Low power Bluetooth advertising with STM32WB5MMG

[aidant]

The goal of my program is to:

          1. Set low power time that runs off HSI for 10 seconds
          2. Enter stop2 WFI 

          3. Wake up from low-power interrupt

          4. Advertise for 5 seconds  

          5. Stop advertising

          6. Repeat Steps

During stop2 at the beginning of program my system is using ~ 74uA (which seems high but is reasonable), and when I wake up to start advertising and call aci_gap_set_discoverable(). This raises the current to ~3.4mA while advertising but when I call aci_gap_set_non_discoverable() it only lowers the current draw to ~3mA.

I would expect the system to go back to the ~74uA it was drawing before advertising was turned on but it is not. 

I am able to connect to the chip over bluetooth successfully just want to know what needs to be done to get the power draw where I would expect 

Any help would be greatly appreciated!

Thank you.

[aidant]

---

@aidant wrote:

The goal of my program is to:

          1. Set low power time that runs off HSI for 10 seconds
          2. Enter stop2 WFI 

          3. Wake up from low-power interrupt

          4. Advertise for 5 seconds  

          5. Stop advertising

          6. Repeat Steps

During stop2 at the beginning of program my system is using ~ 74uA (which seems high but is reasonable), and when I wake up to start advertising and call aci_gap_set_discoverable(). This raises the current to ~3.4mA while advertising but when I call aci_gap_set_non_discoverable() it only lowers the current draw to ~3mA.

I would expect the system to go back to the ~74uA it was drawing before advertising was turned on but it is not. 

I am able to connect to the chip over bluetooth successfully just want to know what needs to be done to get the power draw where I would expect 

Any help would be greatly appreciated!

Thank you.

---

After some more debugging (using aGpioConfigList for cpu2 debugging ) I have seen that when `aci_gap_set_discoverable()` is called, the RNG_PROCESS is constantly toggling for the rest of the runtime of the application even when I call `aci_gap_set_non_discoverable()`. 

 

From the datasheet I know that the TRNG independent clock domain should use 3.80 uA / Mhz however I have seen through testing when I keep my logic the exact same (for the advertising on a period as described in the original post) and only comment out `MX_RNG_Init();` in main.c that my system will advertise and after advertisement go back to the expected idle current (which is the desired behavior I want).

 

This is not perfect though, due to the system not initializing the RNG peripheral each advertisement request will only send out a single advertisement rather then advertising (between min and max retry) until the timeout elapses. Because of this, I know that this temporary fix is not a permanent solution. I have tried to enable/disable the peripheral from cpu1 depending on if the system is advertising however ran into the issue where after cpu1 boots up cpu2 (by setting `PWR_CR4_C2BOOT` I appear to no longer have access to the peripheral (the RNG handle `hrng` instance struct becomes 0x0 for each member and in the SFRs the same is seen). The system will also crash when attempting to clear the bits in `HAL_RNG_DeInit()`. The code I was using to disable is below and this is called in adv_cancel which is generated by CubeMX.

```

while( LL_HSEM_1StepLock( HSEM, CFG_HW_RNG_SEMID ) );

 

HAL_RNG_DeInit(&hrng); //Switch off RNG IP

__HAL_RCC_RNG_CLK_DISABLE(); //Switch off RNG clock

 

LL_HSEM_ReleaseLock( HSEM, CFG_HW_RNG_SEMID, 0 );

```

 

Any help to determine the problem I am having or steps to relieve the issue would be greatly appreciated, thank you!

[aidant]

After a lot of debugging and scrubbing through old forum posts found a suggestion to use the HSI48 for the RNG peripheral rather then the LSI and this in conjuction with staying awake during advertising periods solved the issues I was having.