Atmega Ultra low power development and measurement

Go To Last Post
8 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Dear Group, apologies if this post is duplicated.

Those of you developing commercial products intended for extremely low battery usage, how do you reliably measure uAmps ?

Do you use test instruments that can accurately measure uAmps while not getting destroyed by occasional usage of hundreds of milliamps for operating a relay or display backlight ?

Do you charge up a giant capacitor and measure the time the product can keep on running ?

Or do you use some way to integrate all power usage including occasional spikes ?

I am in awe of Honeywell / Ademco security products, where a friend had wireless equipment installed 4 years ago and still hasn't had to change batteries. I know some of their products use Atmega microcontrollers, but my friend won't let me crack open his entire system.

(Or maybe someone listened to the very funny NPR CarTalk episode where some mechanics kept adding gas to another mechanic's new Fiat, which made the owner brag incessantly about italian engineering, until they changed course and started removing fuel on a daily basis to have him completely humbled by his Fiat. So maybe ADT is sneaking in and changing batteries)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

Those of you developing commercial products intended for extremely low battery usage, how do you reliably measure uAmps ?

Reliably? We just use our decent Fluke handheld meters, like models 79 & 189. On the uA scale the 189 measures nA fairly well. The goodold 79 does OK down to low uA.

Accurately? So far, we haven't really cares when at sub-1uA. The Fluke meters switch in a higher resistance when at uA, so it does affect the reading somewhat. Swapping the leads and reading the [absolute value of the] difference seems to be the meter "overhead".

But I don't think that is really the question. The new AVR generation/models with a "V" model offer very good low power facilities. We have complete apps that draw only a couple of uA in power-down sleep. As a reference, get the Mega48/88/168 datasheet. The numbers/graphs in the datasheet can be accurately reproduced in operation. Your power consumption will depend on your app, and how you construct it. Continuous A/D conversions? You ain't gonna be living off a couple of uA. Wake up once a second, do A/D for temperature measurement, display it, go back to sleep? A few dozen uA average is certainly achievable.

Let's say you get 20uA average. 8760 hours in a year. That would be 175mAH/year. About the capacity of a 20mm-30mm "coin" cell. Cut the average consumption down and number of years go up.

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for your response. Yes, it is the kind of application that sleeps most of the time, yet occasionally has very short duration of spikes with high power usage, e.g. operating a relay and /or wireless link, and occasionally the user activates backlight for LCD. So it's really the total power consumption, which I can calculate based on duration and frequency of the power hungy events superimposed on the low power background, but I was curious about verifying my calculations.

Thanks for confirming that the numbers and graphs in the datasheet can be observed in a lab environment... mine will be the Atmega169 or the newer ones with more segments.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

occasionally has very short duration of spikes with high power usage, e.g. operating a relay and /or wireless link, and occasionally the user activates backlight for LCD.

Then a button cell probably isn't appropriate. You will need to get something like a camera battery that can handle 100mA+ pulse current.

In any case, >>you<< have to calculate your app draw, and make design decisions on limiting backlight intensity, type of relay, etc. The sleeping AVR is the least of your worries.

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

As you want to measure the power usage of your own equipment and not somebody elses the problem can be made easier.

I would measure the current seperatly in each of the known states. Ie measure it in sleep mode, idle mode, backlight on mode etc... Use a special software build to hold the MCU in the modes for measurements.

You should know the time spent in each mode from your coding and specifications. Now you can work out the average over time.

If you were making the measuring equipment, it would be much harder.

Tim

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

For measuring really small currents I use a precision resistor and measure the voltage drop across it. My old but still bitchen HP 3455 has a mode where you can measure the resistance and then use it to automatically calculate current when the voltage measurement is taken.

Tim Ressel
Portland, OR
timr@earthling.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Cool. Thanks for the responses. I guess no one likes the idea to see how long the equipment will run off a known capacitor charge ?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Not really. Too many variables, especially with differing load levels. "How long will it run" isn't really pertinent--the capacitor voltage will be changing over the life of the experiment. The draw for any of your app pieces may change based on the change in supply voltage.

Just clip on a half-decent current meter, or use a low-ohm resistor plus decent voltmeter. Use either bench supply or target battery. [That said, watching the target battery voltage dip when the "big" load turns on is a very educational experience and should be done before shipment. Really ugly things can happen to supply V level if the battery cannot supply the pulse current.]

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.