Testing Current Consumption using datalogger....

[cortinas]

Hi folks

attached is the output of a datalogger which is capturing the current used by the ESP module (ESP001).

It logs the current at 100millisec intervals for about 3000 samples which is approx 5 mins

Config is using a ESP001, running 095 SDK and latest nodemcu/lua firmware. The data above was collected when the ESP is running as Station and connected to nearby access point....


The data shows quite clear repeated patterns --- but what are they ?

[jayc75]

Hi,
Thank you for sharing your insight, but I would suggest you build your test around the latest firmware V1.0 to get better support from Espressif...

thank you,

[cortinas]

Ok, no problem. I rebuilt with 0.9.6_b1 SDK and the grph looks basically the same

[baoshi]

I use a "uCurrent" (8kHz bandwidth) to measure the current consumption, checking output from an oscilloscope. ESP8266 seems to be sucking power on 100ms interval. I guess this is due to some network activity. If you did your log every 100ms, you could have missed something or the result is just alias.

[jayc75]

thanks again, but I was referring to SDK V1.0 found here: viewtopic.php?f=5&t=286
the 9xxx are old...

cheers,

[cortinas]

@jayc75
Ok, didn't realise there was a 1.0 !

@baoshi
... 'sucking power.... due network activity.....' thats exactly the point - but what is it ? Probes / Beacons etc ?

[tve]

Most access points send out a beacon every 100ms, receiving those is bound to take some cpu cycles...

[cortinas]

Hi all

Just to note that 100ms is the sample rate of the datalogger....so each of the very thin lines is a sample....

The steady state is about .0175 - roughly from the graph. The things I am trying to understand are:
- what are the peaks to 0.07 and 0.03 ---

IMHO - I believe that they are either probe requests or some other cpu/mcu timer kicking in...

Maybe someone from espressif can pipe in with some overview of when probes etc are generated

(The test is run in an anechoic chamber so no incoming beacons....also confirmed with a wifi sniffer trace....)

[tve]

Another interesting question is how much the power consumption depends on wifi activity. What I mean specifically is how much power is consumed by the reception of packets that are *not* destined to the esp8266. Clearly the radio has to receive and decode the packets in order to determine that they're destined to a different receiver.

[cortinas]

Ok, its becoming a but clearer now.....

Same test again ---- with a packet sniffer running (on separate machine) to see what is flying around....

As you can see from the wireshark trace, there are lots of wifi mgt packets (subtype 36) --- about 400 of these in the 2-3mins the test is running -- the destination is the wireless router that the ESP is connected to...

There are small numbers of some other QOS and data packets which I also need to investigate