Skip to content

Instantly share code, notes, and snippets.

@tdicola

tdicola/main.py Secret

Created July 1, 2017 00:16
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save tdicola/e12bb4b82291ff436fb2b7dde35f5b4b to your computer and use it in GitHub Desktop.
Save tdicola/e12bb4b82291ff436fb2b7dde35f5b4b to your computer and use it in GitHub Desktop.
Circuit Playground Express and CircuitPython fidget spinner tachometer code
# Adafruit Circuit Playground Express Fidget Spinner Tachometer
#
# This code uses the light sensor built in to Circuit Playground Express
# to detect the speed (in revolutions per second) of a fidget spinner.
# Save this code as main.py on a Circuit Playground Express board running
# CircuitPython (see https://github.com/adafruit/circuitpython). You will
# also need to load neopixel.mpy onto the board's filesystem (from
# https://github.com/adafruit/Adafruit_CircuitPython_NeoPixel/releases).
#
# When the first three NeoPixels light up white you're ready to read the speed # of a spinner. Hold a spinning fidget spinner very close to (but not touching)
# the light sensor (look for the eye graphic on the board, it's right
# below the three lit NeoPixels) and look at the serial terminal from the board
# at 115200 baud to see the speed of the spinner printed. This works best
# holding the spinner perpendicular to the sensor, like:
# ||
# || <- Spinner
# ||
# ________ <- Circuit Playground
#
# Author: Tony DiCola
# License: MIT License (https://en.wikipedia.org/wiki/MIT_License)
import array
import board
import analogio
import time
import neopixel
# Configuration:
SPINNER_ARMS = 3 # Number of arms on the fidget spinner.
# This is used to calculate the true
# revolutions per second of the spinner
# as one full revolution of the spinner
# will actually see this number of cycles
# pass by the light sensor. Set this to
# the value 1 to ignore this calculation
# and just see the raw cycles / second.
SAMPLE_DEPTH = 256 # How many samples to take when measuring
# the spinner speed. The larger this value
# the more memory that will be consumed but
# the slower a spinner speed that can be
# detected (larger sample depths mean longer
# period waves can be detected). You're
# limited by the amount of memory on the
# board for this value.
TARGET_SAMPLE_RATE_HZ = 150 # Target sample rate for sampling the light
# sensor. This in combination with the sample
# depth above controls how slow and fast of
# a signal you can detect. Note that the
# sample rate can only go so high before it's
# too fast for the Python interpreted code
# to run. If that happens the board will
# light up LEDs red to indicate the 'underflow'
# condition (drop the sample rate down to
# a lower value and try again).
# A value of 150 times a second means you can
# measure a spinner going up to 25 revolutions
# per second.
THRESHOLD = 40000 # How big the magnitude of a cyclic
# signal has to be before the measurement
# logic kicks in. This is a value from
# 0 to 65535 and might need to be adjusted
# up or down if the detection is too
# sensitive or not sensitive enough.
# Raising this value will make the detection
# less sensitive and require a very large
# difference in amplitude (i.e. a very close
# or highly reflective spinner), and lowering
# the value will make the detection more
# sensitive and potentially pick up random
# noise from light in the room.
# Configure NeoPixels and turn them all off at the start.
pixels = neopixel.NeoPixel(board.NEOPIXEL, 10)
pixels.fill((0,0,0))
pixels.write()
# Configure analog input for light sensor.
light = analogio.AnalogIn(board.LIGHT)
# Take an initial set of readings and measure how long it takes.
# This is used to calculate a delay between readings to hit the desired
# target sample rate. Use the array module to preallocate an array of 16-bit
# unsigned samples with lower memory overhead vs. a simple python list.
readings = array.array('H', [0]*SAMPLE_DEPTH)
start = time.monotonic()
for i in range(SAMPLE_DEPTH):
readings[i] = light.value
stop = time.monotonic()
print((stop-start)*1000.0)
# Calculate how long it took to take all the readings above, then figure out
# the difference from the target period to actual period. This difference is
# the amount of time to delay between sample readings to hit the desired target
# sample rate.
target_period = 1.0/TARGET_SAMPLE_RATE_HZ
actual_period = (stop-start)/SAMPLE_DEPTH
delay = 0
# Check that we can sample fast enough to hit the target rate.
if actual_period > target_period:
# Uh oh can't sample fast enough--print a warning and light up pixels red.
print('Could not hit desired target sample rate!')
pixels.fill((255,0,0))
pixels.write()
else:
# No problem hitting target sample rate so calculate the delay between
# samples to hit that desired rate. Then turn on the first three pixels
# to white full brightness.
delay = target_period - actual_period
pixels[0] = (255, 255, 255)
pixels[1] = (255, 255, 255)
pixels[2] = (255, 255, 255)
pixels.write()
# Main loop:
while True:
# Pause for a second between tachometer readings.
time.sleep(1.0)
# Grab a set of samples and measure the time it took to do so.
start = time.monotonic()
for i in range(SAMPLE_DEPTH):
readings[i] = light.value
time.sleep(delay) # Sleep for the delay calculated to hit target rate.
stop = time.monotonic()
elapsed = stop - start
# Find the min and max readings from the samples.
minval = readings[0]
maxval = readings[0]
for r in readings:
minval = min(minval, r)
maxval = max(maxval, r)
# Calculate magnitude or size of the signal. If the magnitude doesn't
# pass the threshold then start over with a new sample (run the loop again).
magnitude = maxval - minval
if magnitude < THRESHOLD:
continue
# Calculate the midpoint of the signal, then count how many times the
# signal crosses the midpoint.
midpoint = minval + magnitude/2.0
crossings = 0
for i in range(1, SAMPLE_DEPTH):
p0 = readings[i-1]
p1 = readings[i]
# Check if a pair of points crossed the midpoint either by hitting it
# exactly or hitting it going up or down.
if p1 == midpoint or p0 < midpoint < p1 or p0 > midpoint > p1:
crossings += 1
# Finally use the number of crosssings and the amount of time in the sample
# window to calculate how many times the spinner arms crossed the light
# sensor. Use that period to calculate frequency (rotations per second)
# and RPM (rotations per minute).
period = elapsed / (crossings / 2.0 / SPINNER_ARMS)
frequency = 1.0/period
rpm = frequency * 60.0
print('Frequency: {0:0.5} (hz)\t\tRPM: {1:0.5}\t\tPeriod: {2:0.5} (seconds)'.format(frequency, rpm, period))
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment