Yesterday, the MIPI Alliance announced its new interface specification for RF front-end (RFFE) control. I contacted the MIPI Alliance, and Jori Arrakoski, Senior Architect at Nokia was kind enough to answer my questions.
Yesterday, the MIPI Alliance announced its new interface specification for RF front-end (RFFE) control. The spec takes aim at the problem of control congestion in the RF front end, and includes power amplifiers, low noise amplifiers, filters, switches, power management modules, antenna tuners, and senors, whether discrete or integrated. The press release for the spec and some other information can be found by following this link. But after reading all of that, I still had some questions. So, I contacted the MIPI Alliance, and Jori Arrakoski, Senior Architect at Nokia was kind enough to answer my questions.
What is the history of the origin of this spec? Who were the key drivers?
Cellular radios (in terminals) have become more complex with the need to support multiple radio access standards simultaneously as well as bands. For example, the Nokia N8 incorporates GSM for four bands and WCDMA for five bands (as compared to multi-mode terminals a few years back when even a two-band WCDMA radio was state of the art). The more options, the more there is to control in the front-end devices especially with all the time tightening performance requirements. It seems that front-end devices are becoming more like the main RFICs a few years back, there is a multitude of settings and control to be done to tweak out the required functionality and performance.
The traditional way to control the front-end devices was using dedicated control signals. But a few signals or wires per device soon became far from enough so people started looking after more pin-efficient control methods, i.e. serial buses. There is a huge amount of various serial buses available, some companies even developed their own. Suddenly the industry realized that this can’t continue as the front-end device vendor barely could find more than a single customer (due to compatibility) to its part and the system integrators had difficulties finding compatible parts. There were a lot of discussion among the companies that a standardized control interface would be desirable. One concrete result was that IWPC set up a task force gathering key players and developed a requirements document. Many participants were MIPI members. MIPI, as a well established standardization organization with the right scope, was a quite natural home for the activity to standardize the control bus.
The RFFE IG (Interest Group) was established in August 2008 and became a WG (Working Group) in November 2008. The first task of the WG was to investigate the multitude of already existing interfaces to see if there would be fit. A special feature necessary in the cellular radio application, namely time critical control, played an important role and we had to turn down one candidate after another. Finally the best fit was to use the MIPI SPMI specification as the baseline and tweak it to incorporate the mandatory RF front-end control features. RFFE was born.
Key drivers in the beginning were Nokia, Infineon, Skyworks, RFMD, Renesas, Panasonic, Freescale, Motorola, ST-Ericsson and National Semiconductor (in no particular order). A few of them have later cut down their participation in the WG whereas other companies have joined. (Note, the following companies were listed on the final spec: Agilent Technologies, Inc., Broadcom, Fujitsu Limited, Infineon Technologies AG, LG Electronics Inc., LnK, Motorola, National Semiconductor, Nokia Corporation, Panasonic Corporation, Qualcomm Incorporated, Research In Motion Ltd, Renesas Technology Corp., RF Micro Devices, Rohde & Schwarz Gmbh & Co Kg, Skyworks Solutions, Inc., ST-Ericsson, Synopsys, Inc.,Texas Instruments, Inc., WiSpry, Inc.)
We have had multi-radio systems for years, so what, specifically, is the challenge with the previous system that this new spec addresses?
I agree that we've had multi-radio systems for years but the complexity of the configurations has been continuously increasing with very challenging designs just around the corner. There would be no issue to continue the game with just adding control signals and wires unless tightly constrained by cost and size requirements. There simply comes a point when pins become very costly -- not to mention the extra space needed by the larger packages to incorporate all the pins and all signal routing challenges. A control bus is simply the only viable solution forward.
How will the trend toward integration in the RF front end be effected by this spec? Some RF front-end chip vendors are already integrating control on chip, how will this play?
RFFE actually enables that integration to some extent. We’re talking about combining functionality into larger entities. The more these units do, the more you need to tell them what to do since an on/off switch is not enough anymore. At some point these units start to have logic (or say intelligence) and they need even more sophisticated ways to communicate with the other devices. I would not be surprised to see even small processor cores in future front-end devices. And how do these talk to others? They use buses and communication protocols. RFFE is able to address these needs, and it has growth potential to meet the needs for the next five years, perhaps even longer.
I’ve been following the RF front-end development for almost a decade and it strikes me that although we’ve constantly been talking about increasing the integration level, the progress seems minimal if you count the number of front-end devices in your RF solution. One could even claim that the progress has gone in the other direction with new devices like SMPS and antenna tuners bringing up the number of devices. The explanation is very simple. The cellular radio requirements simply grow in a pace equal to how fast we manage to integrate. There has been a tremendous step in integration level if comparing radios with similar requirements. Take a basic phone of today and compare it to something state of art 10 years ago, you’ll be surprised how similar the features are and how huge the build differences are.
So, again, many thanks to Jori. Now I (Janine) am wondering if you have any questions? Please sound off in the comments section: do you think this spec will help you with your designs? Were you a part of its creation? What do you think about it?