InvenSense ICM-20948 DMP Flash error

[zq649]

Hi,
I am using ICM-20948 by TDK InvenSense, which has on board digital motion processing unit. My host processor is MediaTek MT-2523G.
InvenSense provides DMP firmware, that has to be programed on DMP to access it features. I am using I2C to communicate with ICM-20948. I am able to get the RAW accel data. But when I try to write the firmware bytes on DMP, somehow my writes are not getting to its memory.
I've seen on InvenSense developers form, most of the people are facing this issue. When I write the bytes to DMP memory, I get ACK from the device, but when I try to verify the DMP memory, I recieve garbage values.
I would be very thankful, if I get solution of this problem.

Furthermore, is there any activity recognition library, which take axis data as input and tells us user's activity.

[kriswiner]

Sounds like a question for Invensense. We use the EM7180+MPU9250 to identify sitting, standing, walking, etc. The algorithms have to be developed by the user though, we can't share ours.But it is straightforward.

…

On Fri, Mar 16, 2018 at 5:14 AM, qasimz ***@***.***> wrote: Hi, I am using ICM-20948 by TDK InvenSense, which has on board digital motion processing unit. My host processor is MediaTek MT-2523G. InvenSense provides DMP firmware, that has to be programed on DMP to access it features. I am using I2C to communicate with ICM-20948. I am able to get the RAW accel data. But when I try to write the firmware bytes on DMP, somehow my writes are not getting to its memory. I've seen on InvenSense developers form, most of the people are facing this issue. When I write the bytes to DMP memory, I get ACK from the device, but when I try to verify the DMP memory, I recieve garbage values. I would be very thankful, if I get solution of this problem. Furthermore, is there any activity recognition library, which take axis data as input and tells us user's activity. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#30>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGY1qjUUseW8UTshDO_UixPnrHAwsEWxks5te6yPgaJpZM4StsJn> .

[tatulea]

@QasimZ can you send me an email, please? I am interested in finding more about ICM-20948 ([email protected])

Thanks

[dtquang89]

Hi, do you have any solutions for this issue? Thank you.

[kriswiner]

No, try the Invensense Developer's forum.

…

On Tue, Apr 9, 2019 at 10:24 AM dtquang89 ***@***.***> wrote: Hi, do you have any solutions for this issue? Thank you. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#30 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGY1qrQkH9g0Qq8vsUX2dE3KSyS_cOvzks5vfMzggaJpZM4StsJn> .

[zq649]

Hi, do you have any solutions for this issue? Thank you.

which communication protocol are you using?

[dtquang89]

I’m using SPI.

[zq649]

In ICM20649Transport.h what values have you set for INV_MAX_SERIAL_READ and INV_MAX_SERIAL_WRITE?

I have set them to 1.

I am also posting my spi send and recv routine.

`static int spi_master_transfer_tx(void * context, uint8_t register_addr, const uint8_t * value, uint32_t len)
{

data[0] = register_addr & WRITE_BIT_MASK;
for ( int i = 0 ; i < len ; i++)
{
	data[i + 1] = *value;
	value++;
}

spi_send_and_receive_config.send_length = 1;
spi_send_and_receive_config.receive_length = len + spi_send_and_receive_config.send_length;
spi_send_and_receive_config.send_data = data;
spi_send_and_receive_config.receive_buffer = status_receive;
if(HAL_SPI_MASTER_STATUS_OK != hal_spi_master_send_and_receive_dma_blocking(ICM_SPI_CHANNEL,&spi_send_and_receive_config))
	return -1;
return 0;

}`

`static int spi_master_transfer_rx(void * context, uint8_t register_addr, uint8_t * value, uint32_t len)
{
//hal_gpt_delay_ms(200);
spi_send_and_receive_config.send_length = 1;
spi_send_and_receive_config.receive_length = len + spi_send_and_receive_config.send_length;
data[0] = register_addr | READ_BIT_MASK;
spi_send_and_receive_config.send_data = data;
spi_send_and_receive_config.receive_buffer = status_receive;
if(HAL_SPI_MASTER_STATUS_OK != hal_spi_master_send_and_receive_dma_blocking(ICM_SPI_CHANNEL,&spi_send_and_receive_config))
return -1;
//printf("\r\n %x \r\n",status_receive[1]);
for (int i = 0 ; i < len ; i++)
{
*value = status_receive[i+1];
value++;
//printf("\r\n status = 0x%02x \r\n",status_receive[i+1]);
}
return 0;

}`

[dtquang89]

Thank you for helping me. Actually changing the values doesn't solve the problem. I am using ICM with nRF52. I'm now sure why.

[zq649]

Which solution worked out for you?

…

---- On Tue, 16 Apr 2019 20:57:20 +0500 [email protected] wrote ---- Thank you for helping me. Actually changing the values doesn't solve the problem. I am using ICM with nRF52. I'm now sure why. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[dtquang89]

I haven't found a solution yet. I'm still looking for it.

[mpcrowe]

Just a note. I had a similar problem. The problem was solved by copying the firmware from const memory into RAM. The SAM3U4 we are using does not properly DMA to/from FLASH.