Clear ALevents for DC_sync0 and SM3 might have solved the uneven pulse train. Looking better now.
This commit is contained in:
@@ -9,7 +9,12 @@ private:
|
||||
volatile double_t actualPosition;
|
||||
volatile int32_t nSteps;
|
||||
volatile uint32_t timerPulseSteps;
|
||||
volatile float Tstart;
|
||||
|
||||
public:
|
||||
volatile float Tstartf; // Starting delay in secs
|
||||
volatile uint32_t Tstartu; // Starting delay in usecs
|
||||
private:
|
||||
public:
|
||||
const float maxAllowedFrequency = 100000; // 100 kHz for now
|
||||
HardwareTimer *pulseTimer;
|
||||
uint32_t pulseTimerChan;
|
||||
@@ -28,7 +33,7 @@ public:
|
||||
volatile float frequency;
|
||||
|
||||
static uint32_t sync0CycleTime; // Nominal EtherCAT cycle time
|
||||
volatile uint32_t lcncCycleTime; // Linuxcnc nominal cycle time (1 ms often)
|
||||
volatile float lcncCycleTime; // Linuxcnc nominal cycle time in sec (1 ms often)
|
||||
|
||||
StepGen2(TIM_TypeDef *Timer, uint32_t _timerChannel, PinName _stepPin, uint8_t _dirPin, void irq(void), TIM_TypeDef *Timer2, void irq2(void));
|
||||
|
||||
|
||||
@@ -21,49 +21,52 @@ StepGen2::StepGen2(TIM_TypeDef *Timer, uint32_t _timerChannel, PinName _stepPin,
|
||||
startTimer = new HardwareTimer(Timer2);
|
||||
startTimer->attachInterrupt(irq2);
|
||||
}
|
||||
|
||||
uint32_t cnt = 0;
|
||||
uint32_t StepGen2::handleStepper(void)
|
||||
{
|
||||
|
||||
if (!enabled)
|
||||
return updatePos(0);
|
||||
lcncCycleTime = StepGen2::sync0CycleTime;
|
||||
|
||||
lcncCycleTime = StepGen2::sync0CycleTime * 1.0e-6; // was usec, became sec
|
||||
commandedStepPosition = floor(commandedPosition * stepsPerMM); // Scale position to steps
|
||||
if (initialStepPosition == commandedStepPosition) // No movement
|
||||
return updatePos(1);
|
||||
|
||||
float approximateFrequency = fabs(initialStepPosition - commandedStepPosition) // We must take at least one step
|
||||
/ (float)lcncCycleTime; // from here on
|
||||
if (approximateFrequency > maxAllowedFrequency) // Stay on this position
|
||||
return 1;
|
||||
// digitalWrite(dirPin, cnt++ % 2);
|
||||
float approximateFrequency = fabs(initialStepPosition - commandedStepPosition) // We must take at least one step
|
||||
/ lcncCycleTime; // from here on
|
||||
// if (approximateFrequency > maxAllowedFrequency) // Stay on this position
|
||||
// return 1;
|
||||
|
||||
float kTRAJ = (commandedPosition - initialPosition) / float(lcncCycleTime); // Straight line equation
|
||||
float kTRAJ = (commandedPosition - initialPosition) / lcncCycleTime; // Straight line equation
|
||||
float mTRAJ = initialPosition; // position = kTRAJ x time + mTRAJ
|
||||
// Operating on incoming positions (not steps)
|
||||
if (fabs(kTRAJ * lcncCycleTime * stepsPerMM) < 0.01) // Very flat slope
|
||||
{ //
|
||||
Tstart = 0.5 * lcncCycleTime; // Just take a step in the middle of the cycle
|
||||
Tstartf = 0.5 * lcncCycleTime; // Just take a step in the middle of the cycle
|
||||
frequency = 10000; // At some suitable frequency
|
||||
nSteps = kTRAJ > 0 ? 1 : -1; // Take only one step
|
||||
}
|
||||
else // Regular step train, up or down
|
||||
{
|
||||
if (kTRAJ > 0)
|
||||
Tstart = (ceil(initialPosition * stepsPerMM) - mTRAJ) / kTRAJ;
|
||||
Tstartf = (ceil(initialPosition * stepsPerMM) / stepsPerMM - mTRAJ) / kTRAJ;
|
||||
else
|
||||
Tstart = (floor(initialPosition * stepsPerMM) - mTRAJ) / kTRAJ;
|
||||
frequency = kTRAJ * stepsPerMM;
|
||||
Tstartf = (floor(initialPosition * stepsPerMM) / stepsPerMM - mTRAJ) / kTRAJ;
|
||||
frequency = fabs(kTRAJ * stepsPerMM);
|
||||
nSteps = commandedStepPosition - initialStepPosition; // sign(nSteps) = direction.
|
||||
}
|
||||
updatePos(5);
|
||||
Tstartu = Tstartf * 1e6; // Was secs, now usecs
|
||||
|
||||
startTimer->setOverflow(Tstart + Tjitter, MICROSEC_FORMAT); // All handled by irqs
|
||||
startTimer->setOverflow(Tstartu + Tjitter, MICROSEC_FORMAT); // All handled by irqs
|
||||
startTimer->resume();
|
||||
return 1;
|
||||
}
|
||||
void StepGen2::startTimerCB()
|
||||
{
|
||||
startTimer->pause(); // Once is enough.
|
||||
digitalWrite(dirPin, nSteps > 0 ? 1 : -1);
|
||||
// digitalWrite(dirPin, nSteps > 0 ? 1 : -1);
|
||||
timerPulseSteps = abs(nSteps);
|
||||
pulseTimer->setMode(pulseTimerChan, TIMER_OUTPUT_COMPARE_PWM2, stepPin);
|
||||
pulseTimer->setOverflow(uint32_t(frequency), HERTZ_FORMAT);
|
||||
|
||||
@@ -25,7 +25,8 @@ StepGen2 Step(TIM1, 4, PA_11, PA12, pulseTimerCallback, TIM10, startTimerCallbac
|
||||
void pulseTimerCallback(void) { Step.pulseTimerCB(); }
|
||||
void startTimerCallback(void) { Step.startTimerCB(); }
|
||||
CircularBuffer<uint32_t, 200> Tim;
|
||||
volatile uint64_t nowTime = 0, thenTime = 0;
|
||||
volatile uint64_t irqTime = 0, thenTime = 0;
|
||||
int64_t extendTime(uint32_t in); // Extend from 32-bit to 64-bit precision
|
||||
|
||||
void cb_set_outputs(void) // Master outputs gets here, slave inputs, first operation
|
||||
{
|
||||
@@ -38,8 +39,11 @@ void handleStepper(void)
|
||||
Step.enabled = true;
|
||||
Step.commandedPosition = Obj.StepGenIn1.CommandedPosition;
|
||||
Obj.StepGenOut1.ActualPosition = Step.commandedPosition;
|
||||
|
||||
Step.stepsPerMM = Obj.StepGenIn1.StepsPerMM;
|
||||
Step.stepsPerMM = 4000;
|
||||
Step.handleStepper();
|
||||
|
||||
Obj.StepGenOut2.ActualPosition = Obj.StepGenIn2.CommandedPosition;
|
||||
}
|
||||
|
||||
void cb_get_inputs(void) // Set Master inputs, slave outputs, last operation
|
||||
@@ -49,7 +53,7 @@ void cb_get_inputs(void) // Set Master inputs, slave outputs, last operation
|
||||
Obj.EncFrequency = Encoder1.frequency(ESCvar.Time);
|
||||
Obj.IndexByte = Encoder1.getIndexState();
|
||||
|
||||
uint32_t dTim = nowTime - thenTime; // Debug. Getting jitter over the last 200 milliseconds
|
||||
uint32_t dTim = irqTime - thenTime; // Debug. Getting jitter over the last 200 milliseconds
|
||||
Tim.push(dTim);
|
||||
uint32_t max_Tim = 0, min_Tim = UINT32_MAX;
|
||||
for (decltype(Tim)::index_t i = 0; i < Tim.size(); i++)
|
||||
@@ -60,8 +64,9 @@ void cb_get_inputs(void) // Set Master inputs, slave outputs, last operation
|
||||
if (aTim < min_Tim)
|
||||
min_Tim = aTim;
|
||||
}
|
||||
thenTime = nowTime;
|
||||
thenTime = irqTime;
|
||||
Obj.DiffT = max_Tim - min_Tim; // Debug
|
||||
Obj.DiffT = ESCvar.ALevent;
|
||||
}
|
||||
|
||||
void ESC_interrupt_enable(uint32_t mask);
|
||||
@@ -100,22 +105,27 @@ void setup(void)
|
||||
|
||||
void loop(void)
|
||||
{
|
||||
uint32_t dTime;
|
||||
if (serveIRQ)
|
||||
{
|
||||
nowTime = micros();
|
||||
CC_ATOMIC_SET(ESCvar.ALevent, ESC_ALeventread());
|
||||
DIG_process(DIG_PROCESS_WD_FLAG | DIG_PROCESS_OUTPUTS_FLAG |
|
||||
DIG_PROCESS_APP_HOOK_FLAG | DIG_PROCESS_INPUTS_FLAG);
|
||||
serveIRQ = 0;
|
||||
ESCvar.PrevTime = ESCvar.Time;
|
||||
}
|
||||
uint32_t dTime = micros() - nowTime;
|
||||
dTime = micros() - irqTime;
|
||||
if ((dTime > 200 && dTime < 500) || dTime > 1500) // Don't run ecat_slv_poll when expecting to serve interrupt
|
||||
ecat_slv_poll();
|
||||
}
|
||||
|
||||
volatile uint32_t cmt;
|
||||
void sync0Handler(void)
|
||||
{
|
||||
uint32_t lTime;
|
||||
irqTime = micros();
|
||||
serveIRQ = 1;
|
||||
ESC_read(ESCREG_LOCALTIME, (void *)&lTime, sizeof(lTime)); // Careful! Reads and writes update ALevent also.
|
||||
digitalWrite(Step.dirPin, cmt++ % 2);
|
||||
}
|
||||
|
||||
void ESC_interrupt_enable(uint32_t mask)
|
||||
@@ -125,6 +135,8 @@ void ESC_interrupt_enable(uint32_t mask)
|
||||
uint32_t user_int_mask = ESCREG_ALEVENT_SM2; // Only SM2
|
||||
if (mask & user_int_mask)
|
||||
{
|
||||
ESC_ALeventmaskwrite(ESC_ALeventmaskread() & ~(ESCREG_ALEVENT_DC_SYNC0 | ESCREG_ALEVENT_SM3));
|
||||
|
||||
ESC_ALeventmaskwrite(ESC_ALeventmaskread() | (mask & user_int_mask));
|
||||
attachInterrupt(digitalPinToInterrupt(PC3), sync0Handler, RISING);
|
||||
|
||||
@@ -165,3 +177,23 @@ uint16_t dc_checker(void)
|
||||
StepGen2::sync0CycleTime = ESC_SYNC0cycletime() / 1000; // usecs
|
||||
return 0;
|
||||
}
|
||||
|
||||
#define ONE_PERIOD UINT32_MAX
|
||||
#define HALF_PERIOD (UINT32_MAX >> 1)
|
||||
static int64_t previousTimeValue = 0;
|
||||
|
||||
int64_t extendTime(uint32_t in) // Extend from 32-bit to 64-bit precision
|
||||
{
|
||||
int64_t c64 = (int64_t)in - HALF_PERIOD; // remove half period to determine (+/-) sign of the wrap
|
||||
int64_t dif = (c64 - previousTimeValue); // core concept: prev + (current - prev) = current
|
||||
|
||||
// wrap difference from -HALF_PERIOD to HALF_PERIOD. modulo prevents differences after the wrap from having an incorrect result
|
||||
int64_t mod_dif = ((dif + HALF_PERIOD) % ONE_PERIOD) - HALF_PERIOD;
|
||||
if (dif < -HALF_PERIOD)
|
||||
mod_dif += ONE_PERIOD; // account for mod of negative number behavior in C
|
||||
|
||||
int64_t unwrapped = previousTimeValue + mod_dif;
|
||||
previousTimeValue = unwrapped; // load previous value
|
||||
|
||||
return unwrapped + HALF_PERIOD; // remove the shift we applied at the beginning, and return
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user