Microphone BPM Counter

domi

Hi,
i'm experimenting with the basic stamp and tried to implement a BPM counter·via CPU.count function for a Microphone attached on P15.

To get a counter for 1 second i tried to count within a loop 4·times to 250ms· (value 28802) because of a Maximum Integer value of 32,767.

Now i got 2 problems:
first, it's not counting exactly 1 second ( counter a little bit to slow, someone knows how?)
second, it sometimes counts 1 beat as 2 beats(beat time to long?)


Thanks for help [noparse]:)[/noparse]

Peter Verkaik

Description of CPU.count()

· /**
·· * Count the number of rising or falling edge transitions that occur during
·· * a set period of time.
·· * If the pin is an output then it will be changed to be an input and will remain as
·· * an input when the counting is complete.
·· * <p>
·· * To ensure reliable detection the input signal must maintain a value for at least
·· * 8.68us. This means that a square wave must have a period of at least 17.36us and
·· * a frequency less than 57.6kHz.
·· *
·· * @param timeout the amount of time to count transitions over. Measured in
·· *··············· 8.68us units. The value of 32,767 allows a maximum measurement
·· *··············· time of 284.4ms.
·· * @param portPin···· the I/O pin to measure transitions on.
·· * @param edge··· the type of edge to count. Pass true to count rising edges,
·· *··············· or false to count falling edges.
·· * @return the number of rising edges or falling edges that occured on the pin
·· *········ during the timeout period.
·· */

The maximum time to count rising or falling edges is 0.284 seconds.
So you used a loop of 4 times 250 msec.
Counter too slow (so the real time is longer than 4*250msec)
due to processing time of your loop.
But the counts are measured in 1 second, only there are gaps between
the measurements, which introduces the extra time.

I think the error of 1 in beats is also due to those gaps in measurement.
Can't you just measure 1 period of 250msec and then multiply the counts by 4
to get counts per second?

regards peter

domi

would be to inexact

Peter Verkaik

Instead of a for-loop, use 4 consecutive CPU.count() calls.
That probably gives smaller gaps and better timing.
And use static int to save results.
After the calls you can add the static ints.

regards peter

domi

thanks, you also got an idea because of the 1 beat counted as 2 or 3?

regards dominik

Peter Verkaik

You can try using alternatively count rising and falling edges.
Chances are that if a rising count is 1 up, a falling count is 1 down,
but added the error is cancelled out.

static int res1 = CPU.count(...,true); //count rising edges for 0.25sec
static int res2 = CPU.count(...,false); //count falling edges for 0.25sec
static int res3 = CPU.count(...,true); //count rising edges for 0.25sec
static int res4 = CPU.count(...,false); //count falling edges for 0.25sec
int result = res1+res2+res3+res4;


regards peter