|«Latest ‹Forward News item: 7772 Back› Oldest»|
|5G Heats Up Base Stations|
|USA||Created: 4 Jul 2020|
Before 5G can be deployed commercially on a large scale, engineers have to solve some stubborn problems—including how to make a hot technology a whole lot cooler.
5G-capable modem chipsets are already on the market from Qualcomm, Samsung, Huawei, MediaTek, Intel and Apple, with some 5G service (LTE-Advanced/LTE-Advanced Pro) available in the U.S. But still mostly missing from the 5G equation are base stations powerful enough to shape and direct an individual RF connection to every subscriber within range, while performing feats of electromagnetic geometry to maintain that connection.
A base station in the wireless world is a device that connects other wireless devices to a central hub. It is a wireless receiver and short-range transceiver that consists of an antenna and analog-to-digital converters (ADCs) to convert the RF signals into digital and back again. The 5G base station will have beamforming massive multiple-input, multiple-output (MIMO) antennas—an array of antennas that can focus and steer multiple beams simultaneously to different targets on the ground, such as a cell phone, using the millimeter wave spectrum. Sometimes that means bouncing the signal off an object to reach near the target rather than broadcasting a signal broadly over an area.
Although Ericsson, Samsung, Nokia and Huawei are producing 5G base station technology now, there are gaps in that technology. The base stations are still not powerful enough to track mobile customers and make sure each is connected every nanosecond.
What’s developed for base stations has to work seamlessly with handsets. They also have to be reliable enough to last for years, but the current technology is running too hot. And how that affects reliability and signal integrity isn’t clear because at that point now one is quite sure how the antenna arrays will be tested because there are no exposed leads. Those antennas are essential to form, steer and receive beams, both in the base station and in handsets and other mobile devices, including connected cars, health monitoring devices and even industrial equipment.
“If you embed the antenna into the package, when the package heats up or cools down, that changes how the antennas work,” said Keith Schaub, vice president of business development for Advantest’s U.S. Applied Research & Technology unit. “That affects beam forming, beam steering, and it creates a power loss. It also affects the fabrication process, which needs to be tightly controlled.”
Schaub noted that base stations and handsets are all designed to standards, but the implementation of those standards can vary greatly. For example, when two major chip companies developed their first 5G chips, they adhered to the standards but the chips wouldn’t work with each other due to minor inconsistencies in the drivers.
Despite the moniker, 5G is more of a statement of direction than a single technology. The sub-6GHz version, which is what is being rolled out today, is more like 4.5G. Signal attenuation is modest, and these devices behave much like cell phones today. But when millimeter wave technology begins rolling out—current projections are 2021 or 2022—everything changes significantly. This slice of the spectrum is so sensitive that it can be blocked by clothing, skin, windows, and sometimes even fog.
The result is that many more cells are needed to keep devices connected, and base-stations and handsets will be constantly searching for ways to stay connected. As anyone with a cell phone knows, searching for signals drains the battery faster. But it also keeps the logic circuits active, and that generates heat. In base stations, which are tightly packed with racks of equipment, thermal buildup can cause all sorts of problems. It can have an impact on signal integrity, and it can reduce the lifespan of all components.
“When you have a frequency with a range that’s not as far as a cell tower, you have to add much more density to the network to get the same amount of connectivity,” Michael Foegelle, director of technology development at ETS-Lindgren. “When you design these, you have to assume they’ll be outside, and you have to design in a way to dissipate all that. Since you’re outside and don’t want to risk putting in active cooling, you might have to go fix a lot, that means a lot of ambient cooling,”
Another source of heat stems from the analog circuitry used to generate RF signals. Power amplifiers and converters are needed to get the analog signal onto digital networks. But using silicon for those conversions isn’t efficient, so heat builds up. And while beamforming theoretically can save power, because you’re not broadcasting in every direction, that technology adds its own issues.
“First, you need enough hardware to do the number of digital-to-analog conversions you have to do, and the cost is still prohibitive,” Foegelle said. “But it’s also power-hungry. One of the side effects of the arrays is that the circuits used for them aren’t terrifically efficient. They get hot, and you have to be able to dissipate a lot of heat because of the amount of equipment and conversions and the efficiency issues.”
It’s not entirely clear if this technology will be replaced with digital technology. It’s also not clear how digital technology would impact effects such as heat, particularly if designs are pushed to the most advanced process geometries.
“The 5G standard allows for both,” said David Hall, chief marketer at National Instruments. “Analog circuits are less efficient, which creates more heat in the base station. With a digital beam, there is a change in the waveform itself, particularly with multiple access. So you have to adjust the phase to the wave carriers.”
Hall noted that heat exacerbates non-linear effects. “If you add heat, distortion is not as repeatable.”
That makes it more difficult to identify any heat-related issues. One solution may involve the testing itself. “Historically, we have been using box instruments,” said Heath Noxon, market development manager at NI. “Now you have to hit this more quickly and process test much faster.”
Different materials can help, as well, but they add to the cost. “You can get efficiencies using GaN or GaAS that are probably 60% or 70% compared to silicon, which is more like 20% to 30% efficient, but those are much more /expensive, ” Foegelle says.
That issue could be sorted out if there is enough volume for either gallium nitride (GaN) or gallium arsenide (GaAs) so that economies of scale begin to kick in. Both of those materials are well understood and there is plenty of expertise in working with them. “Engineers have spent 20 years optimizing the efficiency of gallium arsenide power amplifiers,” said NI’s Hall.
“The problem may not be as big as it sounds, though,” Foegelle says. “With millimeter wave the bandwidth is high enough you don’t have to spend much time communicating. It moves quickly, which could minimize heat buildup as well as reduce the amount of energy you broadcast. But we won’t know that until we’re able to see more work on base stations.”
The volume problem
Heat is just one of the many issues cropping up in the 5G world. This is an entirely different wireless technology, particularly when it comes to millimeter wave. The amount of pressure put on technology and service providers trying to move into—and often create—the 5G industry is very high, and few tools are available to test and validate any individual approach early enough in the process to be useful, according to Frank Schirrmeister, senior group director for product management at Cadence.
This is particularly important for dealing with heat, which can impact the lifespan of components. Thermal effects can speed up electromigration, impact performance, and create noise that can impact quality. But engineers are just starting to work with these technologies, and it’s not clear what else might crop up.
“If engineers are used to working at lower frequencies on these earlier cellular applications, and then they transition to working on 5G at higher frequencies, all of sudden all the rules are more stringent, all the rules of thumb go out the door, and you have to do a more thorough design,” said Mike Leffel, an application engineer at Rohde & Schwarz. “It is a more challenging design. Components don’t work as well at the higher frequencies as they used to in lower frequencies, so you really have to retrain yourself on how to make a well-functioning product. Everything gets smaller. Wavelengths get smaller. The ability to adjust the phase of a path is more difficult because now the wavelength is so small so a small change in a wavelength might be 10 degrees instead of 1 degree at lower frequencies.”
Rhode & Schwarz recently started one-day educational conferences to help engineers understand the issues. But for Leffel, preparing engineers for the 5G universe is “one of the biggest challenges for the customers that we have. They have to rethink how things work at higher frequencies. What I see is somebody saying ‘I used to do this at 6GHz, I didn’t even have to calibrate the cable. I would just hook it on and it was good enough. Now when I’m at 40GHz, when I do that, it fails. Everything fails and I have to do this calibration. And when I calibrate, it still doesn’t work right. And the guy came in from Rohde & Schwarz and said you have to use a torque wrench to do this. I never had to use a torque wrench before.’ Yes, but you never worked at 40GHz before. Now everything is touchy. And this is a more expensive, better quality cable at 40GHz. You can’t use that cheap cable anymore. You have to calibrate maybe every day instead of once a week. You have to worry about the length of that line and the insertion loss, so there’s an extra trace on this board, so you can measure how much loss is in that line and then subtract it from the results so that when you measure a path on here, you can correct for that trace. At low frequency you don’t have to do that. At high frequency, that trace is critical to know exactly. So all of these things you didn’t have to do before are suddenly important, and if nobody told you this, then how would you know?”
Millimeter wave technology isn’t new, and a lot of the networking issues in millimeter wave have been addressed before in satellite communications or radar. However, the cost difference between one satellite and a few hundred thousand WLAN-scale access points changes the cost/benefit equations enough that there’s not much direct comparison, said Cadence’s Schirrmeister.
There also are ongoing updates to the 5G standard. “With millimeter wave we’re talking about wavelengths of about a centimeter, so the antennas are also very small and you use two for each subscriber—one upstream, one down,” Foegelle said. “But for base stations we only have a few vendors marketing them. There’s still another version of the standard coming out later this year, so there’s some uncertainty there. And we are getting carriers coming in and trying to figure out what the propagation characteristics on their networks are going to look like and what types of problems they can expect to see in the field, but the prices are still pretty high for distribution of a product that you’re going to have to put out in density more like a WiFi access point than a cell tower.”
It is best to keep things simple with a technology like 5G, which is fantastically complicated to build and test even before the standards or first rounds of implementation are finished and proven, noted Susheel Tadikonda, vice president of engineering at Synopsys’ Verification Group. “The PHY layer is getting very complex. You need high bandwidth, and the latency requirement means you have to do a lot of the processing in the PHY layer itself. We used to have the luxury to send it up the chain and have it done with an algorithm. What you’re doing is moving logic form one portion to another. You still have to convert an analog radio wave. Doing it digitally may be more effective, but in 5G you have not 12 or 14 modems, but hundreds of antennas doing beamforming. It is much more complicated than 4G was, and the transition is more complicated than the transition to 4G was.”
There are good reasons to stick with hybrid approaches, however. All, or nearly all, RF base stations that operate below 6GHz use digital beamforming because it is more power- and heat efficient than analog. At frequencies higher than 6GHz the filters required for conversions take up too much space for digital to be practical, according to a 2018 presentation at MIT by Gabriel M. Rebeiz, a University of California San Diego engineering professor and an expert in high-frequency communications and phased array design.
Hybrid designs that use analog signals for RF and digital for networking are among the most common topologies used in satellite communication radar and other 5G-similar applications of the last two or three decades, communication methods, according to Redeiz, who specializes in millimeter wave and primarily on those issues before the growth in demand for terrestrial demand for high-frequency bandwidth.
Hybrid models are also less computationally complex than digital, though the arrays are larger, which makes digital beamforming much more attractive as the size of the devices and antennas shrink, according to an analysis published by Mostafa Hefnawi, a researcher at the Royal Military College of Canada in Ontario.
People are talking about a lot of ways to mix and match frequencies and protocols and devices in other ways that would deliver a lot of value from 5G, especially for people who don’t necessarily need microsecond latency and 10,000 Gbit/sec wireless network connections, says Gilles Lamant, distinguished engineer at Cadence.
“People are talking about putting RF over fiber, but to cross analog RF at high speed to digital might cause major heat problems. Still, those would be a lot less with a slower wireless interface, or even a smaller geographic area covered by 5G that allowed all that RF data to go straight onto the network digital domains,” Lamant says. “The key here is energy efficiency, so you send the RF across the fiber without converting it first and you can save money and time. You can convert it later or transport that signal straight to another RF domain. It is a little science-fiction-like to think about, and you would have to put more energy in the connection after a certain amount of distance, but if the tradeoff is in cost and energy. It is something to think about rather than sending data out over a heavy, slow coaxial cable.”
Relying too much on the idea of people using smartphones means ignoring a lot of other applications. Analytics providers or IoT network owners could find connecting to 5G access points as attractive a business proposition as a company needing instant high-speed access for mobile video, but companies doing two-way high-definition streaming use the physical network behind the 5G access point is much different than an IoT network sending big chunks of data in batches to the cloud.
“If all you care about is how fast you can post Instagram pictures, that’s a different set of concerns than if you have 100,000 devices spread out across a square kilometer than you want to connect,” Schirrmeister says.
At this point there are many unknowns. Heat is just one more issue, although it is an important one. But how that gets resolved may depend on a lot of other factors, from how much of the base station is digitized to the density of cells and base stations and the millimeter wave frequencies. At this point there is plenty of momentum for 5G, but there are a lot of variables in play that could have a big effect on how this wireless technology is rolled out, how well it works, and how long it lasts.
—Ed Sperling and Susan Rambo contributed to this report.
|Click here to view the source article.|
|Source: Semiconductor Engineering, Kevin Fogarty, 07 May 2019|
|«Latest ‹Forward News item: 7772 Back› Oldest»|