// AVR-Dx SIGROW - Signature Row // Read-only factory calibration data. Pre-loaded with values that produce // correct temperature readings with the D3AA firmware's conversion formula. import { CPU } from 'avr8js/cpu/cpu'; // Typical AVR32DD20 calibration values. // The firmware formula (from arch/avr32dd20.c mcu_temp_raw2cooked): // temp = (sigrow_offset << 4) - measurement // temp *= sigrow_slope // temp += 65536 / 8 // temp >>= 10 // result is Kelvin << 6 // // IMPORTANT: On AVR, the (sigrow_offset << 4) shift is done in 16-bit // arithmetic (int is 16-bit on AVR), so offset MUST be ≤ 0x0FFF (12-bit) // or the shift overflows. The datasheet specifies TEMPSENSE1 as a 12-bit value. // // With slope=0x036F (879), offset=0x0800 (2048): // offset << 4 = 32768 (fits in 16 bits) // At 25°C: ADC_acc16 ≈ 10548, result = 19081 → 23°C (±2°C rounding) // At 80°C: ADC_acc16 ≈ 6447, result = 22601 → 78°C // Range -40°C..125°C: ADC values 3092..15394, all in valid 16-bit range const DEFAULT_TEMPSENSE0 = 0x036F; // slope const DEFAULT_TEMPSENSE1 = 0x0800; // offset (12-bit) const TEMPSENSE0 = 0; const TEMPSENSE1 = 2; export class AVRDxSIGROW { constructor(cpu: CPU, base: number, readonly slope = DEFAULT_TEMPSENSE0, readonly offset = DEFAULT_TEMPSENSE1) { // TEMPSENSE0 (16-bit at 0x1104-0x1105) cpu.readHooks[base + TEMPSENSE0] = () => slope & 0xFF; cpu.readHooks[base + TEMPSENSE0 + 1] = () => (slope >> 8) & 0xFF; // TEMPSENSE1 (16-bit at 0x1106-0x1107) cpu.readHooks[base + TEMPSENSE1] = () => offset & 0xFF; cpu.readHooks[base + TEMPSENSE1 + 1] = () => (offset >> 8) & 0xFF; // Write hooks to prevent accidental writes cpu.writeHooks[base + TEMPSENSE0] = () => true; cpu.writeHooks[base + TEMPSENSE0 + 1] = () => true; cpu.writeHooks[base + TEMPSENSE1] = () => true; cpu.writeHooks[base + TEMPSENSE1 + 1] = () => true; } /** Compute the raw ADC result (16-bit accumulated) for a given temperature in Celsius */ tempCToRawADC(tempC: number): number { const tempK = tempC + 273.15; const kelvin6 = Math.round(tempK * 64); // Reverse the firmware formula: // kelvin6 = ((offset << 4) - measurement) * slope + 8192) >> 10 // kelvin6 << 10 = (offset << 4 - measurement) * slope + 8192 // (kelvin6 << 10) - 8192 = (offset << 4 - measurement) * slope // measurement = (offset << 4) - ((kelvin6 << 10) - 8192) / slope const measurement = (this.offset << 4) - ((kelvin6 << 10) - 8192) / this.slope; return Math.max(0, Math.min(0xFFFF, Math.round(measurement))); } }