See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/319219169

TMS320F28335 DSP programming using MATLAB Simulink embedded coder:

Techniques and advancements

Conference Paper · July 2017

DOI: 10.1109/COMPEL.2017.8013418

CITATIONS READS
5 5,685

3 authors:

Akrem Mohamed Elrajoubi Simon S Ang

University of Arkansas University of Arkansas
6 PUBLICATIONS   17 CITATIONS    321 PUBLICATIONS   2,705 CITATIONS   

SEE PROFILE SEE PROFILE

Ali A. Abushaiba
University of Kansas
4 PUBLICATIONS   15 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Akrem Mohamed Elrajoubi on 28 February 2019.

The user has requested enhancement of the downloaded file.

 TMS320F28335 DSP Programming using MATLAB
Simulink Embedded Coder: Techniques and
Advancements.
Akrem Elrajoubi, Student IEEE member, Simon S. Ang, Fellow, IEEE, Ali Abushaiba, Student IEEE member,
Department of Electrical Engineering Department of Electrical Engineering Department of Electrical Engineering
University of Arkansas University of Arkansas University of Kansas
Fayetteville AR, USA Fayetteville AR, USA Lawrence KS, USA
[email protected] [email protected] [email protected]

Abstract—This paper provides a tutorial on how to program entering the “ver” command into the command window.
Texas Instruments™ (TI) TMS320F28335 Digital Signal Using Simulink code generation is more effective than
Processor (DSP) through Code Composer Studio (CCS) writing line by line any code by CCS which takes a long
version 6 and MATLAB Simulink Embedded Coder. First it time for users to program the DSP [6]. Section II presents
shows how to setup Simulink and Embedded Coder and the required settings for the DSP target configuration, while
produce code to program TMS320F28335 and variant of TI’s section III demonstrates the fundamental blocks used in
C2000 DSPs. It describes how to interact between MATLAB power electronics and motor drives applications from
2015b and CCS V6 and provides an explanation of the vital
steps and settings needed to program the DSP. Basic functions
Simulink and their main parameters. Serial Communication
such as Pulse Width Modulation, Analog Digital Conversion, Interface (SCI) transmit and receive blocks are shown in
and Proportional-Integral controllers are explained. Finally, section IV, and then closed loop control model is developed
closed loop control model for a micro-inverter topology is for the micro-inverter topology in section V.
developed.
II. CCSV6 TARGET CONFIGURATION
Keywords— TMS320F28335 DSP, CCS V6, MATLAB
Simulink. First make sure that Code Composer Studio (CCS)
version 6 or 5 is installed. If you have an older version of
I. INTRODUCTION CCS installed, you should upgrade to version 5 or CCS
Designing closed loop feedback multi PWM control version 6 which is now supported in Simulink for code
schemes is much harder and more time consuming in CCS generation. Embedded Coder (EC) works for previous
than in MATLAB Simulink Embedded Coder for Texas versions but they are no longer supported by Texas
Instruments™ TMS320F28335 Digital Signal Processor Instruments. For setting up xMakefile in Simulink which
(DSP). TMS320F28335 DSP is a cheaper controller which tells EC where the CCS 6 compiler is installed among other
proved excellent convergence, and real time control for programs. So MATLAB can call the command lines
significant reduction of output ripple [1]. The code is provided by CCS 6, and where MATLAB can find the
automatically generated using the embedded coder, and so compiler needed to create the makefile code. However,
time for programming and control implementation is “xmakefilesetup” command in MATLAB is no longer
reduced. TI DSPs present effective and simple control needed because of the Embedded Coder Support package
design for DC motors [2]. MATLAB Simulink for DSP for TI C2000 Processors which permits the settings for
controller is highly valuable as model design, simulation, model configuration parameters easily. Type
code generation, debugging and running can be “supportPackageInstaller” command to launch the Support
accomplished for control algorithm [3]. MATLAB Simulink Package Installer Graphical User Interface in MATLAB.
environment is especially recommended for control Then type this command
algorithm implementation into micro controller [4, 5]. The “checkEnvSetup('ccsv5','f28335','check')” in MATLAB, as
goal of this paper is to provide a simple and clear procedure shown in Fig. 1 the tools must be installed properly. For
to begin learning how to program the TMS320F28335 DSP F28335 we do not need header files but we need to install
from Texas Instruments™ (TI) through Code Composer Flash APIs from TI ControlSUITE webpage. In case you
Studio (CCS) and MATLAB Simulink Embedded Coder. need to check the compiler just type in the command
To begin, you need to make sure that you have the window: mex –setup , or mex.getCompilerConfigurations.
Embedded Coder, MATLAB Coder, and Simulink Coder For Matlab 2015a you will need to install the Embedded
toolboxes installed on your PC. Embedded coder sits on top Coder Support package for TI C2000 Processors. But for
of MATLAB and Simulink coder; it allows you to add Matlab 2011b and 2013a versions the library already exists.
device specific code (ADC’s, DAC’s, CAN, etc.) to what it Then we choose C2833x processor for our DSP control
produces by the respective coders. An easy way to check the card.
toolboxes installed on your version of MATLAB is by
 Fig.1. MATLAB checkEnvSetup command for F28335 DSP.

A. FLASH memory Programming (stand-alone mode)

If you are going to work on RAM programming mode,
you will set the configuration parameters to make the system
target file as either “ert.tlc” or “idelink_ert.tlc” on the code
generation page. For stand-alone mode, the program will be
saved in the flash memory of the DSP, so it will not be
erased when the DSP control card is unplugged from the
computer. Check boot from flash option in configuration
parameters window as shown in Fig. 2. This setting in
Model Configuration Parameters window tells EC what sort
of DSP is being programmed so that it initializes the right
peripherals, uses the correct operation frequency, knows
how much memory is available, etc. Open the configuration
parameters and ensure that the solver is set to fixed-step and
discrete, the fixed-step size should remain auto. The
hardware implementation page should show Texas
Instruments, C2000, and Little Endian [7, 8].
B. Debug Configurations and Code Running
We can generate the code by click Deploy to Hardware
(build model) icon. It will build dot out file which is
downloadable program file in CCSV6. Once you open CCS
you can select the configuration from the debug dropdown
menu and CCS will automatically connect to the DSP and
load the .out file for that project as shown in Fig. 3. You can
check code generation report after loading the model to see
the comments and the highlighted hyperlinks for specific Fig.2. Stand-alone execution configuration parameters.
blocks in the Simulink model.
 Fig.3. Loading the .out file for that project.

A TMS320F28335DSP board is shown in Fig. 4 with the Fig. 7 shows ePWM Block Parameters to generate
blinking GPIO34 LED after running the program. Since this two complementary PWM signals with a frequency of 20
programming was done on the flash memory of the DSP, KHz and duty cycle of 0.5. For frequency of 40 KHz, the
you can turn on and off the USB switch to see that the Timer period would be 1875. But for up counting mode it
program is still working and has not been erased from the would be 3750. The duty cycle is depending on the CMPA
RAM by turning off the power. value and the counting mode specified in Fig.7. Here it is up
counting mode so D= 0.5 since 1875/3750 = 0.5. As can be
III. PWM, ADC, GPIO, AND PI BLOCKS noticed, the frequency and duty cycle can be specified as
TMS320F28335DSP has up to 18 PWM outputs which input ports or via dialog. Under the General tap, the Timer
is adequate to control many three phase power converters. period can be specified and it is calculated as in (1). Every
12 of these outputs are ePWM modules which are shown in action can be chosen from the available choices (Do
Table 1. SYSCLK is set to 150 MHz, which is the nothing, Clear, Set, and Toggle). As can be seen, Action
maximum clock speed of the TMS320F28335 as specified when counter=CMPA on up-count (CAU) is Set, and Action
in the CPU clock. Fig. 5 shows two synchronized enhanced when counter=CMPA on down-count (CAD) is Clear. Also
Pulse Width Modulator (ePWM) blocks together to use the counting mode can be chosen (Up, Down, or Up-Down).
frequency and duty cycle as inputs. Timer period = (150 MHz )/(2 * 20 KHz)=3750 (1)
The pulses shown in Fig. 6 are for the two ePWM
outputs shown in Fig. 5 while they are synchronized with a
phase shift of TBPHS = 1500 (number of cycles for the time
period). Notice that these two ePWM outputs are 180 degree
shifted, the switching frequency is 50 KHz, and each ePWM
has two complementary signals. The duty cycles are 0.33
(=1000/(2*1500)) and 0.23 (=700/(2*1500)) for channel 1
(ePWM1A) and channel 3 (ePWM2A), respectively.

Fig.6. Synchronized ePWM pulses.

Fig.4. TMS320F28335DSP board. Fig.5. Synchronized two ePWM blocks.

 TABLE I. EPWM OUTPUT SIGNALS. The output values are in the range 0 to 4095
ePWM Modeule Module Outputs GPIO Pin because the ADC is a 12-bit converter, the input channel is
ADCINA0. Notice in Fig.9 that the option (Post interrupt at
ePWM1 ePWM1A GPIO00 the end of conversion) has been unchecked, and the sample
ePWM1B GPIO01 time is equal to the time period of the ePWM output pulse
ePWM2 ePWM2A GPIO02 (f=20 KHz). When we build, download and run this simple
model to the DSP board, we connect the power supply DC
ePWM2B GPIO03
voltage to the pin ADCINA0, and the negative to the ground
ePWM3 ePWM3A GPIO04 point. So, we are able to control the duty cycle of this PWM
ePWM3B GPIO05 signal by changing the voltage from 0 to 3V as can be seen
ePWM4 ePWM4A GPIO06 on the oscilloscope. Notice that on the ePWM block you
need to check the option (Enable ADC start of conversion
ePWM4B GPIO07
for module A) as shown in the figure below. Likewise we
ePWM5 ePWM5A GPIO08 can control the frequency of the ePWM output pulse by the
ePWM5B GPIO09 ADC input when we make it an input port.
ePWM6 ePWM6A GPIO10
ePWM6B GPIO11

Fig.8. ADC Example.

Fig.9. ADC and ePWM Blocks Parameters.

Fig.7. ePWM Block Parameters.
Fig. 10 shows an example model of using the PI
Fig.8 shows a simple example to use the Analog controller for changing the duty cycle of the ePWM output
Digital converter (ADC) block to perform analog-to-digital signals. Here the frequency was set to 40 KHz by making
conversion of signals connected to the selected ADC input the time period 3750 and Up counting mode.
pins. The output of the ADC is a vector of uint16 values.

Fig.10. PI controller for duty cycle of ePWM output.

 It is important to have the system controller
designed conveniently to make sure the response is as
desired. Notice that data type conversion block and rate
transition block have been included to convert input signal
to specified data type, and handle transfer of data between
blocks operating at different rates.
These are the basic settings to program the
TMS320F28335 DSP; there are a couple other things that
can be done to improve the code output. For instance, we
can set objectives for the code output in Configuration
Parameters window under the code generation advisor, such
as execution efficiency, rom efficiency, and ram efficiency.
If you click the set objectives button you will see a pop up
and you can put the options in a prioritized list. After you
have a model built you can click the check model button and
Simulink will analyze your model and provide tips to
improve the generated code. Now it is good to start building
your model. You can use any blocks from math, logic, and
discrete block sets. Other blocks can be used as long as they
do not have continuous states, the best way to check is by
trial and error, add it to your model and build the program.
If the block is not compatible Simulink will tell you. In
order to get to the DSP specific blocks (ADC, GPIO, CAN,
PWM, etc.) scroll down to Embedded Coder > Embedded Fig.12. SCI receive block.
Targets > Processors > Texas Instruments C2000 > C2833x.
There are also a number of IQ math and motor control
V. CLOSED LOOP CONTROL EXAMPLE
blocks located in the optimization subgroup. Some of these
blocks include Clark transformation, PID controller, Park This micro inverter has been designed using an
transformation, speed measurement, and space vector interleaved boost series resonant converter (IBSRC). As
generator. shown in Fig. 13, the IBSRC topology has two synchronous
boost converters having 180° phase shift and connected to a
IV. SERIAL COMMUNICATION high frequency transformer [10]. Then the secondary side is
Fig. 11 shows the Serial Communication Interface connected to an H bridge inverter to obtain an AC output
(SCI) transmit block, and Fig. 12 demonstrates the use of voltage.
the SCI receive block to establish the serial communication
between the host computer and the C2000 target hardware.
As such, they work to receive and transmit scalar or vector
data using the specified SCI hardware module.

Fig.13. Interleaved-Boost Series-Resonant Micro-inverter Topology [10].

The Model in Fig. 14 has been developed to control the

gate drivers of the IBSRC by controlling the frequency and
utilizing 4 ADC inputs as feedback from IBSRC sensors.
Each ePWM block generates two complementary pulses, so
this model produces 8 pulses for the micro inverter system.
The DSP will generate the signals for the gate drivers for
these 8 switches. The ePWM blocks are connected to these
8 switches as presented in Table 2. This is based on the
design of the IBSRC PCB board. In Fig. 14 four sensors
were used for the input signals of ADC1 and ADC2, and the
PI controllers implemented a discrete-time controller in
Simulink® model. The PID Controller block allows you to
Fig.11. SCI transmit block. implement setpoint weighting in your controller to achieve
both smooth setpoint tracking and good disturbance TABLE II. EPWM SIGNALS FOR IBSRC 8 SWITCHES.

rejection [9]. The reference values for the PI controllers are ePWM1A S1 ePWM1B S2
for the comparison with the ADC output to obtain the
targeted time period for ePWM blocks. Fig.15 shows the ePWM2A S3 ePWM2B S4
IBSRC was designed and developed in reference [10] with ePWM3A S5 ePWM3B S6
the F28335 DSP Card on the prototype.
ePWM4A S7 ePWM4B S8

Fig.14. TMS320F28335 DSP Model for IBSRC system.

VI. CONCLUSION
Using MATLAB Simulink embedded coder tools with
no hand writing code is time and money saving, more
efficient, faster execution time, and great for research and
industrial control design. This tutorial gives the instructions
on how to program the TMS320F28335 micro-controller
using make file approach with embedded coder in
MATLAB Simulink 2015b, and then debugging the
program in CCS V6. In this paper, the main steps for the
interaction between MATLAB 2015b and CCS V6 are
investigated and explained. Different versions of MATLAB
have some differences to interact with CCS, F28335 DSP
could be programmed using MATLAB 2011b and 2013a,
but there are some errors for flash programming mode.
IBSRC system was controlled by F28335 DSP through
Fig.15. Interleaved Isolated Boost Series Resonant Converter Prototype
MATLAB Simulink Embedded Coder. The Embedded
[10]. Coder Support package for TI C2833x Processor provides
the fundamental blocks needed for any power electronic,
smart grid or motor drives applications.
 REFERENCES
[1]. L. Katzir, Y. Loewenstern, B. Bishara and D. Shmilovitz,
"Implementation of a high voltage power supply with the
MATLAB/Simulink embedded coder," 2014 IEEE 28th Convention of
Electrical & Electronics Engineers in Israel (IEEEI), Eilat, 2014, pp. 1-4.
[2]. R. Duma, P. Dobra, M. Abrudean and M. Dobra, "Rapid prototyping of
control systems using embedded target for TI C2000 DSP," 2007
Mediterranean Conference on Control & Automation, Athens, 2007, pp. 1-
5.
[3]. S. Yuan and Z. Shen, "The Design of Matlab-DSP Development
Environment for Control System," 2012 Third International Conference on
Digital Manufacturing & Automation, GuiLin, 2012, pp. 903-906.
[4]. R. Grepl, "Real-Time Control Prototyping in MATLAB/Simulink:
Review of tools for research and education in mechatronics," 2011 IEEE
International Conference on Mechatronics, Istanbul, 2011, pp. 881-886.
[5]. B. dos Santos, R. E. Araújo, D. Varajão and C. Pinto, "Rapid
Prototyping Framework for real-time control of power electronic converters
using simulink," IECON 2013 - 39th Annual Conference of the IEEE
Industrial Electronics Society, Vienna, 2013, pp. 2303-2308.
[6]. J. C. Molina Fraticelli, “Simulink Code Generation Tutorial for
generating C code from Simulink Models using Simulink Coder” NASA
MARSHALL SPACE FLIGHT CENTER 2012. Available at:
https://www.mathworks.com/matlabcentral/answers/uploaded_files/56491/
Simulink%20Code%20Generation.pdf
[7].http://www.mathworks.com/help/ecoder/ug/tutorial-using-xmakefile-
with-code-composer-studio-4-x-
1.html?searchHighlight=code+composer+studio.
[8]. http://www.ti.com/tool/tmdscncd28335
[9]. http://www.mathworks.com/products/embedded-coder/webinars.html#.
[10]. L. A. Garcia-Rodriguez, C. Deng, J. C. Balda and A. Escobar-Mejía,
"Analysis, modeling and control of an interleaved isolated boost series
resonant converter for microinverter applications," 2016 IEEE Applied
Power Electronics Conference and Exposition (APEC), Long Beach, CA,
2016, pp. 362-369.

View publication stats