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

[Philip Levis]

On Jul 19, 2006, at 9:31 AM, Razvan Musaloiu-E. wrote:

> Hi!
>
> On Wed, 19 Jul 2006, Eric Weddington wrote:
>
>> Philip Levis wrote:
>>
>>>
>>> It's very interesting, though, that the deep sleep state of the
>>> atmega128 is in practice so much lower than what the data sheets   
>>> claim
>>> (and what a lot of papers have used as the basis for  estimates).  
>>> It's
>>> also much lower than what Joe reported measuring  (e.g., we're  
>>> 40% of
>>> what he said is the lowest he's heard reported).
>
> The specs for Atmega128L ([1] page 323) indicate a typical <5uA in
> Power Down for Vcc = 3V.

Yeah. Joe's comment was that 5 is the bottom of the range, though.  
Also, if you look at the Telos paper, it documents the micaZ power  
draw at 27uA for just the RTC. That's a factor of 10 different than  
what we've measured.

>
>>> I'd really love to see what the msp430 pulls; I would not be   
>>> surprised
>>> if it is similarly lower (e.g., < 0.5uA). But measuring it  would be
>>> great. If only my Ubuntu drivers weren't so cranky...
>>
>> I would also be interested in a side-by-side comparison of this data.
>
> Here are some tests on 2 Telos Sky:
>
> 	ACTIVE 1.83mA  1.83mA
> 	LPM0   0.15mA  0.15mA
> 	LPM1   0.15mA  0.15mA
> 	LPM2   26.2uA  26.4uA
> 	LPM3    7.0uA   7.1uA
> 	LPM4    7.0uA   7.1uA
>
> Running basestation disconnected from the PC and no packets
> received I got 20.01mA and 20.17mA.
>
> Note: the input voltage was 3V and I used the same Extech  
> multimeter as
> before.
>
> It would be great if Phil or Joe could post what values they got.

That's interesting. It looks like the micaZ draws less current when  
just running an RTC: I measured 3-5uA, with an average of 4uA. Of  
course, it is handling interrupts more often, and its wakeup time is  
longer, so the integral is probably larger. Measuring that with an  
oscilloscope would be very useful. Still, these values seem to be  
quite different than what has been reported. Based on your  
experiences, I think it's very possible that some of those  
measurements were from "bad" micaZs. To put the data side by side:

MicaZ MCU active draw: ~3.6mA
Telos MCU active draw: 1.83mA

MicaZ deep sleep draw: 3-5uA
Telos deep sleep draw: 7uA

You are using a controlled 3V input? That is what I have been doing  
as well.

Phil