6 HPWM signal generation with PIC18F4431

[pxidr84]

Okay, I've a little bit modified the Bruce's program.
It runs well in ISIS Proteus with my defined duty cycles, and all of the 6 PWM outputs works.

But there is a little problem :


As you can see, U and W phases are working properly, but not the V phase (the low-side PWM is not inverted like the others).

Here's my modified code :

DEFINE LOADER_USED 1
DEFINE OSC 20
    
    ' At 20MHz, to figure a PWM frequency of 19.455kHz
    ' TPWM = time period of PWM frequency
    ' PTPER = 12-bit period register PTPERL and PTPERH
    ' PTMRPS = PWM time base prescaler
    ' 
    '        (PTPER+1)*PTMRPS  257
    ' TPWM = ----------------  =  ------------ = 0.0000514 
    '             Fosc/4            5000000
    '
    ' Frequency = 1/TPWM = 1/0.0000514 = 19.455kHz
    '
    ' PWM resolution (bits resolution for duty cycle)
    '
    '              log(20MHz/19.455kHz)    3.01
    ' Resolution = ------------------ = ----------- = 10 bits
    '                    .301              .301
      
    ' so we'll need a word sized var for Duty
uduty VAR WORD
vduty VAR WORD
wduty VAR WORD
   
    PORTB = 0               ' clear port latch
    TRISB = %11000000       ' PWM0,1,2,3,4,5 outputs
    
    TRISC = 2         ' RC1 = FLTA input (ground RC1 to halt PWM)
                            ' RC1 should be pulled high for normal PWM operation
                            ' when fault A is enabled.
    ' PCPWM init
    DTCON = %00000101       ' ~500nS dead-time (for complementary outputs only)
    PTCON0 = %00000100      ' 1:1 postscale, Fosc/4 1:1 prescale, free running mode
                            ' PTCON0 = %00000100 would give 19.45kHz/4
    PTPERL = 0              ' 
    PTPERH = 1              ' PTPER = $0100 or 256d for ~19.45kHz
	
    ' PWM4,5 independent, PWM0,1,2,3 complementary
    PWMCON0 = %01010100     ' PWM[5:0] outputs enabled
    PWMCON1 = 1             ' updates enabled, overrides sync w/timebase
    PTCON1 = %10000000      ' PWM time base is ON, counts up
    FLTCONFIG = %00000011   ' enable fault A, cycle-by-cycle mode


'Phases duty test
uduty = 800              ' ~50%
vduty = 500
wduty = 300
        
pwmlp: ' PWM update loop

'PWM U phase 
PDC0L = uduty.LowByte
PDC0H = uduty.HighByte

'PWM V phase
PDC1L = vduty.LowByte
PDC1H = vduty.HighByte

'PWM W phase  
PDC2L = wduty.LowByte
PDC2H = wduty.HighByte

goto pwmlp

Thanks in advance for any advice.

[HenrikOlsson]

Hi,
Are you sure it's not the W-phase output that isn't complementary?

Look at the PWMCON0 register, you have it set to %01010100 which means that PWM0 and PWM1 is in complentary mode while PWM2 (the W-phase according to your code) is in independant mode.

/Henrik.

[pxidr84]

Hi,
Are you sure it's not the W-phase output that isn't complementary?

Look at the PWMCON0 register, you have it set to %01010100 which means that PWM0 and PWM1 is in complentary mode while PWM2 (the W-phase according to your code) is in independant mode.

/Henrik.

Thanks a lot, you guided me on the right way.
I've read the datasheet, and especially the "PWMCON0" register description. I've modified the value to PWMCON0 = %01000000, and now the 3 phases are working very well (the low-side V phase is now inverted properly).

Now, I've to modify the duty according to a sinus function. I think about it.

[pxidr84]

A last question : I've doubts about the duty cycle. What's the exact range of the duty cycle?

If I define a duty cycle between 0-10, the simulator oscilloscope gives me weird things (the inverted signal "jumps")...
If I define a duty cycle above 1023, same thing...

The code used :

    ' At 20MHz, to figure a PWM frequency of 19.455kHz
    ' TPWM = time period of PWM frequency
    ' PTPER = 12-bit period register PTPERL and PTPERH
    ' PTMRPS = PWM time base prescaler
    ' 
    '        (PTPER+1)*PTMRPS  257
    ' TPWM = ----------------  =  ------------ = 0.0000514 
    '             Fosc/4            5000000
    '
    ' Frequency = 1/TPWM = 1/0.0000514 = 19.455kHz
    '
    ' PWM resolution (bits resolution for duty cycle)
    '
    '              log(20MHz/19.455kHz)    3.01
    ' Resolution = ------------------ = ----------- = 10 bits
    '                    .301              .301
 
'PIC initialization
DEFINE LOADER_USED 1
DEFINE OSC 40      

'Port registers configuration
PORTB = 0               ' Clear ports
TRISB = %11000000       ' PWM 0,1,2,3,4,5 outputs
TRISC = 2               ' RC1 = FLTA input (ground RC1 to halt PWM), pull-up resistor on RC1
                            
'PCPWM registers configuration
DTCON = %00001010       ' Deadtime (500ns)
PTCON0 = %00000100      ' 1:1 postscale, Fosc/4 1:1 prescale, free running mode
PTCON1 = %10000000      ' PWM time base is ON, counts up, 19.45kHz/4
PTPERL = 0              ' PWM low-side
PTPERH = 1              ' PWM high-side
	
'PWM registers configuration
PWMCON0 = %01000000     'PWM 0,1,2,3,4,5 set in pair mode
PWMCON1 = %00000001     'PWM timer sync configuration
FLTCONFIG = %00000011   'Fault configuration

'U, V and W duty variables
uduty VAR WORD
vduty VAR WORD
wduty VAR WORD

'Phases duty
uduty = 1         
vduty = 1000
wduty = 1000
        
pwmlp: ' PWM update loop

'PWM U phase 
PDC0L = uduty.LOWbyte
PDC0H = Uduty.HIGHByte

'PWM V phase
PDC1L = vduty.LOWbyte
PDC1H = vduty.HIGHByte

'PWM W phase  
PDC2L = wduty.LOWbyte
PDC2H = wduty.HIGHByte

goto pwmlp

[HenrikOlsson]

Hi,
The available resolution depends on the actual PWM frequency, you can see in the code how Bruce has calculated it for you. If you haven't changed the code the resolution is 10 bits (0-1023).

The PCPWM module have a configurable dead-time generator which inserts a deadtime between turning off the upper switch and turning on the lower switch. When aproaching the "ends" of the available resolution this can distort the output waveform and I suspect that is what you're seeing.

If you don't need deadtime you can turn it off. If you do need deadtime you should make sure that the dutycyle never aproaches the values where you're starting to see distortion. You usually can't use extreme dutycycles anyway due to the MOSFET/IGBT highside drivers need to refresh its bootstrap capacitor charge.

[pxidr84]

Okay, the code is now capable to calculate 3 sine waves (one for each phase).

I can modify the frequency (by increasing or decreasing freqdiv), but I've difficulties to set a phase angle (0-120-240° for U, V and W respectively) and I've difficulties to set a sinus amplitude (for exemple, if I want a 0-50% dutycycle, I will write uduty=uduty*0.5, but PBP does not support float maths).

Thanks for your help. I use Darrel Taylor's interrupt system for sine-wave generation loop.

'PIC initialization
DEFINE LOADER_USED 1
DEFINE OSC 40      


'DT interrupt system include
INCLUDE "DT_INTS-18.bas"     ' Base Interrupt System
INCLUDE "ReEnterPBP-18.bas"  ' Include if using PBP interrupts


'PWM calculation variables
uduty VAR WORD     'U phase duty
vduty VAR WORD     'V phase duty
wduty VAR WORD     'W phase duty
t VAR byte         'Incremental value


'Inverter U/F variables
freqdiv var WORD   'Frequency diviser
FREQdiv=0          'Range : 0-65535


'Port registers configuration
PORTB=%0          ' Clear ports
TRISB=%11000000   ' PWM 0,1,2,3,4,5 outputs

                            
'PCPWM registers configuration
DTCON=%110        ' Deadtime (600ns)
PTCON0=%100       ' 1:1 postscale, Fosc/4 1:1 prescale, free running mode
PTCON1=%10000000  ' PWM time base is ON, counts up, 19.45kHz/4
PTPERL=%0         ' PWM low-side
PTPERH=%1         ' PWM high-side

	
'PWM registers configuration
PWMCON0=%1000000  'PWM 0,1,2,3,4,5 set in pair mode
PWMCON1=%1        'PWM timer sync configuration


'Interrupt processor
ASM
INT_LIST  macro    ; IntSource, Label, Type, ResetFlag
        INT_Handler   TMR1_INT,  _pwmint,   PBP,  yes
    endm
    INT_CREATE     ; Creates the interrupt processor
ENDASM


'Timer configuration
T1CON=%10000001            ; Prescaler = 1:1, TMR1ON, 16-bit counter
@ INT_ENABLE  TMR1_INT     ; Enable Timer 1 interrupts


'Main loop
mainlp:

    pause 1000

goto mainlp

'PWM calculation and update interrupt
pwmint:
t=0
    
    for T=0 to 255 step 1

    'Frequency diviser
    pauseus freqdiv

    'PWM U phase calculation
    uduty=((sin(t)+128)*4)  
      
    'PWM V phase calculation
    vduty=((sin(t)+128)*4)
    
    'PWM W phase calculation
    wduty=((sin(t)+128)*4)

    'PWM U phase update
    PDC0L=uduty.LOWbyte
    PDC0H=Uduty.HIGHByte

    'PWM V phase update
    PDC1L=vduty.LOWbyte
    PDC1H=vduty.HIGHByte

    'PWM W phase update
    PDC2L=wduty.LOWbyte
    PDC2H=wduty.HIGHByte

    next t

@ INT_RETURN

[HenrikOlsson]

Hi,
How about:

'PWM U phase calculation
uduty=((sin(t)+128)*4)  
      
'PWM V phase calculation
vduty=((sin(t+85)+128)*4)
    
 'PWM W phase calculation
wduty=((sin(t+170)+128)*4

It's true that you can't mutliply by 0.5 but you can divide by 2. You can also use the ** operator to multiply by units of 1/65536 so instead of doing x=y*0.72 you do x=y**47186 (where 47186 comes from 65536*0.72)

/Henrik.