Thank you for uploading this piece of information! This is quite an ingenious solution.

Dayoung

Hello,

What do you mean by a uart1 ping-pong transmission?

If you are asking about using the UART with ping-pong DMA, this is not possible because the EFM8LB1 does not have a DMA.

Regards,
Dayoung

Hi Nikita,

I have another request. Can you please do the following:

1. Connect to the BGX board via USB
2. Open a terminal program such as TeraTerm
3. Run the 'get al' command
4. Send us the dump of the data from the previous command

Regards,
Dayoung

Hi Bas,

The reason your approach worked for Series 0/1 devices is because the Timer registers for Series 0/1 are clocked differently than in Series 2.

In Series 2, each Timer block has an enable register. Setting the enable register will clock the Timer registers with the SYNC characteristic. If these registers are accessed without the a clock, a bus fault will occur. In Series 0/1, there is no enable register, and so the entire Timer peripheral is clocked with with a CMU_ClockEnable function call. 

I agree with you that the TIMER_Reset function should check if it's enabled. I've created an internal request to have this added.

Regards,
Dayoung

Hi Mélissa,

Allocation of the default frame buffer shouldn't be root cause of the interrupt blockage, since it is used in other examples successfully. I'm not sure why that would be blocking the interrupts in the first place.

A simpler way to implement your solution would be to only use the glib example, and add "EMU_EnterEM3(true);" to the end of the GlibDemo() function. The device will enter EM0 -> update the screen -> enter EM3. When you press a button, the device will again enter EM0, and so on and so forth. I've attached the modified main.c file to this post. Let me know if this works for you.

Regards,
Dayoung

Hi Nikita,

Can you provide us a screenshot of the issue?

Regards,
Dayoung

Hi Nikita,

Did you download the app from the Play Store, or did you build it yourself from the source on Github?

Regards,
Dayoung

Hi Pietro,

You should be able to implement a circular buffer using two LDMA descriptors that are linked to each other (similar to ping-pong mode). But instead of assigning the src/dst pointers of each descriptor to two different RAM buffers, you can assign them to the same buffer with the second descriptor's pointers offset by the expected transfer size. This way, you don't spend double the memory. When the first descriptor finishes it's transfer, the LDMA ISR will trigger. In the ISR, you need to manually update the head/tail variables of the circular buffer and update the descriptor's src/dst pointers as well. You will need to create your own head/tail variables to keep track of the circular buffer pointers.

An example of this can be found in the AN0013 software. Note that this example is for the DMA, so you will need to port to LDMA.

Regards,
Dayoung

Hi Juan,

It looks like the CAN_ConfigureMessageObject function is overwriting the DATAVALID bit to 0, which is why the CAN_ReadMessage function you call afterwards returns false.

if you are basing your code off of the demo, the message objects and filters should be configured once during setup, not repeatedly alongside the CAN_ReadMessage function.

Regards,
Dayoung

Hi Juan,

It looks like the CAN_ConfigureMessageObject function is overwriting the DATAVALID bit to 0, which is why the CAN_ReadMessage function you call afterwards returns false.

if you are basing your code off of the demo, the message objects and filters should be configured once during setup, not repeatedly alongside the CAN_ReadMessage function.

Regards,
Dayoung