer to give a 1 second (or even half a second ) timeout.

 

[oodes]

Cheers Ian , thats sounds great , ill give it a try

 

[Mr RB]

A good simple system is to use TMR2 to generate an interrupt every 100uS. Then just inc a 16bit variable in the interrupt (count the number of interrupts) between each input pulse.

If you use an 8MHz xtal you run TMR2 at 1:1 prescaler and a PR2 period of 200-1 (which is 200 counts, = 100uS).

That gives you a 16bit result which is a period in actual uS *100, with a resolution to 100uS (ie 10000 = 1 second, 10001 = 1.0001 second, 10002 = 1.0002 seconds etc).