Wednesday, September 21, 2011

IBM to Publish 2 GHz Graphene IC at Upcoming Conference

IBM is set to publish a paper on a 2 GHz frequency doubler RF circuit in a CMOS-compatible manufacturing process technology at the upcoming International Electron Device Meeting, due to be held in Washington DC, Dec. 5 to 7. IBM researchers will deliver a paper that is a significant step toward moving graphene from the lab into a manufacturable technology. It will detail how using a 200 mm wafer-scale CMOS-compatible fabrication process can be used to make high-performance graphene FETs and RF passives.


A major obstacle with graphene is the difficulty of building a gate dielectric (insulating layer) on its inherently inert surface. However, graphene layers grown by controlled vapor deposition (CVD) can be transferred to many types of substrates. To take advantage of this property, IBM built silicon wafers containing pre-defined embedded gate structures, and then transferred CVD-fabricated graphene layers onto them. As an example they built a frequency doubler which demonstrated a conversion gain of ~-25 dB at an output frequency of 2 GHz. This performance was nearly constant from 25-200°C, indicating that both n- and p-transconductance are temperature-independent in this range, a new finding for CVD graphene-based devices.



The four images on the right show (a) an 8” graphene FET wafer; (b) single die; (c) SEM image of a typical fully processed device and (d) an enlarged view of the device showing the embedded gate structure with two-finger design. Except for the CVD graphene transfer, all processing was done in a conventional 200 nm fab. Graphene technology is finally making it into production.

Monday, September 19, 2011

New RFIC Greatly Mitigates LightSquared LTE Interference with GPS

There has been a lot in the news the last few months about LightSquared's proposed new broadband LTE network which is close in frequency to the GPS band and could cause interference in critical areas of navigation systems. LightSquared is seeking FCC approval for 40,000 basestations to support 260 million users across the country establishing a new LTE network. They were approved in Jan 2011 but recent tests have indicated that signals from the network could interfere (jam) nearby GPS receivers so the FCC has said they cannot launch their network until the problems are resolved.

LightSquared has indicated they will adjust the direction of their signals to minimize the strength near GPS stations and move the frequency a little farther away. There is a 10 MHz block near the GPS frequency that they will not use and move to another block of spectrum currently used by Inmarsat. They will also reduce the maximum power levels to provide additional protection.

But today Tahoe RF announced an RFIC solution to the problem with an integrated dual channel (L1 & L2) GPS RFIC that substantially mitigates interference from LightSquared and 4G L-Band LTE signals (and other jamming environments). The RFIC also includes a fractional-N synthesizer with a high performance VCO. The receive paths can be configured high linearity or low power operation by setting the ADC bit rate. The two independent receive paths are integrated with 12 bit analog to digital converters providing complete conversion of the GPS signals from RF to digital data. The RFIC has the ability to process L1 and L2 received signal data in the presence of a greater than 60 dBc jammer.

GPS devices are being improved against the backdrop of various 4G and other signals that can interfere with the relatively weak satellite signals with improved filtering and sensitivity. Infineon recently announced highly integrated GPS/GLONASS modules with improved sensitivity plus pre- and post- filtering around the LNA. These modules have out-of-band rejection of greater than 43 dBc in the cellular bands.

This is a very exciting area as another company is trying to bring broadband to the rural market around the country, something the Obama administration has been promoting and supporting for a while but has not seemed to meet expectations. Can all these signals co-exist as we squeeze more and more data into our limited spectrum?