In modern smartphones the bulk of volume is occupied by battery. There is antenna which could be on the right side. There are processor, memory, graphic chips, camera module... What is left - is the small area in the corner for RF end. Normally there is RF chip and discrete components. Even if you like to support maximum number of simultaneous RF bands in a transceiver an OEM typically can’t do that because they can not fit power amplifiers, antenna switch, duplexers, LNAs - all that discrete components. They run out of space. It is big problem everybody is complaining about. Because there is 30 bands in LTE and 10 in 2G/3G.
So what we did? We came up with front end solution that takes the antenna switch and a power amplifier across 2G, 3G and 4G - single PA for the three standards, multiband, and antenna switch integrated in IC mass. And that chip gets packaged in the bottom part of our RF POP solution. On top we put the duplexers and SAW filters - first time in RF world! All components, PA, antenna switch, duplexers, SAW filters integrated into RF POP reduced area you need on PCB by 50%. Now you can add to the platform more discrete elements if you need for example maximize number of RF bands supported. And you can now develop an LTE global phone.
WTR1625L based on the same process - 65 nm, and it has the same footprint as the previous transceiver, but twice the number of bands. And there is also a wafer scale package - smaller than traditional plastic packaging. Package size is the die size (chip size).
And that is not an issue... If you like to support Cat.3 or Cat.4... If you look to about hundred of operators that deployed LTE all over the world. It is only about 10% of them has 2x20 MHz band of spectrum. And to get Cat.4 work with high peak rate you need carrier aggregation. And it is big complexity in a modem and on the RF-side. We were first to demonstrate carrier aggregation on a real chip - MDM9225 and it was integrated into handset chip MSM8974 Snapdragon 800. We have two chips already demonstrating carrier aggregation. And one of them is already in commercial product, there is MiFi router based on it. And Ericsson use it within their commercial infrastructure. This is key enabling features. Not enough to say “I have Cat.4 chips in our pockets”. To serve most global markets you need carrier aggregation. And our solution is even more important for the carrier aggregation because you now can multiply number of bands by two and to have combinations of bands suitable for carrier aggregation. The complexity really start to grow. And you still stack with that tiny little board area to fit all the band support for carrier aggregation. That is why it is really critical that our solution save about 50% of board area - now you have space to support carrier aggregation and global LTE in the same time in a smartphone.
We got third generation of LTE product - Cat.4 + carrier aggregation which we demonstrate now.
Anyone who claim he have product you can ask to show those functionality. And ask to show you the voice support - SCFB across 2G and 3G and VoLTE.
We got VoLTE demos here on MWC, we commercialized voice support - 10 different voice modes: dual-carrier (1x + LTE), CSFB for GSM, CSFB for UMTS, CSFB for 1x, VoLTE single mode, VoLTE with SRVCC to GSM, VoLTE with SRVCC to UMTS,... - all this modes are commercial.
PC: We have healthy respect for the competition, we always paranoid about competition. Issue is from the time of 3G arrival standards are keep moving forward and getting more and more complex. List of features which you need to support as a minimum keeps piling up. It looks great on paper when you hear “Hey, I am catching up!” But reality is - we have had several year and several generations of product. testing every feature, checking the interoperability. This is the foundation of standards today. If you want to compete, you need to go same way from the day one. You can not just to support compatibility with LTE, you need to support 2G and 3G as well. All the voice support as I mention. And the complexity keeps growing. It is not just growing data rates, it is more and more services, like LTE Broadcast. VoLTE is a very complicated feature, now we are talking about HD voice, presence, video telephony. List of features to continue growing up fast. And if you are talking about Cat. 4 support, that is expecting that you product support anything else as well. And you have to have very robust optimized performance in which we perfecting over the years. Each generation of the product we got an improvements. LTE broadcast - we got product, made trials and demonstrations. We keep making enhancements, we expect first launches at the end of the year or next year. VoLTE is already commercial. CSFB is commercial for two years. Performance is getting better and better. So it is not about to say - I can support 100 Mbit/s rate.
I fully expect some of the competitors, quite a few of them will have products to announce. Not all delivering their promises. And as the result they will be some consolidation on the market.
If we talk about one of the competitors you mention earlier - it is quick, but it is quick and dirty. Ny definition of “quick and dirty” is - you support most of the features, but you forgot about power consumption, heat - such a product able to work as Cat.4, but will be very very hot. We test such product - it can’t support data rate when it gets hot.
And we know that their active power consumption is more than double comparing with our MDM9x15, and consumption in standby mode is 4.5 times higher! If you don’t care of such thing, you can claim you have product. Even if they declare that product as commercial, that is not commercial quality. That is what I mean as “quick and dirty”.
AB: Can we expect, lets say any revolution in power consumption of the LTE chips?
PC: Energy consumption is the hard problem for all LTE products. But when we did competitive testing in that area we found our big advantage - active mode power, standby power... For 3G we had time to optimize hardware implementation and have much better results in compare with software defined platform based on general architecture. You can’t optimize it for power consumption. You can only optimize for one process mode. What we have done - we kept in custom silicon the functions that are more power and space efficient, like decoders, and other functions we left in firmware. We got 30% of active power reduction with the processes 25 nm and 28 nm. We were the first with the commercial implementation connected DRX feature (discontinuous reception) for LTE when you stay in connected mode, but stop transmitting and receiving. That is 30% reduction of active power consumption as well. And active mode is most important, because people use their devices much more than few years ago. So battery tense begin to dominate now by active receiving. Just few years ago it was standby power domination. There are many more things we did to improve power consumption - some of them small steps but a lot of them. For examle DRX feature. When you use DRX for VoLTE to save power it is even harder because you only have 40 ms to ramp up you power amplifier and shut it off. You have to perfect this things. And it takes enormous amount of efforts, very sophisticated algorithms, system level interaction. We got all the pieces ready to create such a solution, we did optimize it all together. And when we integrate it into the single chip as we did with our MSM Snapdragon products we got power advantage. The other place is not just to integrate the modem because you need to have less inter chip communications we also integrate WiFi, baseband, BT, so we remove one addition chip as well.
AB: You prefer to keep use only Silicon in you chips? Or you have plan to use gallium arsenide?
PC: We don’t plan to use gallium arsenide in our power amplifier. We compensate that performance with our end-to-end system, that’s controlled by the modem. This PA in this product is actually Si mass based. And the reason is that we have cheaper solution, than those which is based on GaAs. We use envelope tracking and entire system optimization. The modem actually keeps feed PA with the control information for the envelope tracking. It is technology close to ideal - we can use modem information to dynamically change voltage in real time to the exact data envelope. There is no wasting energy technology. And by this we can reduce power consumption from the amplifier by 30%. And of course heat dissipation problem solved as well. Si mass PA is much more practical to manufacture.