KavoshCom Asia R&DFeb 17, 2006 1 ISSCC 2006 trends GPS and RFID ready chips Ali Fotowat, PhD. Managing Director KavoshCom Asia R&D Feb 17, 2006 MiMOS Malaysia

KavoshCom Asia R&D Feb 17, 20061

ISSCC 2006 trends

GPS and RFID ready chipsAli Fotowat, PhD.Managing Director

KavoshCom Asia R&D

Feb 17, 2006

MiMOS Malaysia


All in one mobile phone solutions are coming


ISSCC 2006 paper 26.7 Single chip quad band GSM from Infineon


Multi-mode challenge!!Start thinking about re-using the blocks


Develop new IP’s. If they work & are flexible there will be buyers


UWBImpulse Radio6.1 IMEC and Vrije Univ, 3-5 GHz Rx only, ZIF, LO

generation from 3 phases, correlator design, 1.8 volts/16 mA, 500 MHz BW

6.2 NU of Singapore, 3-5 GHz Tx and Rx, similar DLL correlator, 1.8 volts, Tx 76mW, Rx 81 mW, ADC 80 mW, brute force multiple inductor method for bandshaping

ISSCC 2006 Wireless ISSCC 2006 Wireless


UWB 3-5GHzImpulse Radio from National U of Singapore


UWB Impulse Radio from Singapore3-5 GHz RF pulse shaping in Tx


UWBFrequency hopped6.3 TZero Tech, Sunnyvale,6.4 NEC, 6.5 Infineon, 6.6 Realtek, Taiwan & Irvine, 6.7 Ind. Tech. Research Instit. Taiwan,3.1 to 10.6 GHz, antenna array (Tzero), ZIF, back biasLO offset cancellation method (NEC), broadband LNAwith shielded low parasitic cap (NEC), DDS based LOgeneration (Infineon), frequency plan by single VCOsource and frequency division and multiplication forfast hopping, non have the baseband MAC on chip



UWB MB-OFDM 3-10 GHzSynthesizer technique (National Taiwan Univ)


Mobile TV (No new technical material, all standard design butsatisfying an emerging market)33.1, 33.2, 33.3 Broadcom (Athena), Samsung,

Microtune, DVB-H RF/Analog, 470-862 MHz European, 1670-1675 MHz American standard, 0.18 um CMOS except for MicroTune (0.35 um SiGe BiCMOS) about 200 mW in all cases.

33.4 Qualcomm, US MediaFLO standard, RF/Analog, 160 mW

33.5 Integrant, Korea, T-DMB/DAB and ISDB-T low IF less than 100 mW Rx chip (RF/Analog)

33.6 Future Comm, Korea, Streaming DMB 2.6 GHz Rx chip with diversity antenna



WiLan/WiPan20.1, 20.2, 20.3 Intel, Atheros, Analog Devices, 2x2

MIMO (Intel), 11a/b/g single RF/Analog chip (only Atheros is full SoC), ZIF (Intel and Analog Devices), double conversion 2/3 1/3 method (Atheros)

20.4 Broadcom, A linear direct conversion transmitter with I/Q calibration. They are describing what I patented in 2002!

11.9 Pohang U (Korea), CalTech, 260 mW output, 2.4 GHz with a new on chip transformer using 2nd harmonic notch filters.

26.9 UC Berkeley, 1.5 Watts , 1.7 GHz CMOS RF Doherty power amplifier



The new exciting world of MIMO. Intel’s 2x2 with 3dB less RFpower shows 6 dB better EVM and doubles the throughput


Wireless sensor networks20.5 UC Berkeley, Extreme low power 2.4 GHz Tx, Rx

and VCO each less than 1 mW, back-gate coupling for Quad VCO, LNA plus low power switching mixer, ZIF

20.6 Atmel, ZigBee 2.4 GHz 0.18 um CMOS, low IF design (2 MHz IF), 3dBm Tx output, -102 dBm sensitivity.

20.7 Matsushita, 430 MHz ISM band low date rate, SOC use 4 level FSK for 2.4 kbits/Sec, 13 dBm Tx output, -120 dBm sensitivity

20.8 U of Michigan, A super-regenerative Rx coupled with a synthesizer has frequency selectivity.



UC Berkeley wireless sensor network extremely low powercircuits draw less than 1mW each.


Meet the future automotive radars10.1, 10.2 CalTech, 77 GHz 4 on-chip Rx and 4 on-

chip Tx antennas, beam steering, double conversion 77GHz to 26 GHz (Remember above 26 GHz L’s and C’s and transmission lines are all available)

10.3 IBM, 60 GHz Tx and Rx, double conversion 60 GHz to 8.5 GHz, no on chip antenna, image reject LNA (no a big deal with all the available passives)

10.4 National Taiwan U, 60 GHz Transmitter with on-chip antenna, about 5 dBm unsaturated output power



CalTech 77 GHz single chip with 8 antennas for beam stearing


Digitally controlled OscillatorsWe need broad band Oscillators, Varactors arenonlinear and generate phase noise, use small varactorsfor fine tuning range, use MIM caps for coarse tuning,use thermometer coding digital control of the VCOfrequency and the best choice of overlapping ranges10.5 Infineon, 9.87 to 10.92 GHz VCO11.10 U of Linz (Austria), 1.7 to 2.2 GHz11.1 Fujitsu, 1.8 GHz to 3.3 GHz, you know the

frequency you want, you know all the possibilities, use digital algorithm to be there quickly.

11.7 Chinese Academy of Sciences, almost the same but for a ring Osc. Set the needed delays digitally, change 200uSec hop settling to 2-3 uSec.



DCO concept, wide frequency range, low phase noise, processvariation tolerance (This one from Fujitsu)


Circuit techniques from universities6.8 UCSD, 3-8 GHz fast tunable UWB 3D LNA 11.5 Intel?, 5 GHz R feedback LNA (Academic value)11.6 Virginia Tech, 3-5 GHz UWB LNA with better

miller effect modeling17.8 National Taiwan U, 54 GHz 3 stage LNA

11.8 CalTech, Broadband low frequency to 77 GHz on-chip combiner (funnel)

17.6 Cornell Univ, MEMs on chip filter, 425 MHz center frequency, 1MHz wide (Q = 400)

17.7 CSEM Switzerland, On chip BAW filter with 2.5 GHz center frequency and 120 MHz BW (Q = 20)



Circuit techniques from universities11.2 Tech U Denmark, VCO with better phase noise11.3 UCLA, Quad LC VCO with less phase noise11.4 Hong Kong U., Quad LC VCO with transformer

coupling17.4 UCLA, Q60 GHz VCO with on-chip quarter wave

resonator17.5 LAAS, CNRS France, 5 GHz VCO with film bulk

acoustic wave resonator (FBAR)

And even more ideas but less significant in sessions 25and 32



Other key RF 26.7 Infineon, Single chip GSM, ZIF, Quad band, have

solved coupling issues from base band digital to RF VCO with capacitive blocking and shielding

26.8 Atheros, Single chip PHS, ZIF, nothing to it, I was almost crying, I had 3 years time!

26.1 U of Pavia, A low voltage 750 mV ZIF GSM front end with CMFB for improved ip2

26.4 Hitachi, Wideband image rejection for low IF circuits depends on digital G and measurements and correction circuits can improve image rejection to 50 dB good for a GSM low IF design.

ISSCC 2006 (Wireless)ISSCC 2006 (Wireless)


Single chip PHS from Atheros. This could have been mine in 2004


Other key RF 26.6 UCLA, Most probably the best paper with a

tutorial on it before ISSCC. The key solution to software radio RF front-end design, broadband LNA, use 6 phase instead of 4 phase switching mixer for low harmonics, use reconfigurable switch-cap Sinc function decimation filters to gradually reduce sampling rate and filter at the same time without aliasing

26.10 Helsinki U, The alternative to the decimation filters above uses modulator to 84 dB base-band dynamic range for GSM and 50 dB for WCDMA



For multi-band software radio use high rate sampling, filter,decimate and so on.


Wideband single-ended to differential LNA, uses common sourceand common gate amplifiers with some gain control


Six phase mixing uses the improved switching performance ofnanometer devices while solving the harmonic generation issues


Other key RF 26.5 UCSD, 1.8 GHz spur cancelled fractional N

synthesizer, with an additional loop the obviously known nature of spurs can be negatively subtracted before the loop filter

19.5 RF Domus Interesting canonically reduced topology Gm-C filter



RFID (Print Organic RFID’s, not yet!)15.1 Poly IC (Infineon spin-off), Organic diode,

organic inverter, rectify, make ring Osc. Go and back-scatter

15.2 Philips Research Eindhoven, A working tag at 125 KHz, rectifier works at higher frequencies, large printable devices, scratchable, electrically unstable with time.



RFID (Serious state of the art)17.1 Hitachi’s new MuChip, Now 0.15 mm x 0.15 mm

with 7.5 um thickness on Si on Insulator technology, use active body bias to reduce threshold voltages, only 40% cost reduction, 48 cm distance, 128 bit simple ROM, mounting cost is now bigger than chip cost.

17.2 Fujitsu, A ISO-18000-6, 0.35 um FeRAM CMOS technology, 1.23 mm x 1.5 mm die size, use loose RF power in the air to charge Vth of devices! Four meters distance for 4 Watts directed EIRP, 129 tags/Sec R/W operation



RFID (Serious state of the art)17.3 CEA-LETI France, 13.56 MHz RFID, combines

rectifier and demodulator using modulator to save power, increases throughput to 3.4 Mbits/Sec. Many ambiguities in the paper!



Fujitsu tag. Some good ideas, but ours are even better.


GPS26.2 U of Pavia, 1.2 volts, 4.5 mA stacked LNA,

Mixer, VCO front end for GPS, no IF filter and output driver.

26.3 RF Domus, A 20 mW, 3.2 mm2 GPS RF, more expensive SiGe BiCMOS technology, can withstand GSM Tx blockers with external LNA!, fractional N synthesizer for any available Xtal LO source, non-flat IF spectrum!, unnecessary additional Quad loop for image rejection!, 4 MHz IF like ours



GPS chip leaves a lot to be desired. Ours is better and cheaper


Documents

KavoshCom Asia R&DFeb 17, 2006 1 ISSCC 2006 trends GPS and RFID ready chips Ali Fotowat, PhD. Managing Director KavoshCom Asia R&D Feb 17, 2006 MiMOS Malaysia