Change request double type to float

[mrguen]

Thanks for this great library. I am now using it on STM32 and the double type is 64 bits so it takes a lot of memory when using the actual "double" type that is used everywhere.

I changed them to float. This is much more convenient and does not visibly reduce the precision. For usual Arduino boards like the UNO float and double datatypes are equivalent 32 bits data.

[HorstBaerbel]

See the develop branch and / or pull request #42.

[mrguen]

@HorstBaerbel Thanks. I had the dev branch working well, using a 4096 values float array on my STM32F401RCT6 with Arduino Core STM32

But not working for regular Uno board:

"
In file included from C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\Examples\FFT_speedup\FFT_speedup.ino:32:0:

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h: In member function 'void ArduinoFFT::compute(FFTDirection) const':

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h:141:3: error: 'small_type' was not declared in this scope

small_type index = 0;

^~~~~~~~~~

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h:141:3: note: suggested alternative: 'decltype'

small_type index = 0;

^~~~~~~~~~

decltype

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h:169:31: error: 'index' was not declared in this scope

c2 = pgm_read_T_near(&(_c2[index]));

                           ^~~~~

In file included from C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\Examples\FFT_speedup\FFT_speedup.ino:32:0:

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h: In instantiation of 'class ArduinoFFT':

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\Examples\FFT_speedup\FFT_speedup.ino:57:19: required from here

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h:388:17: error: in-class initialization of static data member 'const float ArduinoFFT::_c1 []' of incomplete type

static const T _c1[] PROGMEM = {

             ^~~

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h:394:17: error: in-class initialization of static data member 'const float ArduinoFFT::_c2 []' of incomplete type

static const T _c2[] PROGMEM = {

             ^~~

In file included from C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\Examples\FFT_speedup\FFT_speedup.ino:32:0:

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h: In instantiation of 'void ArduinoFFT::compute(FFTDirection) const [with T = float]':

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\Examples\FFT_speedup\FFT_speedup.ino:90:36: required from here

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h:169:38: error: 'pgm_read_T_near' was not declared in this scope

c2 = pgm_read_T_near(&(_c2[index]));

                                  ^

C:\Users\TG\Documents\Arduino\libraries\arduinoFFT-develop\src/arduinoFFT.h:169:38: note: suggested alternative: 'pgm_read_ptr_near'

c2 = pgm_read_T_near(&(_c2[index]));

                                  ^

                                  pgm_read_ptr_near

"

[HorstBaerbel]

Damn. Should be fixed in pull request #50.

[mrguen]

@HorstBaerbel

Thanks for the dev version. I am not testing it on an Atmega1284 and saw these:

1. I think there is a little error in the examples. When creating a synthetic wave
   float cycles = (((samples-1) * signalFrequency) / samplingFrequency); //Number of signal cycles that the sampling will read

should be:
float cycles = (((samples) * signalFrequency) / samplingFrequency); //Number of signal cycles that the sampling will read

The duration of the sampling is indeed = samples / samplingFrequency either you index them from 0 or 1.

2. Also I wonder if it is normal to have a peak at "0 H", even after removing the DC component?

[HorstBaerbel]

I don't know as I'm not the maintainer. I'd suggest you open a new issue when you're sure these things are errors, as they have nothing to do with this one. You can also directly point to source code lines using the web interface: Navigate to the file, click the line number, click the "..." menu and then choose the option you'd like, e.g. "create permalink" or even "new issue".

[mrguen]

Ok, I will do that. #52