1 00:00:00,000 --> 00:00:03,629 hello everyone my name is Antone and i will present today how to control the 2 00:00:03,629 --> 00:00:09,599 three face brushless motor using similar ink and a TI launch pad with a three 3 00:00:09,599 --> 00:00:16,410 phase inverter booster back here is the hardware i have my ti launch pad here 4 00:00:16,410 --> 00:00:22,740 with the booster pack on top im using the boost excel dash drv 8301 booster 5 00:00:22,740 --> 00:00:29,759 pack which is a three phase inverter you can see the six mosfets here and i have 6 00:00:29,760 --> 00:00:34,710 my three phase connector here which is connected to my motor you can also 7 00:00:34,710 --> 00:00:40,79 notice here this is my position sensor cable and it's not connected today 8 00:00:40,79 --> 00:00:44,129 because I will demonstrate how to do field oriented control with sensorless 9 00:00:44,129 --> 00:00:48,120 let me start with the end and then i will show how I got there 10 00:00:48,120 --> 00:00:52,260 this is my simulink model that will generate code from it by pressing this 11 00:00:52,260 --> 00:00:57,930 button or pressing ctrl be on my keyboard you can follow the cogeneration 12 00:00:57,930 --> 00:01:03,329 steps on the main map window notice on the side i have these variables that 13 00:01:03,329 --> 00:01:07,920 have been automatically created when I opened my model i can set my cpu 14 00:01:07,920 --> 00:01:13,740 frequency and I can also set my PWM frequency i'm using a PWM frequency of 15 00:01:13,740 --> 00:01:19,259 20 kilohertz today this is also the frequency at which i'm going to run my 16 00:01:19,259 --> 00:01:24,450 current loop so i'm running a new current loop and refreshing my pwm duty 17 00:01:24,450 --> 00:01:30,960 cycles every 50 microseconds or 20 kilohertz i have an automated script 18 00:01:30,960 --> 00:01:36,600 that will load the generated executable on my targets so I just connected to my 19 00:01:36,600 --> 00:01:42,000 target loaded the code and if i go back to my camera I can see now my shaft is 20 00:01:42,000 --> 00:01:52,590 spinning now that my motor spinning i can open this host model that will read 21 00:01:52,590 --> 00:01:57,600 cereal messages coming from the USB cable connected to the launch pad so 22 00:01:57,600 --> 00:02:01,770 here i have three messages coming over the serial port i have the desired speed 23 00:02:01,770 --> 00:02:07,229 the actual speed and torque demand so if I around this model on the host it will 24 00:02:07,229 --> 00:02:12,790 get the serial messages and I can see in real time what's happening on my target 25 00:02:12,790 --> 00:02:19,30 using similar scopes this is my desired speed its per unit speed of 4.5 and this 26 00:02:19,30 --> 00:02:23,290 is the actual measured speed and this is my torque demand coming out of the speed 27 00:02:23,290 --> 00:02:28,870 control PID controller so if i try to slow down my shaft you can see that the 28 00:02:28,870 --> 00:02:32,739 actual speed is slowing down and I need torque to maintain the speed if I 29 00:02:32,739 --> 00:02:38,319 release the shaft then there's a spike of speed because it's catching up with 30 00:02:38,319 --> 00:02:42,879 the with the amount of torque that that was needed to maintain the speed before 31 00:02:42,879 --> 00:02:48,159 I can do that test again if i try to slow down my shaft I can see the Tour de 32 00:02:48,159 --> 00:02:53,560 man climbing up as soon as I release the shaft the amount of torque that built up 33 00:02:53,560 --> 00:02:56,769 created a spike positive speed 34 00:02:56,769 --> 00:03:00,159 how did I get there so first of all let's go through a few installation 35 00:03:00,159 --> 00:03:05,769 steps so i'm using matlab are 2013b everything i'm doing here can be done 36 00:03:05,769 --> 00:03:10,810 also with 13a i wouldn't recommend going to an earlier version of 13a because we 37 00:03:10,810 --> 00:03:16,780 made a lot of improvements for the people f-22 to 1713 a 430 is recommended 38 00:03:16,780 --> 00:03:22,449 version in 13b our support for tis e2000 processor is coming now 39 00:03:22,449 --> 00:03:26,769 yeah a support package that you have to download from the web so you have to 40 00:03:26,769 --> 00:03:33,729 type this target installer command that will help you through that process type 41 00:03:33,729 --> 00:03:38,620 this command it will open a GUI install the Support Package 40 i see two 42 00:03:38,620 --> 00:03:44,49 thousand processors and follow the steps to link also the third party files that 43 00:03:44,49 --> 00:03:48,669 you need ccsb five or before or even be three if that's what you prefer 44 00:03:49,329 --> 00:03:53,79 once this is done you can also run the check 45 00:03:53,769 --> 00:04:00,400 env set up Jackie and setup will help you validate your installation against 46 00:04:00,400 --> 00:04:09,489 the third party tools use check and be setup CCS that's for ccs version 3.3 if 47 00:04:09,489 --> 00:04:16,930 you want to use the five type CCS v5 then f 22 27 check make sure all these 48 00:04:16,930 --> 00:04:21,10 items are pointing to the recommended versions 49 00:04:21,10 --> 00:04:26,80 of the third body files there's another command that you can run also which is 50 00:04:26,80 --> 00:04:32,50 target updater this will make sure that you have the latest version of the 51 00:04:32,50 --> 00:04:38,530 support package and you can also use tools to validate that your third-party 52 00:04:38,530 --> 00:04:44,289 files are configured correctly the paths are correct and similarly knows where to 53 00:04:44,290 --> 00:04:50,140 find the third party tools so three commenced to remember targets installer 54 00:04:50,140 --> 00:04:58,599 target the later check env set up the model I'm using today has been posted on 55 00:04:58,600 --> 00:05:06,10 mass of central so you can find it by typing massive central.com and if you 56 00:05:06,10 --> 00:05:11,349 look at the file exchange and if you type motor control with ti launch bad 57 00:05:11,350 --> 00:05:18,490 you will find this motor control model so download this model it comes with the 58 00:05:18,490 --> 00:05:23,680 installation script or an init script that will initialize my variables each 59 00:05:23,680 --> 00:05:27,970 time i open the model so make sure you put all these files in the same folder 60 00:05:27,970 --> 00:05:33,550 and make this folder your present working directory in matlab so this will 61 00:05:33,550 --> 00:05:37,360 guarantee you have access to all the files there are two models provided 62 00:05:37,360 --> 00:05:43,930 actually there are three models provided 14 CCSD 5 which is the ERT version 14 63 00:05:43,930 --> 00:05:52,210 ccsb three and a host model that allows me to read serial data in real-time now 64 00:05:52,210 --> 00:05:57,310 let's look what's inside one of these models i'm going to open sea 28 to 27 65 00:05:57,310 --> 00:06:05,110 p.m. SFO see on the score ERT why on the score ERT because in my configset i am 66 00:06:05,110 --> 00:06:12,220 using ertiga TLC as my system target file for the older CCS version 3.3 we 67 00:06:12,220 --> 00:06:19,780 are using IE link brt which stands on the ID link automation that we have with 68 00:06:19,780 --> 00:06:27,280 CCS version 3.3 in 13b you can use the generic ERT TLC you can see that you can 69 00:06:27,280 --> 00:06:31,138 select your target hardware directly on the cogeneration pain 70 00:06:31,139 --> 00:06:37,830 so this is new and 30b and you can select your TI processor here and also 71 00:06:37,830 --> 00:06:44,248 your tool chain and that goes through a makefile approach and we also have an 72 00:06:44,249 --> 00:06:49,139 automatic download that you can select here on the corner target pain so let's 73 00:06:49,139 --> 00:06:50,520 look inside this model 74 00:06:50,520 --> 00:06:55,169 let's start with scheduling so for this model i have enabled sample time colors 75 00:06:55,169 --> 00:06:59,938 this is something you can do under display sometime colors and if i press 76 00:06:59,939 --> 00:07:05,669 this control ji get this legend which tells me every discreet an asynchronous 77 00:07:05,669 --> 00:07:11,490 time that I have in this model i have to its discreet x 110 million another one 78 00:07:11,490 --> 00:07:16,319 at $TIME hundred milliseconds and i have one asynchronous time the asynchronous 79 00:07:16,319 --> 00:07:23,610 time is this interrupt trigger here that is represented with this purple color so 80 00:07:23,610 --> 00:07:27,839 this is my current loop this interrupt is coming from the ADC so at the end of 81 00:07:27,839 --> 00:07:31,650 an ADC conversion i'm going to get an interrupt and this will trigger my 82 00:07:31,650 --> 00:07:36,60 current loop so this will guarantee you will start the algorithm as soon as I 83 00:07:36,60 --> 00:07:42,300 have fresh values coming from my analog-to-digital converter at 10 84 00:07:42,300 --> 00:07:46,949 milliseconds i will have my speed control group so i have one PID that 85 00:07:46,949 --> 00:07:51,750 reads the desired speed and the actual speed and the output of that PID is 86 00:07:51,750 --> 00:07:56,909 going to be my torque demand and that will feed the desired torque for the 87 00:07:56,909 --> 00:08:02,188 field oriented control algorithm i have another discreet sample time which is 88 00:08:02,189 --> 00:08:07,379 hundred Millie and that i'm using to send cereal messages to that hosts model 89 00:08:07,379 --> 00:08:11,729 i showed in the beginning of this example so now let's look at side the 90 00:08:11,729 --> 00:08:16,710 FOC algorithm subsystem so first of all you can notice the thick boundaries of 91 00:08:16,710 --> 00:08:21,000 this system this is typical to atomic subsystem and the difference with a 92 00:08:21,000 --> 00:08:25,620 regular subsystem is the fact that in the generated code the boundaries have a 93 00:08:25,620 --> 00:08:30,60 meaning you can define that meaning by right-clicking on the subsystem and go 94 00:08:30,60 --> 00:08:35,640 to block parameters and assigned to specific subsystem to a function or a 95 00:08:35,640 --> 00:08:39,120 separate see file given name to the function set that you 96 00:08:39,120 --> 00:08:43,679 the function to be reusable and not reusable in which case it may take 97 00:08:43,679 --> 00:08:48,870 adventure to stack or global variables and one thing which is important here is 98 00:08:48,870 --> 00:08:54,810 the fact that i can also define the memory section for this subsystem in 99 00:08:54,810 --> 00:08:59,790 this case this is my currently I want this to execute the fastest as possible 100 00:08:59,790 --> 00:09:05,790 so i'm putting this subsystem in code round funks this section has been 101 00:09:05,790 --> 00:09:11,910 defined in a package where we specify that it needs to be loaded in flash but 102 00:09:11,910 --> 00:09:16,439 run from Run so this is what this section will do it will reside in flash 103 00:09:16,440 --> 00:09:22,410 for standalone execution but it will run from round for faster execution and at 104 00:09:22,410 --> 00:09:26,490 any time we're going to copy the code from flashed around to make sure that it 105 00:09:26,490 --> 00:09:32,130 can execute from around the same can be done at the model level in the config 106 00:09:32,130 --> 00:09:43,200 sent i can go to code generation memory section and if i select this ti c2000 107 00:09:43,200 --> 00:09:47,940 demos package you can access it by clicking this refresh package button 108 00:09:47,940 --> 00:09:53,640 then you will see this tis demos package this has the definitions for a few 109 00:09:53,640 --> 00:09:57,449 sections that we have defined like this code round front section that I 110 00:09:57,450 --> 00:10:02,370 described earlier you can select execution this will apply at the model 111 00:10:02,370 --> 00:10:07,350 level and it will try to find all the functions that are being executed at 112 00:10:07,350 --> 00:10:15,360 every time step and you can place them in the code ramp on section so let's 113 00:10:15,360 --> 00:10:20,940 look inside the FOC algorithm subsystem so it is triggered on an ADC and dropped 114 00:10:20,940 --> 00:10:30,120 and it starts with to ADC readings abcb1 and ADCB 3 East correspond to ia and IB 115 00:10:30,120 --> 00:10:35,940 measurements on the hardware the second one which is b3 will trigger the 116 00:10:35,940 --> 00:10:43,950 interrupt so it's using SOC 1y the first one is using SOC 0 so this will happen 117 00:10:43,950 --> 00:10:49,649 after SOC 0 so this is the last one to convert at the end of the converter 118 00:10:49,649 --> 00:10:55,800 I want to post my interrupt and i'm using a DC and one so now an important 119 00:10:55,800 --> 00:11:01,889 thing to notice is the fact that we want to trigger the ADC based on a pwm event 120 00:11:01,889 --> 00:11:07,980 this will guarantee that we launched the conversion outside pwm transition this 121 00:11:07,980 --> 00:11:15,449 can be done using the pwm blog on this block we have an event trigger tab on 122 00:11:15,449 --> 00:11:21,929 the event trigger i enable ABC start of conversion for modulating based on a . 123 00:11:21,929 --> 00:11:27,629 event which means that when the counter reaches the . i will start by a DC 124 00:11:27,629 --> 00:11:33,449 conversion in the general tab i have sent my counter to be up and down so 125 00:11:33,449 --> 00:11:38,459 it's a symmetric moving up moving down counter if i launched my ADC conversion 126 00:11:38,459 --> 00:11:43,859 in the middle is very less likely to launch the conversion when there will be 127 00:11:43,860 --> 00:11:50,69 a pwm transition so I'm going to repeat this workflow pwm has a counter that's 128 00:11:50,69 --> 00:11:56,759 counting up and down when it reaches the . it launches the ADC conversion ADC 129 00:11:56,759 --> 00:12:02,129 converts ini be at the end of conversion i will trigger an interrupt 130 00:12:02,920 --> 00:12:08,290 now i need to react on this interrupt so using that interrupt block i can select 131 00:12:08,290 --> 00:12:14,529 cpu interrupt one pcie one so you have to read it vertically this is a matrix 132 00:12:14,529 --> 00:12:27,519 that you can access in the help so you can view the sec 1 so this is sequence 133 00:12:27,519 --> 00:12:38,889 or 14 the ADC which comes under cpu1 p IE 1 so here you can read pa1 cp1 this 134 00:12:38,889 --> 00:12:45,940 is why I selected cp1 p iu 107 entice priority is said 230 you have to 135 00:12:45,940 --> 00:12:50,79 remember that any synchronous task which is executed 136 00:12:50,79 --> 00:12:56,79 yeah a similar example time will execute at a priority of 40 so 30 means the 137 00:12:56,79 --> 00:13:00,699 priority of this interrupt is higher than the main simulink scheduler 138 00:13:00,699 --> 00:13:06,130 apprehension flag is set to one which means that a higher priority i don't 139 00:13:06,130 --> 00:13:11,709 have any in this model but let's imagine i had a higher priority it would be able 140 00:13:11,709 --> 00:13:19,810 to preempt this particular interrupt so you can notice that this line is dotted 141 00:13:19,810 --> 00:13:24,640 and the reason for this is because it's not data flow going through this line 142 00:13:24,640 --> 00:13:31,510 it's an event so this Hardware interrupt is going to trigger this subsystem you 143 00:13:31,510 --> 00:13:36,250 can also notice the presence of this gpio block the goal of this block is to 144 00:13:36,250 --> 00:13:43,540 enable the DRV 8301 chip and it's using gpio six so i need to set this to one 145 00:13:43,540 --> 00:13:47,19 otherwise my inverter is completely shut down 146 00:13:47,19 --> 00:13:52,269 so let's look inside the store control algorithm have two major subsystem the 147 00:13:52,269 --> 00:13:57,220 first one is input skating this will scale my ia and I be inputs into per 148 00:13:57,220 --> 00:14:01,810 unit value one block that's important to know this is that minus one on the 149 00:14:01,810 --> 00:14:08,589 hardware it's using an inverted often so to feed my main algorithm i need to do 150 00:14:08,589 --> 00:14:12,329 that- one to go back into a positive value for positive 151 00:14:12,329 --> 00:14:18,929 current on the second subsystem i generate myspace vectors and that is the 152 00:14:18,929 --> 00:14:24,600 typical field oriented control algorithm using the park park here i have my PID 153 00:14:24,600 --> 00:14:30,569 controllers for ID and IQ IQ is proportional to torque so it's coming 154 00:14:30,569 --> 00:14:35,998 from my speed controller and then i feed my inverse park space vector generator 155 00:14:35,999 --> 00:14:41,699 and this will feed the duty cycles of the pwm one thing good to notice here is 156 00:14:41,699 --> 00:14:48,689 that sliding mode observer so this will define my position so this allows sensor 157 00:14:48,689 --> 00:14:55,170 less field oriented control so it's using the alpha v beta alpha by beta and 158 00:14:55,170 --> 00:15:00,179 this gets out a theta which is a position signal from 0 to 1 it will be 159 00:15:00,179 --> 00:15:05,369 associated if we're moving at a constant speed and it's also outputting a speed 160 00:15:05,369 --> 00:15:10,529 signal we can look inside the sm 0 subsystem and you can see that it's a 161 00:15:10,529 --> 00:15:15,720 complicated subsystem you can take it out and tested in a unit test feed some 162 00:15:15,720 --> 00:15:22,169 stimulus signals and and watch the output of this subsystem going back to 163 00:15:22,169 --> 00:15:26,429 the root of the model i will show now the speed controller is a pretty simple 164 00:15:26,429 --> 00:15:31,649 system there's just a PID controller that takes a desired speed in this 165 00:15:31,649 --> 00:15:37,259 example is just a constant and feedback will be the measured speed and the 166 00:15:37,259 --> 00:15:42,179 output will be the desired torque so the PID will work on the Delta between the 167 00:15:42,179 --> 00:15:47,279 desired and the actual speed and we'll build up some desire toric depending on 168 00:15:47,279 --> 00:15:53,9 the air measured this subsystem also contains a serial monitor subsystem i'm 169 00:15:53,9 --> 00:15:57,779 going to be sending signals at 100 millisecond and i'm sending the speed 170 00:15:57,779 --> 00:16:01,949 desire d actual speed and the torque command basically the two inputs and the 171 00:16:01,949 --> 00:16:07,949 output of the PID controller i'm going to send the three variables every 172 00:16:07,949 --> 00:16:10,258 hundred millisecond 173 00:16:10,259 --> 00:16:15,569 now you're ready to generate code for this model the other model provided is 174 00:16:15,569 --> 00:16:22,738 the exact same model for ccs version 3.3 it's using the ID link automation so if 175 00:16:22,739 --> 00:16:28,410 i open the config set i will see that it is using ID and link on the score ERT 176 00:16:28,410 --> 00:16:36,988 TLC for the system target file if you want to learn more 177 00:16:37,589 --> 00:16:42,449 you have access to a list of examples if you open the simulink library browser 178 00:16:42,449 --> 00:16:47,998 under embedded code or support package for TI c2000 processor you can click on 179 00:16:47,999 --> 00:16:53,579 this blue example blog and it will point you to the list of examples provided for 180 00:16:53,579 --> 00:16:59,248 this support package i recommend going through the simulation of FOC using pmsf 181 00:16:59,249 --> 00:17:05,789 model this example shows how to test your algorithm using a plant model of 182 00:17:05,789 --> 00:17:10,109 the motor so you can work purely in simulation before you go to the embedded 183 00:17:10,109 --> 00:17:10,979 hardware 184 00:17:10,980 --> 00:17:16,230 this requires some power system so you can use motor simulation i also 185 00:17:16,230 --> 00:17:20,459 recommend going through the steps provided in the motor control example 186 00:17:20,459 --> 00:17:27,600 this is a step-by-step example on how to create the full algorithm and how to 187 00:17:27,599 --> 00:17:33,809 test components one-by-one also recommend going through this example 188 00:17:33,809 --> 00:17:39,000 which provides pretty much the same as what i described today on different 189 00:17:39,000 --> 00:17:45,450 types of hardware you can use the easy ESP from TI with the DM 550 we also have 190 00:17:45,450 --> 00:17:53,669 demos using the DRV 8312 kids with piccolo f28 or 35 + / 69 we have support 191 00:17:53,669 --> 00:17:58,679 for 28 33 528 Wade 2012 192 00:17:58,679 --> 00:18:03,450 thanks for watching I hope you will be successful running your simulation 193 00:18:03,450 --> 00:18:06,510 models and algorithms on the TI launchpad 194 00:18:06,510 --> 00:18:06,899 good luck