how does power consumption for a battery work for a wireless device?

This isn't a good assumption. The mAh specification is only valid under certain load conditions, with a new battery. You may only see

1Ah at a 10 hour discharge rate, for instance. At 1A you may only get a 1/2 hour service. This is also specified to a terminal voltage, which may be below what your device can use and often below where you should discharge the battery, depending on technology. Consult the battery datasheets before making such assumptions.

You're overcomplicating this.

1) Quiescent power - power when the device is sitting there doing nothing 2) Active power - Power when the device is doing something useful. There may be several stages of active power, depending on your device. a) Transmit power levels b) Processor load c) Display backlignt d) ...

Don't ignore anything out of hand. Many modern processors take significant power. Your desktop PC's processor may dissipate up to

100W. Display backlights may not be insignificant either. THere are many things that can swamp transmit power, even. The transmission rate is normally insignificant since the transmitter is powered regardless. You may be able to modulate the transmitter power for different data rates, however. I'd ignore this part unless t was good reason not to (the radio's specification listed this as an operating property).

In any case, you simply integrate the current drawn in the various modes over the time those modes are in use.

I think you're right here. Once you've figured in your transmitter power the data rate shouldn't matter. "It's in there."

Hi Keith,

Thanks for the reply.

A few comments here:

1) So should I be modeling the terminal voltage or the voltage that the device can use? I want to calculate the corresponding mWh for the battery. 2) Since I'm not using real batteries and am just modeling power consumption in software, can I generalize the battery model? It doesn't have to be a complicated model, just something that allows me to compute my "run-time" average power. By that I mean, based on my elapsed time, I can raise or decrease my average based on consumption to date, so if I have an average of 3 mWh, if in the first half an hour my average is 1.5 mW, I can use up to an average of 4.5mW for the second half an hour. Either something like assuming a constant mAh rating and corresponding voltage, or assuming certain load conditions, to be able to calculate that mWh that I'm interested in. 3) "This is also specified to a terminal voltage, which may be below what your device can use and often below where you should discharge the battery, depending on technology." How do I find this information of the voltage the device can use based on the technology it's using?

To simplify my model, I'm assuming that the quiescent power is zero and there is no backlight (as simple of a model that I can use to model power consumption and battery life). This suffices since my simulation is a proof of concept and there is no real way for me to model quiescent power. Correct? In my original post, I specified:

I assume 2) is the "a) Transmit power levels" that you specified. Is

3) considered the "b) Processor load" while using a technology? As mentioned, this would be the energy consumption/bit x data rate in bit/ sec. This would give me power consumption in joules/sec.

Really? I'm working in wireless sensor networks and the biggest problem with building high speed sensor networks is the power consumption required to transmit at high rates. If the transmission rate is normally insignificant because the transmitter is powered regardless, I would think that high speed sensor networks would be the norm. Currently, sensor networks are inherently low rate due to the power requirements of high data rates.

Once I've figured out my transmit power (and thus correspondingly the data rate), is it the transmission power itself that affects battery life, or the energy consumption for the technology (the energy consumption/bit x rate in bits/sec) that affects battery life? or is it both? I agree with you that, "it's in there" so which one should be modeled?

Thanks a lot. If you can answer these questions, I'll be well on my way.

Omar

Your "model" is pretty simple. Multiply current drawn (or power, depending on your load) by time. Yes, you can't count any battery life at a voltage less than your device can call "good".

Your "model" simply has to integrate current(power) over time. This is by definition Ah(Wh). Yes, if you integrate 1.5mW for the first half hour and 4.5mW for the second, average is 3mW.

However, to get to the real world you have to compare this with a real battery's performance at the current(s) in question. A battery may be like a water tank, but only at the loads specified. It may "leak" if the "flow rate" is too high or too low.

Your device should be specified. If it's your design, you should be designing it to match the battery (and verse visa).

FOr the battery end, battery datasheets are the source if information. There used to be a manual called the "Gates Battery Handbook", or some such. It had all the curves for their various products and technologies. It was so useful that someone swiped mine. :-(

Here is a copy of what I think is an updated version:

It is highly recommended if you're going to design battery powered devices. Note to self: I should get a new copy.

If your quiescent power is zero there is no need to model it. It's zero. The battery life will be the shelf-life (primary battery) or the self-discharge (secondary battery).

Technology? Indirectly, perhaps. It's also the design. More gates (or gates switching) more power. If you want to go back that far, uses you can model the number of gates (or gates switching). Each takes a finite charge out of the battery, which is technology dependent. It's a rather circuitious route to get battery life thought.

I don't know exactly what you're calculating, but it may be the maximum data rate at a given power level (or verse visa). Yes, that's true but in practice one builds a device based on the requirements. If the real data rate is less than the design power gets wasted. TO mitigate this transmitters may shift power modes but you can simply model these power modes.

Transmit power level plus losses or the input power to the transmitter. Your data rate no longer matters (much). If you're not transmitting at the maximum rate you waste power. That is, you could have transmitted at a lower power but didn't design for that point.

Hope I helped. If you're designing a widget I really think you're over-analyzing this.

I guess the main concern is how to model the energy consumption/bit x rate in bits/sec for power consumption. This gives me an indication of how much power is used for a transmission using a particular technology and data rate (I am doing a technological analysis of power dissipation with varying rates). As you suggest, I am calculating the data rate at a given power level for various technologies. There is power wastage as well as you suggest, because modeling maximum data rates in software provide significantly higher rates in theory. For reasons such as regulation, devices aren't allowed (or able in some cases) to transmit at that maximum rate so devices would switch power modes to be more efficient. I digress, I could do a gate calculation, but this wouldn't consider the current rate that a device is transmitting at and modeling the corresponding power consumption, which is the key to my study.

Since as you said that the main stages for active power include transmit power levels and processor load, finding the relationship in concrete terms between energy consumption/bit x rate in bits/sec and power consumption is really important. I suppose it does figure into the "processor load" stage that you suggest. What I have done currently is reduce transmitter battery power by a sum of the transmit power level itself (excluding any antenna gain or rate relationship) + power consumption based on the rate and technology used (which I've been denoting as energy consumption/bit x rate in bits/sec). For a receiver, I have only modeled the power consumption in receiving a transmission as energy consumption/bit x rate received in bits/sec (in calculating these values for a transmitter, I've done the integration over time of power to determine how much power I have used and how much I have remaining).

I believe this is the simplest way to model it. What do you think?