Summary: A 25kHz PWM signal (5VDC peak Vpp), using a simple RC low-pass filter, needs to be converted to an analog voltage to be measured using an Arduino MEGA 2560. The filter (with a cutoff frequency of ~25Hz) is assembled on a breadboard with component values of C = 2000uF and R = 3.24Ohms). The analog voltage is read by the Arduino using the analogRead function; however, the output changes from 1023 to 0 for PWM duty cycles of 100% to 50% (and below) instead of 100% to 0% (expected full range). Please advise on any possible mistakes I may have made. Thanks!!!
Let me explain my project to make it clear why I am trying this approach for PWM measurement. Computers (in my case, servers) control the speed of chassis DC fans (4-wire) using a PWM signal based on CPU temperature. I am trying to cool multiple servers stacked on top of each other by a single bank of larger external fans (internal chassis fans are removed). Since each individual server creates a PWM output, I want to measure each signal, compare the values (post conversion to analog DCV and read by the Arduino) and create a PWM signal to the external fans using the Arduino. Please explain why the filter only works between 100% and 50% duty cycles instead of the entire range (100-0). Thank you!!!
Are there issues with my selection of cutoff frequency or components values? I will post a picture of my test setup soon.
You must have the resistor value of 150 ohms or more or you'll burn out
the Arduino pin. That large a capacitance is effectively a short circuit so
the Arduino is putting out far beyond its absolute maximum current trying to
drive it.
While the capacitive reactance of 2000uF at 25Hz is 3.18 ohms the over all load being driven by the arduino is going to be just over 6R. This is way too low to be driven by a logic pin. In other words the input impedance of the filter is way too low.
The absolute value of filter component values are as important as the relative values.
Thanks for the response fellas. As promised, attached is a picture of my setup. The 25kHz PWM signal is supplied by a function generator. The filter is assembled on the breadboard and the output voltage is read by analog pin #8 on the Arduino.
Do you guys have any theories why the filter only works between 100% to 50%? Thanks again!!!
Try the tool here http://sim.okawa-denshi.jp/en/CRlowkeisan.htm
The capacitor looks like a dead short as it charges up.
With a resistor of 3.24ohm, you are practically shorting the output to Gnd.
Try 62K and 0.1uF instead. The current required is much less to change the charge of the capacitor.
Update: I have since tried the following combinations of C and R - (10uF & 680 Ohms) and (4.7uF & 1.5 kOhms). There is no change whatsoever in the response - it still works only between 100% and 50% duty cycles (corresponding output voltage is 5V to 0V).
CrossRoads - I'll try your suggested values next. Thanks for your input.
Be aware that a signal generator might have a different output
impedance from an Arduino pin, but even that doesn't explain non-linear
response - there must be some non-linear element in the circuit.
Update: I have since tried the last suggested combination of 66 kOhms (couldn't find 62k at Radioshack, so had to put two 33k resistors in series) and 0.1uF components. There is no change whatsoever; I still see the output signal go from 5V to 0V in response to duty cycles from 100% to 50%. The output stays zero between 50% and 0%.
Thanks for continuing to provide input. Hopefully, this issue gets resolved soon.
If you're controlling a PWM signal based on CPU temperature ... could this explain the 50-100% control range? The filter wouldn't be the problem for an abrupt cut-off at 50%.
You should manually be able to control the PWM throughout its full range. Check if the variable type for CPU temperature is the problem ... use "unsigned" to eliminate negative numbers.
Update: Based on Ray's input I tried different combinations of resistors and capacitors at my disposal to try different cutoff frequencies. Also, I replaced the Arduino with a datalogger to read the output voltage. See the attached image for the results at different duty cycles (25kHz and 5.25Vpp). As before, we see the output drop from 5.2V to 0V between 100% and 50%. However, it is interesting to observe the that voltages become negative at 30% and 0% and mirror the magnitudes at the positive end.
From you're test results it looks like in fact there is full range control with 0V at center point (50% duty).
EDIT: The circuit looks to be connected OK.
Your function generator for PWM control ... is it set to the correct PWM mode so that the signal is 0 to +5V with 0-100% duty? Or is it -5V to 0 for 0-50% duty and 0 to +5V from 50-100% duty?
