Do Not Buy a 5G Smartphone

by Ron Amadeo – Dec 14, 2018 | Original ARS Technical article here.

We dive into the many ways first-gen 5G hardware will (temporarily?) ruin phone design.

2019 is going to be the year of 5G—at least, that’s what the cellular industry keeps saying. We’re going to see the launch of several 5G smartphones from OEMs like Samsung, Motorola, and OnePlus, and carriers will be tripping over themselves to tell you how awesome their new 5G networks are despite coming with a slew of asterisks. I would like to make something up about how ridiculous the 5G hype has gotten, but it’s hard to top actual quotes from industry executives, like Verizon’s claim that 5G will “dramatically improve our global society.” Faster mobile Internet is coming, but should you care about it yet?

Qualcomm recently had its big 2019 chip announcement, and as the world’s biggest provider of smartphone chips, that gives us a good idea of what the upcoming 5G hardware will look like. The industry is doing its best to hype 5G as The Next Big Thing™, but 5G hardware in 2019 is going to be a decidedly first-generation affair. Early adopters for 5G will have to accept all manner of tradeoffs. And when there might not even be 5G reception in your area, it might be better to just wait the whole thing out for several years.

5G mmWave Primer

Making Use of Spectrum That Nobody Wanted

“5G” is a shorthand reference to the next generation of cellular network technology that is launching in 2019. The whole “G” naming scheme started in the 1990s with the launch of GSM, which was called the “second generation”—aka “2G” — of mobile networking technology. GSM upgraded early networks from analog to digital, and those old analog networks were retroactively given the name “1G.” Since then, we’ve gotten new “G” numbers with major coordinated network upgrades about every 10 years. These iterations brought important features like SMS and MMS messages, IP-based networking and mobile Internet, and, of course, more speed.

Today, modern smartphones run on 4G/LTE, which operates somewhere in the 450MHz to 5.9GHz range. The move to 5G will include improvements to the existing LTE infrastructure, but the defining characteristic of 5G is the addition of a new chunk of spectrum in the 24GHz to 90GHz range. The industry has settled on calling this new 5G spectrum “mmWave” (millimeter wave), and it’s going to require

  1. new hardware in your phone,
  2. new hardware on the towers, and
  3. big changes to current phone and network designs.

We’re used to these “G” network upgrades coming with a compelling sales pitch about how much better everything is going to be, but the move to 5G mmWave is not a slam-dunk.

Since mmWaves runs at a significantly higher frequency than LTE waves, that means it comes with no shortage of tradeoffs:

  • MmWave has worse range and worse penetration compared to LTE.
  • A mmWave signal can be blocked by buildings, trees, and even your hand.
  • MmWave doesn’t work well in the rain or fog, and
  • the ~60GHz chunk of this spectrum can actually be absorbed by oxygen.

That’s right—a slice of mmWave spectrum can be blocked by the air.

With so many issues to overcome, mmWave sounds like a terrible chunk of spectrum to build a mobile network in until you consider two key points: the higher-frequency means mmWave has plenty of bandwidth and low latency if you can get it, and most of all, the spectrum is available.

MmWaves aren’t being used for much right now because it is such a pain in the butt to get them to work well enough. So if you can figure out all the implementation problems, you suddenly have a vast amount of airspace to work with. That’s actually the first thing these companies talk about when they bring up mmWave. It’s all going to be really, really hard and complicated, they say, but it’s going to be worth it.

4G/LTE debuted in 2011, and these past seven years have seen significant progress in making 4G smartphone hardware smaller, faster, and more efficient. With 5G, we’re going to lose plenty of this technical maturity initially by packing in tons of new and expensive 5G hardware.

Discrete 5G Modems

More Components, More Power usage, Smaller Batteries

Smartphones today are almost entirely powered by a single chip, appropriately called an “SoC” or “System on a Chip.” As the name suggests, these are the most basic parts you need to make a computer all on a single, tiny chip. There are usually lots of CPU cores, a GPU, an “ISP” for camera functionality, Wi-Fi, and more. RAM isn’t technically included on this chip, but to save space, the RAM actually gets stacked on top of the SoC. The main off-SoC component is the storage, and across the motherboard there will typically be a sprinkling of tiny chips for power management, audio, Bluetooth, NFC, and other things. From there, it’s the motherboard’s job to connect everything to everything else and then get the hell out of the way so that as much as the phone as possible can be filled with battery.

The point is that space is at a premium inside a smartphone, and while you can’t do much to control the size of core components like the SoC, camera, SIM card, or USB port, the battery is the one part that can be as large or as small as you want it to be. When you think “size” in a smartphone, you should think “battery.” Anything that gets bigger means less battery. Anything that adds an extra component means less battery. The battery gets all the leftover space in a smartphone. (This is, basically, the headphone jack argument.)

These past few years, smartphone manufacturers have all been trying to convince us that we don’t need a headphone jack, and the argument has been that removing them means less complexity and more space for battery. Razer CEO Min-Liang Tan even put a number to this argument: he said that skipping a headphone jack in the Razer Phone meant the company could increase the battery capacity by 500mAh.

Why does this matter in an article about 5G? The short answer is that 5G mmWave is going to require a lot more hardware than 4G/LTE, which brings up all of these battery size and device-complexity concerns.

Qualcomm’s biggest advantage in the 4G era has been its modems. Through a combination of technology and intellectual property rights, Qualcomm is the only chip maker that can combine an SoC and modem into a single chip and sell it around the world at a low price.

This single-chip solution is a huge advantage, resulting in a smaller, less-complex, cheaper motherboard and more room for battery. Merging everything into a single chip also results in power savings while the phone is running, since, generally, one chip takes less power than two chips. For years, Qualcomm users have enjoyed SoCs with onboard 4G LTE modems, and the company rode this design advantage to market domination. Today, as a high-end SoC vendor, Qualcomm is basically a monopoly, with nearly every Android flagship using a Qualcomm SoC.

Qualcomm recently showed off its flagship SoC for 2019, the Snapdragon 855. While the company spent hours hyping up the Snapdragon 855’s 5G compatibility, it won’t actually have a 5G mmWave modem onboard. The 855 will have LTE onboard, as usual, but 5G phones will need a separate modem—Qualcomm is going to lose its single-chip advantage for 5G. As explained above, this means less battery and more power usage.

We’ve already lived through the whole “first-gen network hardware” routine before. When the switch to 4G happened, the first batch of new 4G hardware arrived with the same discrete modem compromise that we’ll see with 5G. The most famous example was the HTC Thunderbolt, the first 4G device on Verizon’s network. This used Qualcomm’s Snapdragon MSM8655 SoC (before the simplified model numbers!) with a separate Qualcomm MDM9600 LTE modem. The Thunderbolt was a disaster, since it included all this new 4G hardware with only a 1400mAh battery. It was thick, hot, slow, buggy, and had terrible battery life. The Thunderbolt regularly makes lists of “the worst phones of all time,” and one HTC employee even apologized for the phone’s creation. New network hardware can be a disaster if you do it wrong

Qualcomm’s Gigantic First-Gen 5G chips

I’m not guaranteeing that the first 5G hardware will be as bad as the Thunderbolt — a lot about smartphone design has changed since then — but the worry about early networking hardware remains. New networking hardware is “new” precisely because this is the first time it’s been made small enough to fit inside a smartphone. Companies don’t wait until they are well under the size requirements for a smartphone to ship a product; they design a new phone the very second new networking hardware will just barely fit. These 2019 devices will likely be bloated with 5G hardware — the biggest 5G hardware that will ever exist.

Let’s compare the internal components you’ll need to make 4G and 5G work. In terms of major Qualcomm chips, a 2019 4G phone will have the Snapdragon 855, and that’s it. There’s an onboard LTE modem in the 855, so you don’t need an extra chip for connectivity, and LTE antennas are tiny wires that are usually integrated into the motherboard.

5G is a totally different story: you’ll need the Snapdragon 855, plus the Snapdragon X50 5G modem, plus a series of “QTM052” 5G antenna modules, which are actual chips instead of wires or motherboard traces.

I haven’t seen Qualcomm publish exact die sizes for its 5G chips or the Snapdragon 855, but the company has a habit of photographing its chips next to pennies for scale. Here’s the Snapdragon 855, the 5G modem, and the 5G RF Module, all pictured next to a US cent. These coin pictures are meant to say, “Look at how small our chip is!” but they also give us a perfect ruler to load everything up in Photoshop and have to-scale Qualcomm chip pictures.

It turns out Qualcomm’s first-gen 5G chips are going to be really big—at least, “really big” compared all the other extremely-small smartphone components. The X50 5G modem and a single RF module take up more space than the Snapdragon 855. Keep in mind the 855 is an entire SoC—nearly everything you need to run a computer—plus a built-in 4G modem, so it’s pretty incredible that a 5G modem and a single RF module takes up just as much space as the entire rest of the core phone components, minus the storage chip. For now, 4G has a clean, single chip design, while 5G is going to have these massive extra chips to deal with.

Exponentially More Complicated Hardware

Lots of design challenges

At Qualcomm’s Snapdragon Summit in Hawaii earlier this month, Qualcomm President Cristiano Amon spoke a bit about just how complicated it is to build a 5G smartphone. One of his slides said that 5G would increase smartphone design complexity “exponentially.”

While Amon said that hoping to make Qualcomm’s engineering sound impressive, to me it sounds like a battery-killing nightmare. Complexity is bad. Complexity is expensive.

It’s not just that 5G hardware is bigger than 4G hardware — you’re also going to have to absolutely pack a phone with 5G hardware to make it work. MmWave’s penetration is so poor you can easily block the signal with your hand, which is kind of a problem for a device that you constantly hold while you’re using it. To avoid situations where “you’re holding it wrong,” Qualcomm’s solution is to pack the phone with multiple 5G antennas.

Your 5G phone won’t just need a Snapdragon 855, an X50 modem, and a QTM052 antenna module — it’s actually going to need multiple QTM052 chips. Qualcomm’s page on the QTM052 says you’ll need “four differently located mmWave modules” in order to work around all of mmWave’s limitations, although more advanced designs might get down to three. Qualcomm’s 5G diagrams all show four 5G antenna modules in a phone — one on each side of the device. Qualcomm’s “reference” 5G prototype (which Anandtech recently had a look at) shows three 5G antennas: top, left, and right. Motorola’s 5G Moto Mod—the closest thing we have to a real, consumer design—has four antennas: left, right, and two on top.

No matter what configuration OEMs go with, it looks like they will all have some kind of multi-antenna design, which will let the phone pick whichever antenna your hands aren’t blocking. For an example, imagine holding a phone in portrait—your hand might block the left and right antennas, so the phone could switch to the top and bottom antennas. If you’re using landscape, your hands might block the top and bottom antennas, so the phone would use the left and right ones. A 5G setup will intelligently switch to whatever antennas can get a signal out.

More antennas mean even more complicated internals and even more space used up. If we go back to our to-scale chip photos, Qualcomm’s 2019 5G package is going to take up an absolutely massive amount of space compared to 4G. I don’t know how thick everything is, but in terms of pixels in Photoshop, 5G uses 3.3 times more area than the “4G” configuration of a single SoC. For an industry that can’t give us headphone jacks anymore because they take up too much space, all this extra 5G hardware seems kind of hard to justify.

The design gets more complicated than just laying everything out on a flat sheet, though. These multiple 5G antennas are there to enable a technique called “3D beamforming.” Rather than send a signal out in a wave (think of the Wi-Fi symbol), a beamforming antenna can locate where the tower is and will fire a concentrated signal directly at it. Similarly, the tower will track your location and fire a beam directly at you.

Qualcomm says 3D beamforming is a key design feature of 5G devices, since a more concentrated beam will help overcome some of the range and penetration problems of mmWaves.

So far, every piece of Qualcomm literature and all the hardware show these antennas placed in a perpendicular plane to the motherboard on the side of the phone.

If consumer 5G really does need a large surface area of the antenna exposed on the sides — which everything has so far indicated is true—that raises some interesting design considerations. Today, wireless charging has seen phones switch from metal unibodies to glass backs, which allows the RF signals to get in and out of the back of the device. Since glass isn’t at all durable, these phones usually use a metal mid-frame for support, which is usually exposed on the sides. It doesn’t seem like a metal-sided smartphone design would work with side-mounted 5G antennas, though—you won’t be able to have metal on the sides of the phone if an antenna needs to be there. Surely, we can’t have glass sides, too, so that leaves … plastic? 5G might usher in the era of plastic-sided phones. I don’t see any other way these side antennas can work.

Also, somebody’s going to have to pay for all these extra 5G components, and it looks like it’s going to be the consumer. OnePlus CEO Pete Lau recently told The Verge that 5G would mean phones that are $200 to $300 more expensive. That’s a pile of money for a phone that won’t have the runtime or more compact size of a 4G phone.

The Hard Reality of the 5G Rollout

Clearly, 5G is going to require a tradeoff in smartphone hardware, at least for the near future. Moving to 5G means lots of hardware compromises over a 4G phone, so what do you get in return? The answer in 2019 is only “possibly faster Internet” depending on a whole host of variables, mostly having to do with your location. First, you need to be in a city that actually has 5G, then you need to be in a specific spot where you can actually receive a 5G signal, and then you have to decide if you even care about the speed increase.

Receiving faster 5G speeds means actually being able to receive a 5G signal, and so far, the 5G rollout sounds like a nightmare. Since 5G runs at a higher frequency and has much worse penetration compared to 4G, it’s not a matter of just upgrading existing towers. Outdoors, the industry’s inelegant solution to 5G’s range and penetration problems is mostly just “build more towers,” which means lots of slow negotiations for land rights and placement.

Right now, it doesn’t even sound like the goal is blanket 5G mmWave coverage everywhere thanks to the limited range.

Qualcomm’s CEO outlined a “5G” network architecture for the United States that only used mmWave in “dense urban areas”—everywhere else would only use 4G/LTE.

Keeping mmWave to urban areas makes sense when you consider the range problems. The need to build so many more towers for 5G probably isn’t worth the cost outside of a city, so 4G/LTE will have to do. Even if you’re in a 5G city, getting 5G mmWave coverage indoors is going to be a problem thanks to the penetration problems. Usually, you’re going to need a tower inside the building to have a chance at reception (i.e., a femtocell) or you’ll drop down to LTE. No one in the industry claims that 4G/LTE is going away, and in fact, the current plan includes 4G/LTE improvements in the “5G” bucket.

So while there’s plenty of hype around how fast 5G might be eventually, in terms of actual, 2019 modems, it’s not the “10x” improvement in networking speed that often gets cited—especially as LTE keeps improving. The Snapdragon 855’s onboard 4G LTE modem—the single-chip solution with none of the 5G tradeoffs — has a new-and-improved theoretical top speed of 2Gbps down, according to Qualcomm. The X50 5G modem and a pile of RF modules can more than double that, with a theoretical download speed of 5Gbps.

Both of these cited speeds are theoretical velocities that you will never achieve in real life, but keep in mind the following:

  • You only need about 0.006 Gbps to perfectly stream high-quality 1080p, 60fps video.
  • Even something crazy like 4K streaming only takes around 0.025Gbps

. . . so current 4G speeds are more than enough for anyone.

I am sure some day when we are all streaming virtual reality data to the heads-up display on our face computers, 5G’s faster Internet speed may be useful. But for my next smartphone, I think 4G/LTE — which, again, still isn’t done getting faster — will be just fine.

Check Back in 2025 — or Even Never

5G in 2019 will be a flop. For 5G mmWave in 2019, we’re going to get thicker, hotter, more complicated phones that use more energy and cost more money. With no commercial devices to look at, the exact extent of all of these downgrades is still up in the air, but it’s undeniable that first-gen 5G hardware is going to be inferior to more mature 4G designs. With 5G networks only in their infancy and a supposed $200-$300 premium for 5G-compatible phones, this really doesn’t seem worth it for consumers. You are not obligated to spend your money on 5G now — or ever, for that matter.

Soon you’ll start seeing the launch of the first 5G phone on network X, and every carrier on Earth will be advertising 5G as the best thing since sliced bread. Don’t believe the hype. If you can buy a 4G version of a phone, you probably should.