Thread: Higher power consumption with SSED compared to SED

[BertL]

I am trying to get the lowest possible power consumption for my Thread device, so I am experimenting with different configurations.

Currently I'm testing with SSED (Synchronized Sleepy End Device) vs SED (Sleepy End Device). My assumption was that a SSED should be more power efficient.

My config is shown below. The only thing I change is CONFIG_CHIP_THREAD_SSED.

CONFIG_OPENTHREAD_NORDIC_LIBRARY_MTD=y
CONFIG_OPENTHREAD_MTD=y
CONFIG_OPENTHREAD_FTD=n
CONFIG_CHIP_ENABLE_ICD_SUPPORT=y
CONFIG_CHIP_THREAD_SSED=y

CONFIG_CHIP_ICD_SLOW_POLL_INTERVAL=2500
CONFIG_CHIP_ICD_SIT_SLOW_POLL_LIMIT=15000
CONFIG_CHIP_ICD_FAST_POLLING_INTERVAL=500
CONFIG_CHIP_ICD_IDLE_MODE_DURATION=120
CONFIG_CHIP_ICD_ACTIVE_MODE_THRESHOLD=0

My results are as follows (1 minute average):

SED average power consumption: 42.5 uA 
SSED average power consumption: 67.5 uA

So with these settings a SSED is consuming more than a SED. Does this make sense? Is a SSED mainly interesting for LIT devices?

[dejans]

Hi,

You can look at the SED and SSED comparison which provides information when to use SSED. You can also have a look at this power consumption reference. For Matter over Thread, you can look at SED configuration and SSED configuration. There is also general information about SED and SSED in the documentation.

Best regards,
Dejan

[BertL]

Thanks for the links! It is interesting that for my application I get different results than in the power consumption reference if I use the SSED settings. 

I'm using a Google Nest Display as thread border router for my tests. Can it be that it is using CSL parameters that are not working well for a SSED? Is it possible to debug this somewhere?

[dejans]

Hi,

You can check typical values for SCL parameters in SSED CSL configuration. You can also read about the SoC and power charge depenency in dependency between consumed electrical charge and SoC.

Best regards,
Dejan