circuitpython/atmel-samd/common-hal/microcontroller/Processor.c

238 lines
10 KiB
C

/*
* This file is part of the MicroPython project, http://micropython.org/
*
* The MIT License (MIT)
*
* Copyright (c) 2017 Dan Halbert for Adafruit Industries
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to deal
* in the Software without restriction, including without limitation the rights
* to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
* copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
* THE SOFTWARE.
*/
/*
* Includes code from ASF sample code adc_temp.h and adc_temp.c,
* and so includes this license:
*
* Copyright (C) 2015 Atmel Corporation. All rights reserved.
*
* License
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* 1. Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright notice,
* this list of conditions and the following disclaimer in the documentation
* and/or other materials provided with the distribution.
*
* 3. The name of Atmel may not be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* 4. This software may only be redistributed and used in connection with an
* Atmel microcontroller product.
*
* THIS SOFTWARE IS PROVIDED BY ATMEL "AS IS" AND ANY EXPRESS OR IMPLIED
* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT ARE
* EXPRESSLY AND SPECIFICALLY DISCLAIMED. IN NO EVENT SHALL ATMEL BE LIABLE FOR
* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
* STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include "common-hal/microcontroller/Processor.h"
// Don't reorder these includes because they are dependencies of adc_feature.h.
// They should really be included by adc_feature.h.
#include <compiler.h>
#include "asf/sam0/drivers/system/clock/gclk.h"
#include "asf/sam0/utils/cmsis/samd21/include/component/adc.h"
#include "asf/sam0/drivers/adc/adc_sam_d_r/adc_feature.h"
#include "asf/sam0/drivers/adc/adc.h"
#define ADC_TEMP_SAMPLE_LENGTH 4
#define INT1V_VALUE_FLOAT 1.0
#define INT1V_DIVIDER_1000 1000.0
#define ADC_12BIT_FULL_SCALE_VALUE_FLOAT 4095.0
typedef struct nvm_calibration_data_t {
float tempR; // Production Room temperature
float tempH; // Production Hot temperature
float INT1VR; // Room temp 2's complement of the internal 1V reference value
float INT1VH; // Hot temp 2's complement of the internal 1V reference value
uint16_t ADCR; // Production Room temperature ADC value
uint16_t ADCH; // Production Hot temperature ADC value
float VADCR; // Room temperature ADC voltage
float VADCH; // Hot temperature ADC voltage
} nvm_calibration_data_t;
// Decimal to fraction conversion. (adapted from ASF sample).
STATIC float convert_dec_to_frac(uint8_t val) {
float float_val = (float)val;
if (val < 10) {
return (float_val/10.0);
} else if (val < 100) {
return (float_val/100.0);
} else {
return (float_val/1000.0);
}
}
STATIC void configure_adc_temp(struct adc_module *adc_instance) {
struct adc_config config_adc;
adc_get_config_defaults(&config_adc);
// The parameters chosen here are from the temperature example in:
// http://www.atmel.com/images/Atmel-42645-ADC-Configurations-with-Examples_ApplicationNote_AT11481.pdf
// That note also recommends in general:
// "Discard the first conversion result whenever there is a change
// in ADC configuration like voltage reference / ADC channel change."
config_adc.clock_prescaler = ADC_CLOCK_PRESCALER_DIV16;
config_adc.reference = ADC_REFERENCE_INT1V;
config_adc.positive_input = ADC_POSITIVE_INPUT_TEMP;
config_adc.negative_input = ADC_NEGATIVE_INPUT_GND;
config_adc.sample_length = ADC_TEMP_SAMPLE_LENGTH;
adc_init(adc_instance, ADC, &config_adc);
// Oversample and decimate. A higher samplenum produces a more stable result.
ADC->AVGCTRL.reg = ADC_AVGCTRL_ADJRES(2) | ADC_AVGCTRL_SAMPLENUM_4;
//ADC->AVGCTRL.reg = ADC_AVGCTRL_ADJRES(4) | ADC_AVGCTRL_SAMPLENUM_16;
}
// Extract the production calibration data information from NVM (adapted from ASF sample).
//
STATIC void load_calibration_data(nvm_calibration_data_t *cal) {
volatile uint32_t val1; /* Temperature Log Row Content first 32 bits */
volatile uint32_t val2; /* Temperature Log Row Content another 32 bits */
uint8_t room_temp_val_int; /* Integer part of room temperature in °C */
uint8_t room_temp_val_dec; /* Decimal part of room temperature in °C */
uint8_t hot_temp_val_int; /* Integer part of hot temperature in °C */
uint8_t hot_temp_val_dec; /* Decimal part of hot temperature in °C */
int8_t room_int1v_val; /* internal 1V reference drift at room temperature */
int8_t hot_int1v_val; /* internal 1V reference drift at hot temperature*/
uint32_t *temp_log_row_ptr = (uint32_t *)NVMCTRL_TEMP_LOG;
val1 = *temp_log_row_ptr;
temp_log_row_ptr++;
val2 = *temp_log_row_ptr;
room_temp_val_int = (uint8_t)((val1 & NVMCTRL_FUSES_ROOM_TEMP_VAL_INT_Msk) >> NVMCTRL_FUSES_ROOM_TEMP_VAL_INT_Pos);
room_temp_val_dec = (uint8_t)((val1 & NVMCTRL_FUSES_ROOM_TEMP_VAL_DEC_Msk) >> NVMCTRL_FUSES_ROOM_TEMP_VAL_DEC_Pos);
hot_temp_val_int = (uint8_t)((val1 & NVMCTRL_FUSES_HOT_TEMP_VAL_INT_Msk) >> NVMCTRL_FUSES_HOT_TEMP_VAL_INT_Pos);
hot_temp_val_dec = (uint8_t)((val1 & NVMCTRL_FUSES_HOT_TEMP_VAL_DEC_Msk) >> NVMCTRL_FUSES_HOT_TEMP_VAL_DEC_Pos);
room_int1v_val = (int8_t)((val1 & NVMCTRL_FUSES_ROOM_INT1V_VAL_Msk) >> NVMCTRL_FUSES_ROOM_INT1V_VAL_Pos);
hot_int1v_val = (int8_t)((val2 & NVMCTRL_FUSES_HOT_INT1V_VAL_Msk) >> NVMCTRL_FUSES_HOT_INT1V_VAL_Pos);
cal->ADCR = (uint16_t)((val2 & NVMCTRL_FUSES_ROOM_ADC_VAL_Msk) >> NVMCTRL_FUSES_ROOM_ADC_VAL_Pos);
cal->ADCH = (uint16_t)((val2 & NVMCTRL_FUSES_HOT_ADC_VAL_Msk) >> NVMCTRL_FUSES_HOT_ADC_VAL_Pos);
cal->tempR = room_temp_val_int + convert_dec_to_frac(room_temp_val_dec);
cal->tempH = hot_temp_val_int + convert_dec_to_frac(hot_temp_val_dec);
cal->INT1VR = 1 - ((float)room_int1v_val/INT1V_DIVIDER_1000);
cal->INT1VH = 1 - ((float)hot_int1v_val/INT1V_DIVIDER_1000);
cal->VADCR = ((float)cal->ADCR * cal->INT1VR)/ADC_12BIT_FULL_SCALE_VALUE_FLOAT;
cal->VADCH = ((float)cal->ADCH * cal->INT1VH)/ADC_12BIT_FULL_SCALE_VALUE_FLOAT;
}
/*
* Calculate fine temperature using Equation1 and Equation
* 1b as mentioned in data sheet section "Temperature Sensor Characteristics"
* of Electrical Characteristics. (adapted from ASF sample code).
*/
STATIC float calculate_temperature(uint16_t raw_code, nvm_calibration_data_t *cal)
{
float VADC; /* Voltage calculation using ADC result for Coarse Temp calculation */
float VADCM; /* Voltage calculation using ADC result for Fine Temp calculation. */
float INT1VM; /* Voltage calculation for reality INT1V value during the ADC conversion */
VADC = ((float)raw_code * INT1V_VALUE_FLOAT)/ADC_12BIT_FULL_SCALE_VALUE_FLOAT;
// Hopefully compiler will remove common subepxressions here.
/* Coarse Temp Calculation by assume INT1V=1V for this ADC conversion */
float coarse_temp = cal->tempR + (((cal->tempH - cal->tempR)/(cal->VADCH - cal->VADCR)) * (VADC - cal->VADCR));
/* Calculation to find the real INT1V value during the ADC conversion */
INT1VM = cal->INT1VR + (((cal->INT1VH - cal->INT1VR) * (coarse_temp - cal->tempR))/(cal->tempH - cal->tempR));
VADCM = ((float)raw_code * INT1VM)/ADC_12BIT_FULL_SCALE_VALUE_FLOAT;
/* Fine Temp Calculation by replace INT1V=1V by INT1V = INT1Vm for ADC conversion */
float fine_temp = cal->tempR + (((cal->tempH - cal->tempR)/(cal->VADCH - cal->VADCR)) * (VADCM - cal->VADCR));
return fine_temp;
}
// External interface.
//
float common_hal_mcu_processor_get_temperature(void) {
struct adc_module adc_instance_struct;
system_voltage_reference_enable(SYSTEM_VOLTAGE_REFERENCE_TEMPSENSE);
configure_adc_temp(&adc_instance_struct);
nvm_calibration_data_t nvm_calibration_data;
load_calibration_data(&nvm_calibration_data);
adc_enable(&adc_instance_struct);
uint16_t data;
enum status_code status;
// Read twice and discard first result, as recommended in section 14 of
// http://www.atmel.com/images/Atmel-42645-ADC-Configurations-with-Examples_ApplicationNote_AT11481.pdf
// "Discard the first conversion result whenever there is a change in ADC configuration
// like voltage reference / ADC channel change"
// Empirical observation shows the first reading is quite different than subsequent ones.
adc_start_conversion(&adc_instance_struct);
do {
status = adc_read(&adc_instance_struct, &data);
} while (status == STATUS_BUSY);
adc_start_conversion(&adc_instance_struct);
do {
status = adc_read(&adc_instance_struct, &data);
} while (status == STATUS_BUSY);
// Disable so that someone else can use the adc with different settings.
adc_disable(&adc_instance_struct);
return calculate_temperature(data, &nvm_calibration_data);
}
uint32_t common_hal_mcu_processor_get_frequency(void) {
return system_cpu_clock_get_hz();
}