[TriEmbed] Trail Traffic Counter Update and Lessons Learned

Pete Soper pete at soper.us
Wed Nov 27 14:52:18 CST 2013

On 11/26/2013 02:38 PM, Charles McClelland wrote:
> - Changed from 18650 LiON batteries to 4 AA alkaline batteries in 
> series.  Not sure if this is any better than 3 AA batteries so I plan 
> to test that next.  Thoughts?

Turns out the SD interface uses an amazing amount of current to write, 
perhaps 10 or 20 times the current the Mini's CPU is drawing.

> - Moved from 5V to 3.3V - this helped on two fronts.  1) The micro SD 
> card is a 3.3V device so I eliminated some components. and 2) I was 
> able to use the voltage regulator on the Pro Mini for more of my 
> battery pack's useful life
> - Put in a switch to turn off the LED - it turned out that the 
> indicator LED I was using for setup / testing was a major drain - 
> added a latching push button switch to turn it off for normal use
> 4) Sleep mode - I played with this but was disappointed.  It turns out 
> that the millis() count is stopped in all but the lightest sleep mode. 
>  This was a real problem before I moved to the real

If you don't need the clock chip's alarm feature you could program the 
INT bar/SQW pin to put out a square wave and periodically wake the cpu 
up with that. Another possibility is varying the CPU clock speed between 
active periods.

> time clock but I could also not get the interrupt working consistently 
> with the piezo sensor (too transient a signal?).  May revisit but it 
> looks like I hit my goal for battery life without this step.
> 5) Sensor - I heard there was discussion about using an accelerometer 
> instead of a piezo for sensing movement.  I would be interested in 
> this but have a concern about using a $10 for the most exposed part of 
> the system.  I recently bought a lot of 10 piezos from eBay for $6 and 
> my expectation is that this is the component which will get the most 
> wear and abuse.  I am looking at a few alternatives to the basic piezo 
> <https://www.sparkfun.com/products/10293> - a larger version 
> <https://www.sparkfun.com/products/9196>, one with a mass attached 
> <https://www.sparkfun.com/products/9197> (as I am looking for low 
> frequency signal), and perhaps a vibration switch 
> <http://www.mouser.com/ProductDetail/Mountain-Switch/107-2009-EV/?qs=sGAEpiMZZMuRFIlaTiDGn6wXOgQixfplWXe4vgIGDyc=> which 
> could solve my interrupt problems.  Would be interested in any 
> thoughts folks have on this.
My concern would be the second 10 piezos you get won't behave the same 
way. Not that I have any idea whether there is a lot of variability with 
these and how critical their performance is to your application. But 
with an accelerometer chip something very repeatable could be arranged 
and the sensitivity would be programmable both in terms of the G range 
mapping to the output bits and by filtering the samples. If you're 
intending to make your own board the right number for an accelerometer 
cost should be $2-3 tops.

It might be important to arrange to clamp the output of the piezo 
transducer so it doesn't blow a transistor junction when the crystal 
gets smacked hard. Notice that gas ignition gadgets are based on 
striking a crystal and getting very high potentials to arc over. :-)


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.triembed.org/pipermail/triembed_triembed.org/attachments/20131127/a5815ef9/attachment.html>

More information about the TriEmbed mailing list