The first one that we liked a lot was the QCC5100 headphone chipset, the other half of the pairing we told you about in December. In Hawaii Qualcomm talked about the tech, how it saves power, the Bluetooth extensions, and where the initial signal comes from. What they didn’t talk about is the receiver but they did show off headphones that used a lot of the same technologies.
Custom chipsets are all the rage
The idea is simple enough, bring wireless *HUZZZZZK*PTUIIIII* earbuds to the masses by allowing even tier 12 accessory makers access to the same tech as the Apple buds but at realistic prices. Much of the claimed 65% drop in energy use comes in via the Bluetooth extensions that allow multi-pairing of devices. There is one signal sent to both ears rather than the source -> left -> right that is done now. Rebroadcasting a signal eats battery. Bluetooth doesn’t travel through large unsightly hunks of fat and water all that well either.
This means that to go from left to right ear buds, that retransmission needs to crank up the power, use other transmission mechanisms like magnetic resonance, or pull tricks that don’t tend to work well. On top of the energy use issue there is the problem of keeping the two buds in sync, something that Apple reportedly struggled with on their reassuringly expensive coffee shop entry pass earrings.
One thing that stood out, other than the spiffy limited edition Cristiano Amon QC5100 chips pictured above, was Qualcomm’s plans for wireless earbuds. There were numerous mentions of dropping the price of these plugs low enough to include in the box for phones, something that would be a welcome change over the $100 or so price tag they now command. The down side? It really says that those who know what is coming for the next several years see the roadmaps littered with devices lacking headphone jacks. Yes the robot masters won and it sucks.
RFFE is really important. Repeat, Repeat again.
A little more technically oriented news is the announcement of a tunable 3/4/5G RF Front End (RFFE) from Qualcomm. It started out with the news that they had wins from Google, HTC, LG, Sony, and Samsung, basically everyone that matters. Then it got a bit technical
The idea is to take the data from the modem and feed it directly back to the RFFE chip to take the current coarse-grained tuning and make it more fine grained. Why? Because you have to. For those of you not familiar with the RFFE world, not having a tunable RFFE is what gave Apple a bit of heartburn with Antennagate on the iPhone 4. If you can’t tune your antenna, your performance suffers badly, and if you can’t tune it precisely, quickly, and do so over a wide range, good luck in the 5G world.
And this brings us to the other problem, something Qualcomm has been saying for a while. The band combinations available have mushroomed of late. 3G was pretty simple, 4G added a lot of bands. When you start looking at 2/3/4 band Carrier Aggregation (CA) and dual SIMs, you get a lot more complex in a hurry to over 1000 band combinations. Add in 5G along with sub-6, 28GHz, and 50GHz bands and you get a combinatorial explosion to 10K+ combinations. Good luck with the RFFE on that, if anyone can do it, I would put my bet on Qualcomm, and so did the top 5 Android phone makers.
More on the business side was the announcement of a few new automakers using Qualcomm as their infotainment silicon vendor including Jaguar/Land Rover and Honda for the 2018 Accord. My memory might be a bit hazy but weren’t both those vendors the halo marquee wins for other silicon makers in the very recent past? I wonder if this means the dumptrucks full of cash stopped coming, either that or it was initially win on merit, you decide.
More interesting was the win with Jaguar/Land Rover which also uses a Qualcomm 820A for their connectivity as well as the infotainment side. I think this is the first use of Qualcomm’s SoC as an I/O device versus a plain modem. Anyone see an entry for edge compute and future services in this? Related to that the Chinese car maker BYD announced they would be using a Quaclomm 820A for the infotainment system and the instrument cluster. Remember how we laughed at a certain other silicon vendor with a penchant for questionable statements when they said that the explosions of screens in a car would be matched by an explosion in SoCs sold to the automakers? One of us was accurate, not just SemiAccurate.
On top of this there were a bunch of announcements from customers using Qualcomm silicon highlighted by the Xaiomi Mi VR Standalone and it’s Oculus Go clone. There were a handful of smart speaker partnerships, basically everyone that matters precertified Qualcomm solutions for their ecosystems, and the same for mesh Wi-Fi. In short this day was about customers implementing previously announced Qualcomm solutions, not new chips. Then again new silicon announcements doesn’t make Qualcomm money, people buying the products do.S|A
Latest posts by Charlie Demerjian (see all)
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018
- Intel can’t supply 14nm Xeons, HPE directly recommends AMD Epyc - Sep 7, 2018
- AMD reintroduces the Athlon name with two CPUs - Sep 6, 2018