[Tinyos-devel] more bugging about current consumption on MicaZ in tinyos-2.x

[Philip Levis]

On Mon, 2006-07-17 at 13:18, Philip Levis wrote:
> On Wed, 2006-07-12 at 16:08, Razvan Musaloiu-E. wrote:
> > Hi!
> > 
> > A few weeks ago I wrote [1] on the tinyos-help list about some high
> > consumptions on MicaZ in tinyos-2.x. I rerun the tests using the beta2 and
> > the results I'm getting are still wrong. I attached a small application
> > that demonstrate this. Its based on the TestSleep [2] (which doesn't
> > compile) and the only thing it does is to call the McuSleep.sleep() in
> > Boot.booted(). The current value I'm getting is 0.14mA. Could someone else
> > run the program and confirm or disprove this problem? :-)
> 
> I tested with the app Null on a micaz and measured a current draw of 2
> uA. Can you verify whether you measure this (or a similar value,
> expected is <5uA and max is 10uA) as well?

Some further data. This is average/min/max current consumption, just
measured with a multimeter. So for apps such as the last one (a 60s
timer), the data does not capture sharp changes (e.g., the interrupt
handlers). But it does capture the general overhead (e.g., oscillator,
counter, etc).

Null
  Avg: 2uA
  Min: 2uA
  Max: 2uA

Blink (with no Leds, so a 4 Hz, 2 Hz and 1 Hz timer):
  Avg: 52  uA
  Min: 14  uA
  Max: 124 uA

Blink (with Leds):
  Avg: 3.6 mA
  Min: 11  uA
  Max: 7.1 mA

BaseStation (no packets received/transmitted, no Leds):
  Avg: 23.6 mA
  Min: 23.6 mA
  Max: 23.6 mA

60s Timer (app that does nothing except maintain a 1/60 Hz timer):
  Avg: 4 uA
  Min: 3 uA
  Max: 5 uA

The final app better measures the overhead on a very low duty cycle app:
it just measures the cost of maintaining a low-rate wakeup timer. 

For the really low numbers, I'm sure the particular atmega128L comes
into play. The mote I used seemed to have a reasonably good one.

Phil