作者：Terry Deng

本文檔概述了一種基于 SCI/UART 輸入信號，可以自動校準(zhǔn)本設(shè)備SCI/UART波特率的方法，該方法適用與所有第三代C2000芯片，比如F2807x/37x，F(xiàn)28004x，F(xiàn)28002x等等。

一 原理說明

假設(shè)有2塊電路板通過SCI進(jìn)行通信。“Transmitter”向“Receiver”發(fā)送未知波特率的數(shù)據(jù)，“ Receiver”則使用 eCAP 測量未知的波特率，然后修改其自身的波特率和“Transmitter”匹配。

下面款圖是一種情況，其中“Transmitter” 的波特率設(shè)置為 9889，而“Receiver”的初始波特率設(shè)置為 9601 ，相比之下“Receiver”的波特率為 -3% 偏差。 經(jīng)過算法的自動校準(zhǔn)以后，“Receiver”將會把自身波特率校正為與“Transmitter”相同的9889。

下面框圖則是另一種情況，假如“Receiver”和“Transmitter”的初始波特率都是9889，但“Receiver”的內(nèi)部晶振INTOSC有-3%的偏差。使用上述完全相同的方法原理和步驟，“Receiver”波特率設(shè)置將會從9889校準(zhǔn)成9601，這樣“Receiver”的波特率設(shè)置被自動校準(zhǔn)抵消內(nèi)部晶振的偏差。在測量實際信號時，“Receiver”輸出到“Transmitter”的信號會是正確的 9889 波特率。

二 Receiver 的校準(zhǔn)代碼

1. 初始化

需要配置以下模塊來校準(zhǔn)波特率：

• 時鐘：使用 INTOSC2 并選擇 100MHz 的 LSPCLK

#define DEVICE_SETCLOCK_CFG      (SYSCTL_OSCSRC_OSC2 | SYSCTL_IMULT(20) |  \
                                     SYSCTL_FMULT_NONE | SYSCTL_SYSDIV(2) |   \
                                     SYSCTL_PLL_ENABLE)
    //
    // Set up PLL control and clock dividers
    //
    SysCtl_setClock(DEVICE_SETCLOCK_CFG);

    //
    // Make sure the LSPCLK divider is set to the default (divide by 4)
    //
    SysCtl_setLowSpeedClock(SYSCTL_LSPCLK_PRESCALE_1);

• SCI 模塊：通訊數(shù)據(jù)使用，發(fā)出校準(zhǔn)以后的波形

    // Initialize SCIA and its FIFO.
    //
    SCI_performSoftwareReset(SCIA_BASE);

    //
    // Configure SCIA for communications.
    //
    SCI_setConfig(SCIA_BASE, DEVICE_LSPCLK_FREQ, TARGETBAUD, (SCI_CONFIG_WLEN_8 |
                                                        SCI_CONFIG_STOP_ONE |
                                                        SCI_CONFIG_PAR_NONE));
    SCI_resetChannels(SCIA_BASE);
    SCI_resetRxFIFO(SCIA_BASE);
    SCI_resetTxFIFO(SCIA_BASE);
    SCI_clearInterruptStatus(SCIA_BASE, SCI_INT_TXFF | SCI_INT_RXFF);
    SCI_enableFIFO(SCIA_BASE);
    SCI_enableModule(SCIA_BASE);
SCI_performSoftwareReset(SCIA_BASE);

• Xbar 輸入：將 GPIO28/SCI 內(nèi)部連接到 INPUTXBAR7 與 ECAP1 配合使用

//
    // Configure GPIO 28 as eCAP input
    //
    XBAR_setInputPin(XBAR_INPUT7, 28);

• ECAP 模塊：監(jiān)控接收到的 SCI 通信脈沖寬度

//
    // Disable ,clear all capture flags and interrupts
    //
    ECAP_disableInterrupt(ECAP1_BASE,
                          (ECAP_ISR_SOURCE_CAPTURE_EVENT_1  |
                           ECAP_ISR_SOURCE_CAPTURE_EVENT_2  |
                           ECAP_ISR_SOURCE_CAPTURE_EVENT_3  |
                           ECAP_ISR_SOURCE_CAPTURE_EVENT_4  |
                           ECAP_ISR_SOURCE_COUNTER_OVERFLOW |
                           ECAP_ISR_SOURCE_COUNTER_PERIOD   |
                           ECAP_ISR_SOURCE_COUNTER_COMPARE));
    ECAP_clearInterrupt(ECAP1_BASE,
                        (ECAP_ISR_SOURCE_CAPTURE_EVENT_1  |
                         ECAP_ISR_SOURCE_CAPTURE_EVENT_2  |
                         ECAP_ISR_SOURCE_CAPTURE_EVENT_3  |
                         ECAP_ISR_SOURCE_CAPTURE_EVENT_4  |
                         ECAP_ISR_SOURCE_COUNTER_OVERFLOW |
                         ECAP_ISR_SOURCE_COUNTER_PERIOD   |
                         ECAP_ISR_SOURCE_COUNTER_COMPARE));

    //
    // Disable CAP1-CAP4 register loads
    //
    ECAP_disableTimeStampCapture(ECAP1_BASE);

    //
    // Configure eCAP
    //    Enable capture mode.
    //    One shot mode, stop capture at event 4.
    //    Set polarity of the events to rising, falling, rising, falling edge.
    //    Set capture in time difference mode.
    //    Select input from XBAR7.
    //    Enable eCAP module.
    //    Enable interrupt.
    //
    ECAP_stopCounter(ECAP1_BASE);
    ECAP_enableCaptureMode(ECAP1_BASE);
    ECAP_setCaptureMode(ECAP1_BASE, ECAP_ONE_SHOT_CAPTURE_MODE, ECAP_EVENT_4);

    ECAP_setEventPolarity(ECAP1_BASE, ECAP_EVENT_1, ECAP_EVNT_FALLING_EDGE);
    ECAP_setEventPolarity(ECAP1_BASE, ECAP_EVENT_2, ECAP_EVNT_RISING_EDGE);
    ECAP_setEventPolarity(ECAP1_BASE, ECAP_EVENT_3, ECAP_EVNT_FALLING_EDGE);
    ECAP_setEventPolarity(ECAP1_BASE, ECAP_EVENT_4, ECAP_EVNT_RISING_EDGE);

    ECAP_enableCounterResetOnEvent(ECAP1_BASE, ECAP_EVENT_1);
    ECAP_enableCounterResetOnEvent(ECAP1_BASE, ECAP_EVENT_2);
    ECAP_enableCounterResetOnEvent(ECAP1_BASE, ECAP_EVENT_3);
    ECAP_enableCounterResetOnEvent(ECAP1_BASE, ECAP_EVENT_4);

    ECAP_selectECAPInput(ECAP1_BASE, ECAP_INPUT_INPUTXBAR7);

    ECAP_enableLoadCounter(ECAP1_BASE);
    ECAP_setSyncOutMode(ECAP1_BASE, ECAP_SYNC_OUT_DISABLED);
    ECAP_startCounter(ECAP1_BASE);
    ECAP_enableTimeStampCapture(ECAP1_BASE);
    ECAP_reArm(ECAP1_BASE);

    ECAP_enableInterrupt(ECAP1_BASE, ECAP_ISR_SOURCE_CAPTURE_EVENT_4);

2. 中斷

捕獲傳入 SCI 通信的脈沖寬度，每捕獲 4 次就中斷一次。 將這 4 個捕獲添加到陣列中。

__interrupt void ecap1ISR(void)
{
    if(stopCaptures==0)
    {
        //
        // Get the capture counts, interrupt every 4. Can be 1-bit or more wide.
        // add one to account for partial eCAP counts at higher baud rates
        // (e.g. count = 40, but if had higher resolution, this would be 40.5)
        //
        capCountArr[0] = 1+ECAP_getEventTimeStamp(ECAP1_BASE, ECAP_EVENT_1);
        capCountArr[1] = 1+ECAP_getEventTimeStamp(ECAP1_BASE, ECAP_EVENT_2);
        capCountArr[2] = 1+ECAP_getEventTimeStamp(ECAP1_BASE, ECAP_EVENT_3);
        capCountArr[3] = 1+ECAP_getEventTimeStamp(ECAP1_BASE, ECAP_EVENT_4);

        //
        // Add samples to a buffer. Get average baud and tune INTOSC if buffer filled.
        //
        capCountIter = 0;
        for (capCountIter=0; capCountIter<4; capCountIter++)
        {
            //
            // if we still have samples left to capture, add it to the samples array
            //
            if(samplesArrIter

捕獲陣列滿后，計算陣列的平均脈沖寬度 (也就是波特率)，并更新SCI波特率寄存器，使其盡可能接近計算的平均值。

//
    // Loop forever. Suspend or place breakpoints to observe the buffers.
    //
    for(;;)
    {
        //
        // Array is filled, begin tuning
        //
        if(stopCaptures==1)
        {
            //
            // Get an average baud rate from the array of samples
            //
            uint32_t avgBaud = getAverageBaud(samplesArr,NUMSAMPLES,TARGETBAUD);

            //
            // if the baud function returns the error code '0', then flag an error
            //
            if(avgBaud==0)
            {
                ESTOP0;
            }

            //
            // Update the device's baud rate to match the measured baud rate
            //
            SCI_setBaud(SCIA_BASE, DEVICE_LSPCLK_FREQ, avgBaud);

            //
            // (OPTIONAL) Continuously send data to SCITX once tuning
            // is complete for external observation (by logic analyzer or scope)
            //
            //unsigned char *msg;
            //while(1)
            //{
            //    msg = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\0";
            //    SCI_writeCharArray(SCIA_BASE, (uint16_t*)msg, 91);
            //}

            //
            // Wait for user to view the results in "Expressions" window
            //
            ESTOP0;

            //
            // If continuing, reset the array iterator and unlock the ISR for new captures
            //
            samplesArrIter=0;
            stopCaptures=0;
        }
}

4. 平均脈沖寬度

對于許多應(yīng)用的SCI 通信，傳輸?shù)臄?shù)據(jù) (例如 0xA5)是變化不固定的，因此SCI的高低電平脈沖寬度就是變化的。所以必須對樣本陣列進(jìn)行如下的預(yù)處理，然后才能計算平均脈沖寬度。

a) 丟棄大于 10 位寬的脈沖寬度 (丟棄空閑時間)

b) 將 n 位值除以 n

c) 對修改后的樣本數(shù)組進(jìn)行平均化

uint32_t getAverageBaud(volatile float arr[], int size, float targetBaudRate)
{
    //
    // clean up variable width array to single-bit-width array
    //
    uint16_t pass = arrTo1PulseWidth(arr, size, (float)DEVICE_SYSCLK_FREQ/targetBaudRate);

    //
    // pass only if enough good samples provided
    //
    if(pass == 0)
    {
        return 0;
    }

    //
    // convert 2-bit width, 3-bit width, etc. to 1-bit width values by dividing, and average these values.
    // skip unrelated values
    //
    float averageBitWidth = computeAvgWidth(arr, size);

    //
    // get the rounded baud rate from the average number of clocks and the sysclk frequency
    //
    return (uint32_t)(((float)DEVICE_SYSCLK_FREQ/(float)averageBitWidth)+0.5);
}

以下是平均脈寬計算的原理和代碼流程圖

三 結(jié)果

按照以下設(shè)置進(jìn)行測試，結(jié)果詳見表格，校準(zhǔn)以后的誤差從3% 改善為0.1%左右甚至更小。

1. “Transmitter”設(shè)置為正確的波特率 (我們嘗試匹配的波特率)
2. “Receiver”設(shè)置為錯誤波特率 (-3% 或 +3%)
3. “Receiver”運行校準(zhǔn)程序以匹配“Transmitter”