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

[Philip Levis]

On Jul 31, 2007, at 6:36 AM, Murray, Ben wrote:

> Hi
> sorry to drag up an old thread but I'm getting some rather strange  
> results
> wrt micaz current draw vs input voltage for the null application  
> compiled in
> TinyOS-2.x (I have tried the srec posted by Phil and get the same  
> results).
> Using a bench PSU (100mA limit) and a Fluke 75 multimeter so should be
> resonably accurate down to ~10uA steps. Accurate enough to spot  
> some trends
> anyway...
>
> The current drawn by a mote running the null application seems to  
> depend on
> its input voltage and in more than just a I=V/R way. Results are  
> given below
> for 3 motes (M1 and M2 bought in 2004, M3 bought in 2007). Currents  
> are
> shown in uA and values of 5 are shown when the value was recorded  
> as "less
> than 10uA". Values ending in a 5 record when the multimeter noticeably
> flipped between two readings. These are the stable idle currents,  
> typically
> reached after a couple of seconds, occasionally taking longer to  
> settle.
>
> VCC   M1   M2   M3
> ------------------
> 2.1  430  430  440
> 2.2  520    5    5
> 2.3   10    5    5
> 2.4   15    5   20
> 2.5   40    5   40
> 2.6  125   15   50
> 2.7  600   30   60
> 2.8  350   90   60
> 2.9  380  140   70
> 3.0  500  200  100
>
> I have only been messing around with the motes and TinyOS for a  
> short while
> so am certainly not adept with programming them yet - is there an  
> easy way
> to tell the mode what power state to go into as all I seem to be  
> able to
> find is McuSleep.sleep() which will set it to what the mote thinks  
> is the
> correct sleep level. I'm having trouble locating in which component  
> the
> McuPowerOverride interface is used?

I believe that the only component which implements McuPowerOverride  
is the atmega128 timer stack. It uses the interface to make sure the  
MCU does not drop into a sleep state with a long wakeup if there is a  
timer that needs to fire very soon.

When I've measured sleep current, I've always used 3.0V as an input,  
and I've seen sleep currents down at the 4uA level. I'll check that  
this is still the case tomorrow (I'm not in the lab today).

Phil