MISO lines skipping a bit on nrf52840dk using the stm lsm6dsv16x.

[acand001]

Hey everyone. I am having an issue reading the WHO_AM_I register from the stm lsm6dsv16x. 
I am using the spi1 configuration as shown in the academy classes. I have very short cables. 

The issue I get is. I am constantly reading 0x60, but I should be reading 0x70. 
I also see people have similar issues from this stm link
community.st.com/.../810200
I tested in multiple configurations in Mode 0 and 3 ( supported only by lsm6dsv16x). But for some reason I could never get the frist bit correct. I am following similar code you guys have for the spi read on the academy course, but still get 0x60. 

the boards I am using is the lsm6dsv16x sparkfun breakout board( I made sure it is configured for spi) and the nrf52840dk

&spi2 {
    compatible = "nordic,nrf-spim";
    status = "okay";
    cs-gpios = <&gpio0 30 GPIO_ACTIVE_LOW>;
    pinctrl-0     = <&spi2_default>;
    pinctrl-1     = <&spi2_sleep>;
    pinctrl-names = "default", "sleep";

    lsm6dsv16x: lsm6dsv16x@0 {
        status = "okay";
        compatible = "zephyr,custom-lsm6dsv16x";
        reg = <0>;      
        spi-cpol;
        spi-cpha;              
        spi-max-frequency = <4000000>;
        int1-gpios = <&gpio1 9  (GPIO_PULL_DOWN | GPIO_ACTIVE_HIGH)>;
        int2-gpios = <&gpio0 3  (GPIO_PULL_DOWN | GPIO_ACTIVE_HIGH)>;
    };
};
&{/aliases} {
    imu0 = &lsm6dsv16x;   /* flip to &imu_icm to build with the other driver */


};

&pinctrl {
    i2c0_default: i2c0_default {
        group1 { psels = <NRF_PSEL(TWIM_SCL, 0, 27)>,
                         <NRF_PSEL(TWIM_SDA, 0, 5)>; };
    };
    i2c0_sleep: i2c0_sleep {
        group1 { psels = <NRF_PSEL(TWIM_SCL, 0, 27)>,
                         <NRF_PSEL(TWIM_SDA, 0, 5)>;
                 low-power-enable; };
    };

    spi2_default: spi2_default {
        group1 { psels = <NRF_PSEL(SPIM_SCK, 0, 28)>,
                         <NRF_PSEL(SPIM_MOSI, 0, 29)>,
                         <NRF_PSEL(SPIM_MISO, 0, 31)>;
                         };
    };
    spi2_sleep: spi2_sleep {
        group1 { psels = <NRF_PSEL(SPIM_SCK, 0, 28)>,
                         <NRF_PSEL(SPIM_MOSI, 0, 29)>,
                         <NRF_PSEL(SPIM_MISO, 0, 31)>;
                 low-power-enable; 
                };
    };
};

int32_t platform_read(void *handle, uint8_t reg, uint8_t *bufp, uint16_t len)
{
    const struct lsm6dsv16x_config *cfg = handle;

    uint8_t tx_buffer = reg | 0x80;  // Set MSB to indicate read
    uint8_t rx_buffer[len + 1];      // Temporary buffer to receive extra byte

    struct spi_buf tx_bufs[] = {
        { .buf = &tx_buffer, .len = 1 },
    };
    struct spi_buf rx_bufs[] = {
        { .buf = rx_buffer, .len = len + 1 },
    };

    struct spi_buf_set tx_set = { .buffers = tx_bufs, .count = 1 };
    struct spi_buf_set rx_set = { .buffers = rx_bufs, .count = 1 };

    int ret = spi_transceive_dt(&cfg->spi, &tx_set, &rx_set);
    if (ret < 0) {
        return ret;
    }

    memcpy(bufp, &rx_buffer[1], len);  // Skip dummy byte
    return 0;
}

[hmolesworth]

Often traces such as those included in the post are misleading as they are not true signals but instead interpreted signals displayed as binary on a logic analyser  when in fact they are sloppy-edge signals with a finite rise- and fall-time, not square at all. Should an analogue display mode or better still an analogue oscilloscope be available this would become apparent. Worse, the logic analyser may not use the same voltage thresholds as the nRF52840 to determine the edge locations.

However, no matter; the usual fix is to start by boosting the drive strength for /CS, SCK and MOSI to H0H1 (High-Drive) to get sharper edges (MISO doesn't matter):

    spi2_default: spi2_default {
        group1 { psels = <NRF_PSEL(SPIM_SCK, 0, 28)>,
                         <NRF_PSEL(SPIM_MOSI, 0, 29)>,
                         <NRF_PSEL(SPIM_MISO, 0, 31)>;
                 nordic,drive-mode = <NRF_DRIVE_H0H1>;
		};

nrf-pinctrl.html