Slashdot Mirror
Student Invention May Significantly Extend Mobile Device Battery Life
imamac writes with this excerpt from news out of Carleton University:
"Atif Shamim, an electronics PhD student at Carleton University, has built a prototype that extends the battery life of portable gadgets such as the iPhone and BlackBerry, by getting rid of all the wires used to connect the electronic circuits with the antenna. ... The invention involves a packaging technique to connect the antenna with the circuits via a wireless connection between a micro-antenna embedded within the circuits on the chip. 'This has not been tried before — that the circuits are connected to the antenna wirelessly. They've been connected through wires and a bunch of other components. That's where the power gets lost,' Mr. Shamim said."
The story's headline claims the breakthrough can extend battery life by up to 12 times, but that seems to be a misinterpretation of Shamim's claim that his method reduces the power required to operate the antenna by a factor of about 12; 3.3 mW down from 38 mW. The research paper (PDF) is available at the Microwave Journal. imamac adds, "Unlike many of the breakthroughs we read about here and elsewhere, this seems like it has a very high probability of market acceptance and actual implementation."
I mean my phone lasts for days if i don't use it and many hours if i'm just talking. The vast majority of power seems to be used when I'm watching video, playing games, or browsing the web. My guess would be this is more CPU related.
So even if it saves 10x in the transmit/receive it still might only be a 2x overall savings or less. I suppose it depends on usage patterns.
They also do this in recording studios. It takes far less power and wiring(or can be done via RF or IR) to have each speaker have its own small amplifier than to try to power the whole room with a rack of giant units.
This also would create less interference, believe it or not, since running wires near live electrical components(even the tiny components in a circuit board make a difference - just stick an AM radio near your computer's motherboard) tends to cause interference. This is the other reason recording studios do this. They can run a very heavily shielded or wireless line level signal to each speaker directly. Less power, less clutter, less interference.
For once, something that I'm actually qualified to post on!
I was a Weapons system depot level tech in the navy, doing lots of work with waveguides, radar, etc. I went on to work in the private sector, doing among other things antenna design at Nortel.
I can't help but say this is a bunch of shit. It is ALWAYS more energy-expensive to do wireless, it's just the way things are.
If it is just the journalist making a mistake, I can see some possible advances in energy conservation using a waveguide, or even a virtual waveguide; anything else would only start to be possible if you enter the realm of high energy physics.
Unless this guy's name is Tesla, and/or they have developed a completely new principle...
Why, yes, I AM a Pagan Libertarian.
what are the security ramifications? that a 3rd party might be able to intercept the wireless transmission just like they already can? whether you use this technique or not, you're still going to be broadcasting the signal wirelessly. that's why GSM signals are supposed to be encrypted.
the GSM encryption was broken earlier this year. the security ramifications of that are far more serious. why would you be worried about someone intercepting this weak wireless signal when attackers can already eavesdrop on your conversation from miles away?
heck, if they're close enough to intercept this signal, then they're already within earshot of you. they wouldn't need to intercept the wireless signal to the antenna. anyone silly enough to do so would look rather conspicuous standing there with a laptop and a directional antenna pointed at your phone.
"The on-chip antenna feeds the LTCC patch antenna through aperture coupling, thus negating the need for RF buffer amplifiers, matching elements, baluns, bond wires and package transmission lines."
From the systems perspective he made a better RF transmitter block. Digging into that block and looking at the RF design level, he simplified the circuitry normally used such as a matching network for the antenna, transmission lines, oscillator (for modulating the information over the carrier frequency), etc into a discrete chip as opposed to multiple printed circuit board components to do that same job.
Beyond that I'd need to study the paper and find more detailed examples of cell phone architecture to have a better idea of the advantages and disadvantages over the legacy design.
I'm not as qualified as paganizer, as I usually work at much higher frequencies (mmwave). However, losses from the PA to the antenna are typically pretty low. The claim of 12x improvement imply the current interconnects are at best 8% efficient (utter BS!).
From the PA to the radiated signal you typically have:
1. On PA losses because of their design. For example they typically have at least 3 different output stages to span from just a few milli-watts (single HBT cell), up to full power (hundreds of milli-watts, hundreds of HBT cells). The parasitics of driving the unused cells at less than full power operation creates small losses, but I don't know a hard number for this.
2. Baluns/impedance transforms. PA's are typically class B operation with a load line that is just a few Ohms (3V Vcc, and hundreds of mA of DC power, so the RF loadline is pretty steep). Solutions are matching structures, or a push-pull architecture through a balun to transform up to 50 Ohms. These usually account for 0.5-1 dB of loss (10-20%) of power. The invention ignores this part of a cell phones design.
3. Multi-band switch. Missing in this article is that most phones are designed to operate on at least 2, often 3 frequency bands. Several PA's are used, each designed to cover only one band. A GaAs phemt switch is usually used to switch between the two or more PA die. The invention does not address this aspect of cell phone design. These chips are either integrated in with the PA chip (separate die in the same carrier), or in some cases done in a different chip.
4. Small line loss from the PA chip to the antenna do have modest loss, usually just a few tenths of a dB (few percent). The article addresses this aspect of things.
5. The antenna is a clusterfuck of design hassles, as it is often dual, or tri-band in nature. A lot of compromises go on with the antenna. Making it have multiple resonances to cover the bands is hard. Making it small is hard. Making it work with the crappy ground plane, user's hand and head, and technicolor plastic case is damn hard. The article glosses over all this, and talks about a single narrow band antenna scenario.
Nevermind that he's apparently ignoring the true cause of a lot of the "lost" power - which is in the various bandlimiting filters that any real cellphone pretty much can't do without. It's tough to get a good multiband filter that doesn't have 1 to 2 dB insertion loss. The apertures are also geometric, so you are automatically sensitive to odd-order harmonics in both directions.
And I wonder how his aperture's impedance matches the amplifier out of band? From what I've seen in bleeding-edge RF architectures over the last 20 years or so, it's far easier to make a poor oscillator than a good amplifier, with any given set of components.
Less is more.
It's not so much that the path between circuitry and the antenna is so full of junk because of poor designs, it's because prior to this "discovery" no one knew how to get rid of that junk.
Now this guy shows us a way to bypass all that and gain the efficiency of removing all those components so that less power is used to get the same amount of radiation out of the antenna.
I KUT J00 M4NG!!!