Merge pull request #3327 from opentx/projectkk2glider/new_jitter_filter_for_next2

Projectkk2glider/new jitter filter for next2
This commit is contained in:
Bertrand Songis
2016-02-28 19:58:45 +01:00
8 changed files with 220 additions and 88 deletions
+38 -16
View File
@@ -133,11 +133,11 @@ void adcInit()
#define SPI_S1 4
#define SPI_6POS 5
#define SPI_S2 6
#define SPI_S3 7
#define SPI_S4 8
#define SPI_LS 7
#define SPI_RS 8
#define SPI_TX_VOLTAGE 9
#define SPI_RS 10
#define SPI_LS 11
#define SPI_L2 10
#define SPI_L1 11
const uint16_t adcCommands[MOUSE1+2] =
{
@@ -148,13 +148,13 @@ const uint16_t adcCommands[MOUSE1+2] =
MANUAL_MODE | ( SPI_S1 << 7 ),
MANUAL_MODE | ( SPI_6POS << 7 ),
MANUAL_MODE | ( SPI_S2 << 7 ),
MANUAL_MODE | ( SPI_L1 << 7 ),
MANUAL_MODE | ( SPI_L2 << 7 ),
MANUAL_MODE | ( SPI_LS << 7 ),
MANUAL_MODE | ( SPI_RS << 7 ),
MANUAL_MODE | ( SPI_S3 << 7 ),
MANUAL_MODE | ( SPI_S4 << 7 ),
MANUAL_MODE | ( SPI_TX_VOLTAGE << 7 ),
MANUAL_MODE | ( 0 << 7 ),
MANUAL_MODE | ( 0 << 7 )
MANUAL_MODE | ( 0 << 7 ), // small joystick left/right
MANUAL_MODE | ( 0 << 7 ) // small joystick up/down
};
void adcRead()
@@ -170,22 +170,44 @@ void adcRead()
DMA2_Stream1->CR |= DMA_SxCR_EN; // Enable DMA
ADC3->CR2 |= (uint32_t)ADC_CR2_SWSTART;
// A dummy command to get things started
// (because the sampled data is lagging behind for two command cycles)
ADC_CS_LOW();
delay_01us(1);
SPIx_ReadWriteByte(*command++); // Discard
ADC_CS_HIGH();
delay_01us(1);
ADC_CS_LOW();
delay_01us(1);
SPIx_ReadWriteByte(*command++); // Discard
SPIx_ReadWriteByte(*command); // still sampling the old MUX channel
ADC_CS_HIGH();
delay_01us(1);
for (uint32_t adcIndex=0; adcIndex<MOUSE1; adcIndex++) {
// MUX is changed to the channel that was sent in the previous command
// but the data is from the old MUX position
ADC_CS_LOW();
delay_01us(1);
adcValues[adcIndex] = (0x0fff & SPIx_ReadWriteByte(*command++));
SPIx_ReadWriteByte(*command);
ADC_CS_HIGH();
delay_01us(1);
// This delay is to allow charging of ADC input capacitor
// after the MUX changes from one channel to the other.
// It was determined experimentally. Biggest problem seems to be
// the cross-talk between A4:S1 and A5:MULTIPOS. Changing S1 for one extreme
// to the other resulted in A5 change of:
//
// delay value A5 change
// 0 76
// 100 26
// 300 13
// 500 undetectable
delay_01us(500);
++command; // move to the next channel
// First time the wanted ADC channel is actually sampled.
// This is delayed to allow MUX to settle (high input impedance)
// Also the MUX channel is set to the next channel here,
// but the actual MUX change will happen in the next round
ADC_CS_LOW();
delay_01us(1);
adcValues[adcIndex] = (0x0fff & SPIx_ReadWriteByte(*command));
ADC_CS_HIGH();
delay_01us(1);
}