Qualcomm Announces Fourth Gen Cat 6 LTE Modem – MDM9x35by Brian Klug on November 20, 2013 2:47 PM EST
- Posted in
In addition to the IPQ8064 news this morning, Qualcomm released a host of other product announcements we're working our way through. Among them is news of its fourth generation multimode LTE baseband, MDM9x35, which is a Category 6 part capable of 300 Mbps on the downlink and 50 Mbps on uplink. Getting to 300 Mbps of downstream throughput requires aggregation of two 20 MHz wide LTE carriers. By comparison MDM9x25 which we haven't seen quite yet in discrete form (although it is the IP block inside MSM8974) is a Category 4 part capable of 150 Mbps of downstream throughput on either a single 20 MHz LTE carrier, or aggregation of two 10 MHz carriers, and is
built on TSMC's 28nm LP process
Update: Qualcomm's third gen LTE modem, MDM9x25, is built on 28nm HPm. Qualcomm also notes its MDM9x25 has shipped in two discrete WiFi hotspot products. Also just like previous generations, there's a version of the fourth generation LTE modem suffixed with M for a package that includes stacked memory, and sans M, 9x35M includes memory, 9x30 does not (yes it's 9x30 not 35 for the version sans stacked memory).
MDM9x35 is built on TSMC's 20nm SOC process, making it the first publicly announced product on Qualcomm's roadmap to use 20 SOC. It'll be interesting to see whether the switch to 20 SOC goes smoothly for Qualcomm, and just how much volume there is.
MDM9x35 inherits all the legacy air interfaces below it that you'd expect, including DC-HSPA, EVDO Rev. B, CDMA 1x, GSM and TD-SCDMA. In addition the modem adds dual carrier HSUPA for faster uplink speeds, and aggregation of even more carriers on the downlink across more bands.
As part of the announcement there's also a new transceiver, WTR3925 which is Qualcomm's first single-chip carrier aggregation solution, confirming my suspicions that WTR1625L and WFR1620 were both required to achieve aggregation with the MDM9x25 solution. In addition WTR3925 is built on a 28nm RF CMOS process, a significant jump from the current 65nm RF CMOS process used in WTR1605 and WTR1625L.
Interestingly enough Qualcomm claims that MDM9x35 will be available for pairing (Fusion) with Snapdragon 805 which was announced today, in addition to MDM9x25. Qualcomm notes that MDM9x35 and WTR3925 will be available to sample early 2014.
Update: One more addition to note is that although Qualcomm won't disclose the port configuration (High/Mid/Low band primary Rx/Tx) for WTR3925, they have noted that it will support all the currently published 3GPP carrier aggregation combinations across all bandwidth combinations. In addition MDM9x35 is 3GPP Release 11 capable and supports all release 11 mandatory features.
Post Your CommentPlease log in or sign up to comment.
View All Comments
DanNeely - Wednesday, November 20, 2013 - linkHow much will the process switch on the transceiver help (or is it just TSMC wanting to shut down the old line)? I thought that part was mostly analog components; and that unlike digital, analog parts don't shrink well with new processes.
metafor - Wednesday, November 20, 2013 - linkAnalog is more difficult to implement as components get smaller; that's what is meant by "didn't shrink well". If you are able to achieve the same levels of gain, noise and linearity required at a smaller process, you'll still get significant power reductions.
Hrel - Wednesday, November 20, 2013 - linkWhen do we get Gbit/s speeds wirelessly? Preferably with 1 tower reaching 100km with 10 Million connections. Cause that's what we're gonna need to blanket the US in quality internet.
jchambers2586 - Wednesday, November 20, 2013 - linkwhy do u want that fast speed for? The data caps make it useless.
Doh! - Wednesday, November 20, 2013 - linkthere is a whole new world outside of your world where data cap is not an issue or does not exist.
errorr - Wednesday, November 20, 2013 - linkIt is all about power consumption. Just like faster processors allow race to sleep speeding up the data transmission helps keep the power consumption lower and battery life's better.
Margalus - Thursday, November 21, 2013 - linkwhat data caps?
DanNeely - Wednesday, November 20, 2013 - linkWhen the laws of physics are repealed. LTE is already approaching the Shannon information density limits. To get higher capacity you need to either increase the amount of RF spectrum used (which is what this update does); or space the towers closer together (this is called expanding urban capacity).
While LTE does support cell sizes as large as 100km line of sight issues mean it's only viable in areas that are as flat as a pancake; while the larger cell sizes actually have a smaller total capacity than the smaller ones both because the data takes longer to travel back and forth and because less dense encoding and more ECC are needed to keep the data from being lost in noise. For phone applications as opposed to wireless residential internet the increased transmit power needed on the phone/hotspot will reduce battery life.
If you want wireless internet with the same speed and per user capacity as wired internet it's going to take going from having a cell tower every few miles in suburban areas to base stations every block or so (although they probably won't need to be as tall since long distance line of sight will be less of an issue. Probably they would end up piggybacked on the big boxes your wired internet providers are already scattering all over the place.
nafhan - Wednesday, November 20, 2013 - linkA tower with 100km range runs counter to the goal of "high speed connections for everyone" for at least two reasons: 1: there's a fixed amount of spectrum. Shorter range, means the same channels can be used by many physically distant connections. 2: transmit power for the phones. Long distance + low power = very problematic SnR situation (if you want reasonable throughput).
DanNeely - Wednesday, November 20, 2013 - linkThe huge maximum range that LTE can be configured for was added because while not suitable for general usage it's beneficial in a limited number of circumstances. With Verizon having largely blanketed the midwest in LTE already, I suspect the main area it will be deployed is along the edges of flatter parts of the Australian outback where it can push relatively low latency internet out into areas that were previously only serviced by satellite connections.
In theory something similar could be done to provide connectivity to ships traveling along coast lines but beyond the range of standard cells; but since it would be competing with land based services with much larger numbers of customers for spectrum I doubt it will happen.