UM0585

User manual
STM32F10x
DSP library

Introduction
This user manual describes the STM32F10x DSP (digital signal processing) library, which is
a suite of common digital signal processing functions:
● PID controller
● Fast Fourier transform
● FIR and IIR filters
The library contains C and assembly functions. The assembly code is ported on ARM®,
GCC and IAR Systems™ assemblers.

June 2010 Doc ID 14988 Rev 2 1/25

www.st.com
Contents UM0585

Contents

1 DSP Library description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 PID controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 DSP library functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.1 DoPID function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.2 DoFullPID function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.3 PID_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3 Complex 16-bit radix-4 FFT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2 DSP library functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2.1 cr4_fft_64_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2.2 cr4_fft_256_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2.3 cr4_fft_1024_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3 FFT performance improvement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4 16-bit FIR filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

4.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2 DSP library function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2.1 fir_16by16_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

5 16-bit IIR filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

5.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5.2 DSP library functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.2.1 iiarma_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.2.2 iir_biquad_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

6 STM32F10x DSP library benchmark . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

6.1 Function code footprint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
6.2 Function execution time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
6.2.1 PID controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
6.2.2 Fast Fourier transform (FFT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
6.2.3 FIR filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2/25 Doc ID 14988 Rev 2

UM0585 Contents

6.2.4 IIR filters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

7 STM32F10x DSP demo description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

9 Revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Doc ID 14988 Rev 2 3/25

List of tables UM0585

List of tables

Table 1. STM32F10x DSP library functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Table 2. DoPID function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Table 3. DoFullPID function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Table 4. PID_stm32 function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Table 5. cr4_fft_64_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Table 6. cr4_fft_256_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Table 7. cr4_fft_1024_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Table 8. fir_16by16_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Table 9. iirarma_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Table 10. iir_biquad_stm32 function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Table 11. STM32F10x DSP library functions code footprint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Table 12. PID controller, error computed outside the routine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Table 13. PID controller, error computed within the routine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Table 14. Complex radix 4, 16-bit FFT, coefficients in Flash memory . . . . . . . . . . . . . . . . . . . . . . . . 20
Table 15. Complex radix 4, 16-bit FFT, coefficients in RAM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Table 16. 16-bit, 32-tap FIR filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Table 17. 16-bit canonic form, 4 biquad IIR filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Table 18. 16-bit, 4th-order IIR filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Table 19. Document revision history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

4/25 Doc ID 14988 Rev 2

UM0585 List of figures

List of figures

Figure 1. Block diagram of PID controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Figure 2. Block diagram of an FIR filter of length N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Figure 3. Block diagram of the direct form I of second-order IIR filter . . . . . . . . . . . . . . . . . . . . . . . . 15
Figure 4. Block diagram of the canonical form of a second-order IIR filter . . . . . . . . . . . . . . . . . . . . 16
Figure 5. FFT of a sine wave with frequency f1 and of a dual sine wave with
frequencies f1 and f2, both sampled at a frequency Fs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Doc ID 14988 Rev 2 5/25

DSP Library description UM0585

1 DSP Library description

The STM32F10x DSPLib is a suite of common functions for signal processing. It includes
the following functions:

Table 1. STM32F10x DSP library functions

Function name Description

DoPID PID controller in C, error computed outside the routine

DoFullPID PID controller in C, error computed inside the routine
PID_stm32 PID controller in ASM, error computed outside the routine
cr4_fft_64_stm32 Complex radix-4 FFT for a 16-bit, 64-point vector
cr4_fft_256_stm32 Complex radix-4 FFT for a 16-bit, 256-point vector
cr4_fft_1024_stm32 Complex radix-4 FFT for a 16-bit, 1024-point vector
fir_16by16_stm32 16-bit FIR filter
iiarma_stm32 16-bit, 4th-order IIR filter
iir_biquad_stm32 16-bit, 8th-order, biquad IIR filter

6/25 Doc ID 14988 Rev 2

UM0585 PID controller

2 PID controller

2.1 Description
The proportional-integral-derivative or PID controller is commonly used in the industry. It is a
feedback loop controller that manages the process command with a view to reducing the
error between the desired set point and the measured process variable.
The following block diagram shows the parallel structure of a PID controller. This is the
structure implemented in this DSP library.

Figure 1. Block diagram of PID controller

Ref Error + command Output

+- I + Process
+
In

ai15447

2.2 DSP library functions

The DSP library provides three PID functions:
● DoPID: a PID core loop coded in C (the error is computed outside the function)
● DoFullPID: a full PID controller coded in C (with error computing)
● PID_stm32: an optimized PID core loop written in assembly

2.2.1 DoPID function

Table 2 describes the DoPID function.

Table 2. DoPID function

Function name DoPID

Prototype uint16_t DoPID(uint16_t Error, uint16_t *Coeff)

Behavior description PID in C, error computed outside the function.
– Error: difference between reference and measured values
Input parameter
– Coeff: pointer to the coefficients table
Output parameter None
Return parameter PID output command

Example
/* Fill the coefficients table */
Coeff[0] = Kp;/*proportional coefficient*/

Doc ID 14988 Rev 2 7/25

PID controller UM0585

Coeff[1] = Ki;/*integral coefficient*/

Coeff[2] = Kd;/*derivative coefficient*/
/* Compute the error */
Error = Target_Signal - Measured_Signal;
/* PID control process */
Command = DoPID(Error, Coeff);

2.2.2 DoFullPID function

Table 3 describes the DoFullPID function.

Table 3. DoFullPID function

Function name DoFullPID

uint16_t DoFullPID(uint16_t In, uint16_t Ref,

Prototype
uint16_t *Coeff)
Behavior description PID in C, error computed inside the function.
– In: Input (measured value)
Input parameter – Ref: reference (target value)
– Coeff: pointer to the coefficients table
Output parameter Computed error
Return parameter PID output command

Example
/* Fill the coefficients table */
Coeff[0] = Kp;/*proportional coefficient*/
Coeff[1] = Ki;/*integral coefficient*/
Coeff[2] = Kd;/*derivative coefficient*/

/* PID control process, the error is computed inside the function */

Command = DoFullPID(Measured_Signal, Target_Signal, Coeff);

2.2.3 PID_stm32 function

Table 4 describes the PID_stm32 function.

Table 4. PID_stm32 function

Function name PID_stm32

Prototype uint16_t PID_stm32(uint16_t Error, uint16_t *Coeff);

Behavior description Assembly optimized PID controller with error computed outside the function.
– Error: difference between reference and measured values
Input parameter
– Coeff: pointer to the coefficients table
Output parameter None
Return parameter PID output command

8/25 Doc ID 14988 Rev 2

UM0585 PID controller

The PID_stm32 function is used in the same way as the DoPID function. The error must be
computed, then the PID_stm32 function is called to improve the PID control process and to
return the appropriate command according to the coefficients table.

Doc ID 14988 Rev 2 9/25

Complex 16-bit radix-4 FFT UM0585

3 Complex 16-bit radix-4 FFT

3.1 Description
The discrete Fourier transform (DFT) converts N complex values from the time domain to
the frequency domain.
The fast Fourier transform (FFT) is an optimized algorithm designed to compute the DFT
efficiently.
The STM32F10x DSP library provides a complex radix-4, with decimation-in-time, linear-
order FFT.
Let x[N] be the time signal samples. To use the FFT functions of the DSP library, the
following conditions must be satisfied:
● N is a power of 4
● All the signal samples must be 32-bit data containing the 16-bit real part followed by the
16-bit imaginary part (in the little Endian order: imaginary_real).

3.2 DSP library functions

The DSP provides three complex 16-bit radix-4 FFT functions:
1. cr4_fft_64_stm32: an optimized FFT function to compute 64-point DFT
2. cr4_fft_256_stm32: an optimized FFT function to compute 256-point DFT
3. cr4_fft_1024_stm32: an optimized FFT function to compute 1024-point DFT

3.2.1 cr4_fft_64_stm32 function

Table 5 describes the cr4_fft_stm32 function.

Table 5. cr4_fft_64_stm32 function

Function name cr4_fft_64_stm32

void cr4_fft_64_stm32(void *pssOUT, void *pssIN,

Prototype
uint16_t Nbin);
Behavior description complex 16-bit, 64-point radix-4 FFT
– pssOUT: pointer to the output array data
Input parameter – pssIN: pointer to the input array data
– Nbin: the number of points, must be 64.
Output parameter None
Return parameter None

10/25 Doc ID 14988 Rev 2

UM0585 Complex 16-bit radix-4 FFT

3.2.2 cr4_fft_256_stm32 function

Table 6 describes the cr4_fft_256_stm32 function.

Table 6. cr4_fft_256_stm32 function

Function name cr4_fft_256_stm32

void cr4_fft_256_stm32(void *pssOUT, void *pssIN,

Prototype
uint16_t Nbin);
Behavior description complex 16-bit, 256-point radix-4 FFT
– pssOUT: pointer to the output array data
Input parameter – pssIN: pointer to the input array data
– Nbin: the number of points, must be 256.
Output parameter None
Return parameter None

3.2.3 cr4_fft_1024_stm32 function

Table 7 describes the cr4_fft_1024_stm32 function.

Table 7. cr4_fft_1024_stm32 function

Function name cr4_fft_1024_stm32

void cr4_fft_1024_stm32(void *pssOUT, void *pssIN,

Prototype
uint16_t Nbin);
Behavior description complex 16-bit, 1024-point radix-4 FFT
– pssOUT: pointer to the output array data
Input parameter – pssIN: pointer to the input array data
– Nbin: the number of points, must be 1024.
Output parameter None
Return parameter None

Example
#define N 64 /*Number of points*/
uint32_t x[N],y[N]; /* input and output arrays */
uint16_t real[N], imag[N]; /* real and imaginary arrays */
/* Fill the input array */
for(i=0; i<N; i++)
x[i] = (((uint16_t)(real[i])) | ((uint32_t)(imag[i]<<16)));
cr4_fft_64_stm32(y, x, N); /*computes the FFT of the x[N] samples*/

3.3 FFT performance improvement

The FFT coefficients table is stored in the Flash memory since it is declared as code in the
assembly file. The performance of the FFT function can be improved by placing the FFT
coefficients in RAM. This is done as described below:

Doc ID 14988 Rev 2 11/25

Complex 16-bit radix-4 FFT UM0585

1. Comment all the FFT coefficients in the FFT function assembly file.
/* TableFFT_V7
;N=16
DC16 0x4000,0x0000, 0x4000,0x0000, 0x4000,0x0000
DC16 0xdd5d,0x3b21, 0x22a3,0x187e, 0x0000,0x2d41
DC16 0xa57e,0x2d41, 0x0000,0x2d41, 0xc000,0x4000
DC16 0xdd5d,0xe782, 0xdd5d,0x3b21, 0xa57e,0x2d41
; N=64
DC16 0x4000,0x0000, 0x4000,0x0000, 0x4000,0x0000
DC16 0x2aaa,0x1294, 0x396b,0x0646, 0x3249,0x0c7c
...
*/
2. Then, in the main, include the table_fft.h file, which is a part of the DSP library.
3. Finally, go to the FFT function assembly code, and inverse the comment in the following
lines:
ADRL R0, TableFFT_V7 /* Coeff in Flash */
//LDR.W R0, =TableFFT /* Coeff in RAM */

12/25 Doc ID 14988 Rev 2

UM0585 16-bit FIR filter

4 16-bit FIR filter

4.1 Description
The finite impulse filter (FIR) is a digital filter that is linearly dependent on a fixed finite
number of input samples.
The FIR filter can be defined by the number of coefficients to be processed (also called
taps), which gives the number of MAC (multiply-accumulate) operations to be done.

Figure 2. Block diagram of an FIR filter of length N

xn xn-1 xn-2 xn-N
Z-1 Z-1 Z-1

c0 c1 c2 cN-1

yn
+ + + +
ai15448

The FIR filter of the DSP library is a direct-form real FIR filter that uses an array of M 16-bit
coefficients to filter N 16-bit samples.
Let us put:
● a, the output vector of length N
● c, the coefficients vector of length M
● x, the input vector
So, x must have a length of M + N – 1.

4.2 DSP library function

The FIR function of the DSP library is an optimized assembly function that takes into
account the load-store architecture of the Cortex™-M3. Therefore, and as an optimization
constraint, the number of taps and the number of output samples must be a multiple of 4.

Doc ID 14988 Rev 2 13/25

16-bit FIR filter UM0585

4.2.1 fir_16by16_stm32 function

Table 8 describes the fir_16by16_stm32 function.

Table 8. fir_16by16_stm32 function

Function name fir_16by16_stm32

void fir_16by16_stm32(int *a,short *x,struct COEFS

Prototype
*p,unsigned int N)
Behavior description Block Fir 16-bit filter.
– a: output array
– x: input array
Input parameter
– p: pointer to the coefficient structure of the COEFS type
– N: number of output samples
Output parameter None
Return parameter None

The filter coefficients and their number must be filled into a structure of the COEFS type.
The coefficients structure is defined as follows:
typedef struct {
short *h;
unsigned int nh;
}COEFS;

Example
#define M 32 /*number of coefficients*/
#define N 32 /*number of output samples*/

COEFS fir_coefs;/*coefficients structure*/

int a[N];/*filter output vector*/

short x[M+N-1] = {x0,x1...,xM+N-1};/*filter input vector*/
short h[M]={h0,h1...,hM-1};/*filter coefficients vector*/

fir_coefs.nh = M; /*Number of Coefficients for FIR*/

fir_coefs.h = h; /*Pointer on FIR coefficient vector*/

fir_16by16_stm32(a,x,&fir_coefs,N);/*performs the FIR filtering*/

14/25 Doc ID 14988 Rev 2

UM0585 16-bit IIR filters

5 16-bit IIR filters

5.1 Description
The infinite impulse response (IIR) filter is a digital filter that depends linearly on a finite
number of input samples and a finite number of previous filter outputs.
The IIR filter is represented by Equation 1 below.

Equation 1
M N

yn =  bi x  n – i  –  ai y  n – i 
i=0 i=1

Equation 1 is known as an auto-regressive moving average form (ARMA). The first sum of
Equation 1 represents the moving average part, which is similar to an FIR block, and the
second sum represents the auto-regressive part, which is the feedback from previous
outputs.
This IIR filter structure is called direct form I.
Figure 3 shows the structure of the direct form I of a second-order IIR filter. Direct form I
uses four delays to realize a second-order IIR filter.
To reduce the number of delays, the canonical form may be used. Figure 4 represents the
block diagram of the canonical form of a second-order IIR filter. The number of delays is
reduced from 4 to only 2.

Figure 3. Block diagram of the direct form I of second-order IIR filter

xn yn
b0 + +

Z-1 Z-1

b1 + + -a1

Z-1 Z-1

b2 + + -a2

ai15449

Doc ID 14988 Rev 2 15/25

16-bit IIR filters UM0585

Figure 4. Block diagram of the canonical form of a second-order IIR filter

xn wn yn
+ b0 +

Z-1

+ -a1 b1 +

Z-1

-a2 b2
wn-2
ai15450

The DSP library implements the two forms (direct form I and the canonical form) of the IIR
filter:
● Direct form I is used to design an IIR filter of order 4 (ARMA IIR)
● The canonical form is used to design an IIR filter of order 8, by using 4 second-order
IIR filter sections (biquads) arranged in series (biquad IIR filter).

5.2 DSP library functions

The DSP library provides two IIR filters:
1. iiarma_stm32: an ARMA IIR filter, designed with 4 auto-regressive and 5 moving-
average filter coefficients, so, in Equation 1, M = 4 and N = 4. The function code is
written in assembly.
2. iir_biquad_stm32: a biquad IIR filter, designed by connecting 4 biquads in series.
The function code is written in C.

5.2.1 iiarma_stm32 function

Table 9 describes the iirarma_stm32 function.

Table 9. iirarma_stm32 function

Function name iirarma_stm32

void iirarma_stm32(short *y, short *x, short *h2, short

Prototype
*h1, int ny)
Behavior description 16-bit, auto-regressive moving-average IIR (ARMA) filter.
– y: output array of length ny+4
– x: input array of length ny+4
Input parameter – h2: filter coefficient vector, moving-average part.
– h1: filter coefficient vector, auto-regressive part.
– ny: number of output samples
Output parameter None
Return parameter None

16/25 Doc ID 14988 Rev 2

UM0585 16-bit IIR filters

To use the iirarma_stm32 function, the following conditions must be satisfied:

● The moving-average coefficient vector, h2, must have a length of 5 shorts:
– short h2[5];
● The auto-regressive coefficient vector, h1, must have a length of 5 shorts. The h1[0]
value is not used, so 4 coefficients remain:
– short h1[5];
● The number of output samples, ny, must be a multiple of 4 greater than or equal to 8.
● Input and output vectors must have a length of (ny+4) shorts.
● The first four elements of the output vector must have the previous outputs.

Example
#define NY 32/*number of outputs, must be a multiple of 4 and >= 8*/
/* Coefficients for the ARMA IIR filter */
short h2[5] = { 0x09c2, 0x270a, 0x3a8f, 0x270a, 0x09c2 };
short h1[5] = { 0x7fff, 0xd24a, 0x72ca, 0xcf4e, 0x1ad4 };
/* Input and output vectors */
short x[NY+4],y[NY+4];

/* Fill the input vector x */

...
/* Fill the 4 previous outputs */
y[0] = y0;
y[1] = y1;
y[2] = y2;
y[3] = y3;
/* Improve the filtering of NY samples */
iirarma_stm32(y, x, h2, h1, NY);

5.2.2 iir_biquad_stm32 function

Table 10 describes the iir_biquad_stm32 function.

Table 10. iir_biquad_stm32 function

Function name iir_biquad_stm32

void iir_biquad_stm32(uint16_t *y, uint16_t *x, int16_t

Prototype
*IIRCoeff, uint16_t ny)
Behavior description 8th-order 4 biquad IIR filter
– y: output array
– x: input array
Input parameter
– IIRCoeff: IIR filter coefficients
– ny: number of output samples
Output parameter None
Return parameter None

Doc ID 14988 Rev 2 17/25

16-bit IIR filters UM0585

Example
#define NY 32/*number of outputs*/
/* Coefficients for the biquad IIR filter: 4 sections, with 5
coefficients in each section */
int16_t Coeff[20] = {...};
/* Input and output vectors */
short x[NY],y[NY];
/* Fill the input vector x */
...
/* Improve the filtering of NY samples */
iir_biquad_stm32(y, x, Coeff, NY);

18/25 Doc ID 14988 Rev 2

UM0585 STM32F10x DSP library benchmark

6 STM32F10x DSP library benchmark

This section provides the STM32F10x DSP library benchmark results, which are computed
using the IAR EWARM 5.20 toolchain, with high-speed optimization.

6.1 Function code footprint

Table 11. STM32F10x DSP library functions code footprint
Function name Code size (bytes)

DoPID 52
DoFullPID 58
PID_stm32 72
cr4_fft_64_stm32 718(1)
cr4_fft_256_stm32 1486(1)
cr4_fft_1024_stm32 4560(1)
fir_16by16_stm32 162
iiarma_stm32 156
iir_biquad_stm32 294
1. FFT code size was computed with FFT coefficients table stored in Flash memory. If the FFT coefficients
are stored in SRAM, the code size of the three FFT functions is equal to 480 bytes.

6.2 Function execution time

6.2.1 PID controller

Table 12. PID controller, error computed outside the routine

24 MHz 48 MHz 72 MHz
0 wait state 1 wait state 2 wait states
PID
cycle count time cycle count time cycle count time

ASM function 45 1.87 µs 51 1.06 µs 59 0.819 µs

C function 47 1.96 µs 50 1.04 µs 54 0.75 µs

Analysis of the PID timing shows that assembly code is not as fast as C code. The compiler
is more efficient in accessing variables than manual optimization (offset computation and
data placement in literal pool).

Doc ID 14988 Rev 2 19/25

STM32F10x DSP library benchmark UM0585

Table 13. PID controller, error computed within the routine

24 MHz 48 MHz 72 MHz
0 wait state 1 wait state 2 wait states
PID
cycle count time cycle count time cycle count time

C function 48 2 µs 52 1.08 µs 57 0.79 µs

6.2.2 Fast Fourier transform (FFT)

Table 14. Complex radix 4, 16-bit FFT, coefficients in Flash memory

24 MHz 48 MHz 72 MHz
FFT 0 wait state 1 wait state 2 wait states
(ASM funct.)
cycle count time cycle count time cycle count time

64 points 3847 0.16 ms 4 472 0.093 ms 5 661 0.078 ms

256 points 21 039 0.876 ms 24 964 0.52 ms 31 527 0.437 ms
1024 points 100 180 4.174 ms 114 350 2.382 ms 153 930 2.138 ms

Table 15. Complex radix 4, 16-bit FFT, coefficients in RAM

24 MHz 48 MHz 72 MHz
FFT 0 wait state 1 wait state 2 wait states
(ASM funct.)
cycle count time cycle count time cycle count time

64 points 3 847 0.16 ms 4 025 0.084 ms 4 764 0.066 ms

256 points 21 039 0.876 ms 22 176 0.462 ms 26 065 0.362 ms
1024 points 100 180 4.174 ms 102 057 2.126 ms 127 318 1.768 ms

6.2.3 FIR filter

Table 16. 16-bit, 32-tap FIR filter

24 MHz 48 MHz 72 MHz
FIR 0 wait state 1 wait state 2 wait states
(ASM filter)
cycle count time cycle count time cycle count time

32 samples 3516 146.5 µs 3525 73.4 µs 3727 51.76 µs

20/25 Doc ID 14988 Rev 2

UM0585 STM32F10x DSP library benchmark

6.2.4 IIR filters

Table 17. 16-bit canonic form, 4 biquad IIR filter

24 MHz 48 MHz 72 MHz
IIR 0 wait state 1 wait state 2 wait states
(C filter)
cycle count time cycle count time cycle count time

32 samples 3478 144.9 µs 3636 75.75 µs 3929 54.57 µs

Table 18. 16-bit, 4th-order IIR filter

24 MHz 48 MHz 72 MHz
IIR 0 wait state 1 wait state 2 wait states
(ASM filter)
cycle count time cycle count time cycle count time

32 samples 1696 70 µs 1761 36.69 µs 1986 27.58 µs

Doc ID 14988 Rev 2 21/25

STM32F10x DSP demo description UM0585

7 STM32F10x DSP demo description

The STM32F10x DSP demo is an example that illustrates how to use the 64-point FFT
function of the STM32F10x DSP library. It consists of an LCD that displays the FFT
transformation.
The demo runs on the STM3210B-EVAL board, and shows the 64-point FFT transformation
of two signals:
1. a sine wave
2. a dual sine wave
The two types of wave are displayed at variable frequencies and for each frequency the FFT
is computed and the result is plotted as shown in Figure 5:
● The frequency (f1) of the sine wave is increased from 40 Hz to 4 kHz by steps of 30 Hz.
● The dual sine wave is the sum of a constant-frequency (f2) sine wave and the above
described variable-frequency (f1) sine wave.

Figure 5. FFT of a sine wave with frequency f1 and of a dual sine wave with
frequencies f1 and f2, both sampled at a frequency Fs
Sine wave

–Fs –f1 f1 Fs
2 2

Dual sine wave

–Fs –f2 –f1 f1 f2 Fs

2 2

ai15649b

22/25 Doc ID 14988 Rev 2

UM0585 Conclusion

8 Conclusion

This user manual describes the STM32F10x DSP library, which contains:
● a PID controller
● complex 16-bit radix-4 FFT optimized functions for 64, 256 and 1024 points
● a 16-bit FIR filter
● a 16-bit direct-form I IIR filter
● a 16-bit canonical-form IIR filter designed by biquads

Doc ID 14988 Rev 2 23/25

Revision history UM0585

9 Revision history

Table 19. Document revision history

Date Revision Changes

13-Oct-2008 1 Initial release.

Introduction modified.
Table 1: STM32F10x DSP library functions updated.
Section 7: STM32F10x DSP demo description added.
04-Jun-2010 2 u32 changed to uint32_t.
u16 changed to uint16_t.
s16 changed to int16_t.
Small text changes.

24/25 Doc ID 14988 Rev 2

UM0585

Please Read Carefully:

Information in this document is provided solely in connection with ST products. STMicroelectronics NV and its subsidiaries (“ST”) reserve the
right to make changes, corrections, modifications or improvements, to this document, and the products and services described herein at any
time, without notice.
All ST products are sold pursuant to ST’s terms and conditions of sale.
Purchasers are solely responsible for the choice, selection and use of the ST products and services described herein, and ST assumes no
liability whatsoever relating to the choice, selection or use of the ST products and services described herein.
No license, express or implied, by estoppel or otherwise, to any intellectual property rights is granted under this document. If any part of this
document refers to any third party products or services it shall not be deemed a license grant by ST for the use of such third party products
or services, or any intellectual property contained therein or considered as a warranty covering the use in any manner whatsoever of such
third party products or services or any intellectual property contained therein.

UNLESS OTHERWISE SET FORTH IN ST’S TERMS AND CONDITIONS OF SALE ST DISCLAIMS ANY EXPRESS OR IMPLIED
WARRANTY WITH RESPECT TO THE USE AND/OR SALE OF ST PRODUCTS INCLUDING WITHOUT LIMITATION IMPLIED
WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE (AND THEIR EQUIVALENTS UNDER THE LAWS
OF ANY JURISDICTION), OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.
UNLESS EXPRESSLY APPROVED IN WRITING BY AN AUTHORIZED ST REPRESENTATIVE, ST PRODUCTS ARE NOT
RECOMMENDED, AUTHORIZED OR WARRANTED FOR USE IN MILITARY, AIR CRAFT, SPACE, LIFE SAVING, OR LIFE SUSTAINING
APPLICATIONS, NOR IN PRODUCTS OR SYSTEMS WHERE FAILURE OR MALFUNCTION MAY RESULT IN PERSONAL INJURY,
DEATH, OR SEVERE PROPERTY OR ENVIRONMENTAL DAMAGE. ST PRODUCTS WHICH ARE NOT SPECIFIED AS "AUTOMOTIVE
GRADE" MAY ONLY BE USED IN AUTOMOTIVE APPLICATIONS AT USER’S OWN RISK.

Resale of ST products with provisions different from the statements and/or technical features set forth in this document shall immediately void
any warranty granted by ST for the ST product or service described herein and shall not create or extend in any manner whatsoever, any
liability of ST.

ST and the ST logo are trademarks or registered trademarks of ST in various countries.

Information in this document supersedes and replaces all information previously supplied.

The ST logo is a registered trademark of STMicroelectronics. All other names are the property of their respective owners.

© 2010 STMicroelectronics - All rights reserved

STMicroelectronics group of companies

Australia - Belgium - Brazil - Canada - China - Czech Republic - Finland - France - Germany - Hong Kong - India - Israel - Italy - Japan -
Malaysia - Malta - Morocco - Philippines - Singapore - Spain - Sweden - Switzerland - United Kingdom - United States of America
www.st.com

Doc ID 14988 Rev 2 25/25