SED drawing too much current after autojoin fails

RE: SED drawing too much current after autojoin fails

Edvin — Mon, 09 Dec 2024 11:42:21 GMT

1: Did you cut the solder bridge SB 40 on the DK before doing measurements?

Is this current consumption only related to the autojoin feature?

At this point, I am not sure what those spikes on your custom board are coming from, and I don't know how to, or if it is possible to reproduce them on a DK, or if it is some component on your custom board drawing these current spikes.

Best regards,

Edvin

RE: SED drawing too much current after autojoin fails

Mary — Fri, 06 Dec 2024 15:46:37 GMT

My custom board:

once it joins a network successfully and is commissioned, draws ~13 uA on average
If it does not join a network, then calls otPlatRadioSleep() draws ~590 uA (see PPK2 plot above)

nRF52840DK

If it does not join a network, then calls otPlatRadioSleep() draws 2.6 mA (see PPK2 plot above)

Mary

RE: SED drawing too much current after autojoin fails

Edvin — Fri, 06 Dec 2024 14:07:49 GMT

[quote user="Mary"]I was not able to reproduce the same current on the DK.[/quote]

What does that mean? The current consumption on the DK is lower? Does that mean that you may have some other components on your custom PCB that is drawing more power?

Best regards,

Edvin

RE: SED drawing too much current after autojoin fails

Mary — Wed, 04 Dec 2024 20:28:04 GMT

We are using custom hardware. Yes, I want it to reboot and retry joining every hour.

I was not able to reproduce the same current on the DK.

I modified the coap_client example to autojoin, and added otRadioPlatSleep() after running for 10 seconds.

After 10 seconds, the current is 2.6 mA.

Changes to prj.conf

# Same network Master Key for client and server
#CONFIG_OPENTHREAD_NETWORKKEY="00:11:22:33:44:55:66:77:88:99:aa:bb:cc:dd:ee:ff"

# Thread Joiner
CONFIG_OPENTHREAD_JOINER=y
CONFIG_OPENTHREAD_JOINER_AUTOSTART=y
CONFIG_OPENTHREAD_JOINER_PSKD="J01NUS"
CONFIG_OPENTHREAD_MANUAL_START=n
CONFIG_OPENTHREAD_NETWORKKEY=""

Code added to coap_client.c

int main(void)
{
	int ret;

	LOG_INF("Start CoAP-client sample");

	if (IS_ENABLED(CONFIG_RAM_POWER_DOWN_LIBRARY)) {
		power_down_unused_ram();
	}

	ret = dk_buttons_init(on_button_changed);
	if (ret) {
		LOG_ERR("Cannot init buttons (error: %d)", ret);
		return 0;
	}

	ret = dk_leds_init();
	if (ret) {
		LOG_ERR("Cannot init leds, (error: %d)", ret);
		return 0;
	}

#if CONFIG_BT_NUS
	struct bt_nus_cb nus_clbs = {
		.received = on_nus_received,
		.sent = NULL,
	};

	ret = ble_utils_init(&nus_clbs, on_ble_connect, on_ble_disconnect);
	if (ret) {
		LOG_ERR("Cannot init BLE utilities");
		return 0;
	}

#endif /* CONFIG_BT_NUS */

	coap_client_utils_init(on_ot_connect, on_ot_disconnect, on_mtd_mode_toggle);

	// My Test: added the next two lines
	k_msleep(10000);		
	otPlatRadioSleep(openthread_get_default_instance());

	return 0;
}

Mary

RE: SED drawing too much current after autojoin fails

Edvin — Mon, 25 Nov 2024 10:28:14 GMT

Hello,

sorry for the late reply. So you managed to get the current consumption quite a lot down, it seems. What hardware are you running this on? Is that current consumption possible to reproduce on a DK? If so, can you please upload your application here? (or a strip down version of the application that replicates this current pattern).

It is not an alternative to enter system off when it doesn't find a network? You want it to reboot (and I assume retry joining) every hour?

Best regards,

Edvin

RE: SED drawing too much current after autojoin fails

Mary — Tue, 19 Nov 2024 19:00:14 GMT

I changed the code to call otPlatRadioSleep() instead of otThreadSetLinkMode() and reduced the current quite a bit.

		while(otDatasetGetActive(instance, &dataset) != OT_ERROR_NONE)
		{
			wait++;
			if (wait < 60)				// 5 sec x 60 = 5 min
					{
#ifdef CONFIG_SERIAL				
				printk(".");			// wait for joining to complete, blink LED up to 30 seconds
#endif
				set_led(COLOR_PURPLE);	// blink purple, 5 sec
				k_msleep(100);
				set_led(COLOR_OFF);
				k_msleep(4900);
			}
			else if (wait > (12 * 60))	// 5 sec x 12 x 60 = 60 min, reboot in an hour
			{
				LOG_INF("Active Dataset not found.  Rebooting in 2 seconds\r\n");
				//k_msleep(2000);
				//NVIC_SystemReset();
				g_reboot = true;
			}
			else 
			{
				
				if (otPlatRadioGetState(instance) != OT_RADIO_STATE_SLEEP)
					otPlatRadioSleep(instance);
					
				// openthread_stop(instance);

				// if (otPlatRadioIsEnabled(instance))
				// 	otPlatRadioDisable(instance);	

				// otError error;
				// otLinkModeConfig mode;
				// struct openthread_context *context = openthread_get_default_context();

				// __ASSERT_NO_MSG(context != NULL);
				// openthread_api_mutex_lock(context);
				// mode = otThreadGetLinkMode(context->instance);
				// if (mode.mRxOnWhenIdle)											// disable receiver
				// {
				// 	mode.mRxOnWhenIdle = false;
				// 	error = otThreadSetLinkMode(context->instance, mode);
				// }
				// openthread_api_mutex_unlock(context);

				k_msleep(5000);

			}

			if (g_reboot)
				k_work_submit(&reboot_work);
		}

The current looks like this:

This looks like some peripheral waking up every ~4ms. Maybe a timer?

See also this similar issue where I found the suggestion to call otPlatRadioSleep():

github.com/.../10302

Mary