Smartphones gear-up for envelope tracking and antenna changes
Smartphones are geared up to crunch through a lot of data in delivering apps and Internet service, but all this new funky stuff places new drains on battery staying-power. Emerging envelope-tracking technologies may be able to help.
Smartphones like to communicate. Very much. So much so that Internet-based companies are busy buying stacks of server hardware.
Rob Ober, fellow at chipmaker LSI, spelled out how much hardware Internet-connected phones and tablet PCs need at the recent International Electronics Forum in Bratislava, Slovakia: "For every 120 tablets sold, you need one more server in the cloud. Apple's iPhone 5 launch generated a need for 10,000 servers in the cloud".
By 'cloud' Ober ostensibly means data centres, but the rapid proliferation of the constantly Internet-connected smart device is not just helping to fill data centres. It's draining the battery on the phones and, although moves to upgrade the wireless network to 4G will let people view more video and play faster games, the charger could end up being plugged in a lot more often.
Improvements in recent years in battery power strength and longevity will be obviated by this increased demand on power at the endpoint device. It is hardly surprising that the industry has been very focused on examining any technologies that might help this apparent dichotomy.
Although governments have in many cases now allocated frequency bands to 4G that were previously unavailable to mobile phones, these slivers of spectrum do not have more inherent data capacity than those used for 3G. This is why EE is able to roll out its 4G service on frequencies originally allocated to its owner for 3G networks. The increase in data-rates is not coming through more radio, but by squeezing more out of each frequency band by employing increasingly complex modulation techniques.
GSM-based 2G phones, for example, use a modulation technique known as Gaussian minimum shift keying (GMSK). This is a technique that is not dissimilar to the concepts used in digital music synthesisers. The core of the transmitted signal is a sine wave centred on the carrier frequency. To put digital information on to it, the modulator phase shifts this sine wave backwards and forwards relative to the phase of the carrier.
The receiver detects whether the received bit is one or zero by comparing the phase of the incoming signal versus what the carrier's phase would be. It would be conceptually simpler to use a technique such as amplitude modulation (AM) ' like old analogue radio receivers ' to do the same thing. However, GMSK has one important advantage. The sine wave is always at maximum amplitude. It lets the transistors inside the power amplifier responsible for boosting the signal to transmittable levels operate mostly in the saturated region, where their power efficiency is greatest. Even better, the transistors do not have to guarantee a particularly linear response ' unlike those used in audio amplifiers where linearity is king ' because only the behaviour in the flat-topped saturated region is important. Circumstances have changed since the days of 2G. The demand for higher data rates that came with 3G pushed the protocol designers into adopting more complex modulation schemes, such as quadrature amplitude modulation.
These alter both the phase, as with GMSK, and amplitude. This presents a problem: the amplifier has to operate in a more linear way, which is bad for efficiency in general. It also means that, for much of its time, it will back off its power output to fairly low levels that are even less energy-efficient.
The Long-Term Evolution ' LTE ' standard introduces a more complex scheme, based on the same technology used in Wi-Fi: the transmit band is split into very narrow strips, and each one is used to transmit a small number of bits. The combined signal handled by the power amplifier requires even greater linearity to ensure that the receiver will decode the signal correctly.
One key metric is the peak-to-average power ratio (PAPR) or crest factor ' the amount of power that the amplifier could generate at its maximum-rated output power versus its long-term actual average. GSM had a PAPR rating of 3dB. 3G saw this ratio increase to 7dB. With LTE, RF designers are looking at a PAPR of 10dB and, with it, much lower power efficiency.
As a result, phone users get higher data rates, but they use a scheme that punishes the battery inside. This is one reason why one recommendation for users worried about making it through the day on one charge is to switch over to 2G-only mode if they do not have a lot of data to send or receive.
A secondary problem for handset makers is the sheer number of radio interfaces they have to squeeze into one design. Jonathan Borrill, director of marketing at RF test specialist Anritsu, says: "There could be 13'or 14 radios on one handset, including LTE, 3G, Bluetooth and Wi-Fi. And we might expect Bluetooth, Wi-Fi and LTE to all run at once."
These days, most of those bands are handled by dedicated circuitry, using physical switches to take the RF signal from each antenna to the receiver circuitry, and back out the other way. As the number of bands has grown, so have the switches.
“We’ve gone from supporting 4-throw architectures to smartphones that require up to 32 switch paths, and this is increasing,” says Rodd Novak, chief marketing officer at Peregrine Semiconductor. Phone makers would prefer to move much of the burden of supporting many different radio interfaces into software, using broadband circuitry to generate and decode the narrowband signals they need to work with. The reason they have not done that so far is that power efficiency would plummet even further.
Broadband amplifier takes a bow
Although a power amplifier can conceivably operate across a wide range of frequencies, each one has a sweet spot. This is what contemporary power-amplifier makers exploit. But, to provide consumers with LTE handsets that can roam around the world, something has to give if the phone is not to balloon to iPad size. The broadband amplifier is coming ' and curiously, the secret to making it work was developed when vacuum tubes were used for transmitting radio signals rather than transistors.
The power output of an amplifier depends very strongly on what voltage it is fed by the power supply. If you shift the supply voltage up on, the power output moves with it.
While doing so, however, the transistor can always be operating at its peak efficiency. This is the core principle that lies behind envelope tracking ' separating the signal into two components so that the power supply circuitry that feeds the power amplifier handles the bulk of the work in determining signal amplitude.
The amplifier stage continues to take care of doing what it does so efficiently: providing a 'fixed' power boost to a phase-shifted sine wave. Power amplifier manufacturers such as RFMD see envelope tracking as being the main method that they will use to recover the efficiency lost in the move to broadband designs.
"A year and a half ago the industry wasn't sure about it," says Jeremy Hendy, vice president of sales and marketing at envelope-tracking specialist Nujira. "Now the snowball is rolling down the hill. But it's going to happen." Yet it is not straightforward for them to adopt. "There are four things you need to get envelope tracking into smartphone," Hendy explains. "You need power amplifiers that are capable of being envelope tracked, the envelope tracking chips themselves, an output from the baseband modem to provide the envelope signal and firmware to calibrate the system in production."
One issue is technological, although vendors such as Nujira and US-based competitor Quantance believe they have that problem cracked. Nujira is already shipping its first-generation design and expect to go into production phones with a new version during 2013. For envelope tracking to succeed a traditional power supply will not work. It needs to be able to react fast enough to cope with a signal that changes amplitude millions of times a second.
If implemented using a traditional switched-mode power converter ' the standard used throughout consumer electronics ' such high switching-rates would lead to massive energy losses. Hendy likens Nujira's approach to speakers: a low-rate switching 'subwoofer' that provides a relatively steady flow of power; a mid-range digital supply that operates at a higher frequency and which can respond in microseconds. But the very rapid changes in power envelope are handled by the 'tweeter' a purely linear power converter. This type of converter is less efficient than a switching design when delivering a steady amount of power. But it is far more responsive. Having the linear supply deliver the tiny rapid voltage changes on which envelope tracking relies provides better overall efficiency.
"It's all about how you balance switching elements and linear amplifiers," says Hendy.
The OpenET group
To implement envelope tracking, chipmakers have had to agree on an interface between the various parts. To try to obtain agreement, Nujira was instrumental in setting up the OpenET group, which has defined one standard interface. MIPI, which has produced a number of standards now used across the mobile-phone industry, has defined its own. Hendy says the two are very similar making it possible to support both and that compatible parts are likely to appear in 2013. But vendors did not want to move until it was clear that baseband modems would drive envelope-tracking designs.
The wait now, says Nujira's Hendy is "really down to the availability of power amplifiers. They are still six to nine months away from production. It's frustrating in some ways that that the simplest element is taking the time".
One further issue is how to calibrate each phone to ensure the amplifier is correctly matched to the power supply. This is something that the phone makers have never had to do before. The chipmakers are putting self-test features into phones to avoid forcing the manufacturers to stock up on test further equipment and trained staff. Although envelope tracking is likely to arrest the increase in power consumption, the power amplifier is still far from being the only problem for smartphone designer engineers.
Ideally, each radio interface would use its own antenna but this has become more or less impossible to achieve. The introduction of networks that use more than one antenna to capture a signal ' a technique called MIMO ' is making matters even worse. New versions of Wi-Fi, such as IEEE802.11ac, use two more than one antenna to provide greater data rates. The combination of multiple signals makes it easier to detect data buried in noise. 4G networks expand the use of MIMO to cellular communications.
To include extra antennas for MIMO, other radio interfaces have to share. This is bad news for signal reception. The poor suitability of a one-size-fits-all antenna is not the last of its problems. After the launch of the iPhone 4, Apple discovered how a flawed antenna design can be exposed simply by holding it differently. Depending on how a phone is held or where it is, much of the radiation can be reflected back inside the case ' suddenly the number of bars shown on the display plummets simply because the phone has been placed on a metal surface, like a table top.
The issue of reflected power can be exacerbated by legislation. Governments limit the amount of power in a mobile phone as a precautionary measure in case excessive microwave emissions cause damage to the brain or body. Phones measure their output from the power amplifier, but not necessarily what makes it out of the antenna.
The result: the phone may limit its output not realising that electrical mismatch is preventing it from being heard. Then data rates fall, and phones spend more time on the network trying to send information, thereby clogging it up for other users.
Novak says: "RF performance is now very important, and operators are worried about how it performs in the field. What happens when the phone is against your head or on a metal table? Or when the phone gets hot?"
One approach to dealing with these antenna issues is to tune it dynamically. Peregrine, for example, can vary the capacitance of the circuit attached to the antenna, helping to match it electrically to the conditions at hand. To deal with the problems of what happens when the phone is placed on a table or clasped by a hand against the user's head, some manufacturers are considering using internal sensors to detect tell-tale movements that point to these conditions. Other vendors, such as Black Sand, have developed power amplifier interfaces that measure the reflected power in real time. But the more advanced techniques again need reliable cooperation between components inside the device.
"The world of antenna-tuning is a bit like envelope-tracking a few years ago," says Hendy. "It's like a fuel injector: you need an engine-management unit to control it."
As features such as MIMO become more commonplace, along with the networks on which the technique relies, Hendy says he sees a shift in the mobile-device market: "With 4G, suddenly there is a lot more freedom to have different bandwidth and capabilities within the chipsets. Then at the other end, there are cheap and cheerful options. I see a wide dynamic range of what vendors can do."
4G to the fore: Network headaches
In contrast to 3G, even 4G voice calls are handled using bursts of data packets rather than allocating the continuously transmitting channels employed by 2G and 3G networks. 4G networks based on the Long-Term Evolution (LTE) standard potentially could support data rates measured in hundreds of megabits per second - achieving performance levels encountered on Wi-Fi nets. Even though some networks are up and running and more will be going into action next year, LTE is still something of a moving target.
For example, the first public test of a handover of a voice call from LTE, which uses packets of data to 3G, which relies on the network forming virtual circuits, only took place in Spring 2012. Most operators expect to handle voice calls on the older networks, reserving 4G for high-speed data in hotspots. They expect to upgrade their networks when they are up and running. The use of MIMO - in which data is sent and received using more than one antenna in parallel - complicates further network design, says Jonathan Borrill at Anritsu: "MIMO makes it really quite difficult for a planner to predict what kind of coverage they might get. The MIMO algorithm almost doubles the data rate but uses non-line of sight reflections to get improvement. Traditionally, if you stand in direct line of sight of the basestation you will have a good data-rate. But with MIMO it can work better when you are in the shadow of a building."
|To start a discussion topic about this article, please log in or register.|
"There has been a lot of talk about the reported £30bn cost of the Sochi Games, so we go behind the scenes to find out where all that money has been spent"
- 3D Magnetic field rotation of light [10:55 am 09/03/14]
- Repeated Alternator Failure on Power Plant rated 16MVA/ 11,000V using 12x 415V generators [09:01 am 09/03/14]
- IEEE salary survey [04:59 am 09/03/14]
- EMC compliance - retrospective [01:10 pm 08/03/14]
- Microprocessor, integrated circuit Failure Modes IEC61508 [01:03 pm 08/03/14]
The essential source of engineering products and suppliers.
Tune into our latest podcast