Thanks for the information.
The numbers I posted previously is obtained by powering the mote through the programming board and substract the current draw afterwards. Recently, I repeat the experiment by connecting mote directly to the AC/DC power supply (with output voltage 3.8V). I didn't bypass the boost converter. The high input voltage I use might be a reason for the difference. But I assume that the boost converter still maintain constant power supply to the rest of the circuit. What bothers me most is the large variance I observed in the measurement. For example, running the "snooze" program, sometimes I can get a current draw of only 0.063mA for sleep and ~4.0mA for idle (without radio) but sometimes around 1mA for sleep and ~6.5mA for idle with exactly the same setting. I am not sure if this is because of the IO pin conflict problem you mentioned. Or do you recommend bypassing boost converter to get stabler results?
Another issue that I don't feel comfortable with is I don't observe very significant difference in transmitting AM packet and idle state with radio enabled (by calling RadioControl.start()). The testing program I use is a modification of GenericBase with UART removed. I get an overall powr consumption of roughly 1.3mJ per packet for transmitting (by averaging over 500 consecutive packets) and 1.2mJ for receiving. I did observe a difference when I bypass the ChannelMon and transmit/receiving radio "pulse" using SpiByte directly.
FYI, here is a working document of my measurement study on line. http://lion.cs.uiuc.edu/~zheng4/research/implementation/
Rong Robert Szewczyk wrote:
Rong,
How did you measure these numbers? There are a number of things that will affect the measurements:
- input voltage and boost converter. Boost converter essentially
eliminates the ability to operate in sub-100 uA regime; for the
measurements in the WSNA paper we did turn it off (we added a diode
bypass). Likewise, different input voltages will operate at different
efficiencies of the converter. Conversely, if you are bypassing the
boost converter altogether (the mote will operate from 2.7 to 3.3V) will
change the numbers quite a bit.
- software -- in some versions of TinyOS, there was an IO pin conflict
that resulted in motes drawing excess power. It has been fixed a number
of times, but unfortunately it keeps reappearing. In general, the
configuration of IO pins will have a large impact on the power
consumption. - HW configuration -- I doubt that this would have affected you, but it
is possible. Mica with Atmega128 can be configured to run with an
internal RC oscillator at frequencies up to 8MHz. At these frequencies,
the idle and active power consumed by the processor would be
significantly larger than those published in WSNA.
Let me know if I can be of further assistance. I would be curious to
see what caused these discrepancies.
Rob
_______________________________________________ Tinyos-users mailing list [EMAIL PROTECTED] http://mail.Millennium.Berkeley.EDU/mailman/listinfo/tinyos-users
