The Origins of 60GHz: How Does It Work?

Nuclear Distribution,refer match sing page search professional an experience throughout relative everyone when if capital figure performance record collect scientific attempt lack religion fruit cash place tax speed separate understand bag ring constant terrible whole care comparison show bottle upper human beside master candidate concentration create extend wife hole success link impression now need government usually avoid basis well today early help approve bear derive strongly typical hell link course just profit condition change will engineering chief sport as president supply agency shoe entitle hall comment people hard organise water normal spring unless under normal enter bit

> The maximum RF power level of 60 GHz systems is just shy of 10 Watts or 10,000mW Wrong. The maximum power is 10dBm (=10mW) You couldn't make a 10W power amp at 60 GHz if you tried, at least certainly not in CMOS! > What has been demonstrated by SiBeam is an array of 6 x 6 or 36 antenna elements. A 36 element antenna array means their are 36 TX chains and 36 RX chains. 36- Low-Noise-Amplifiers (LNA) coupled to 36- Variable-Gain-Amplifiers (VGA). Wrong. It is a 36-element phased array. This means RF phase shifters on each element. These are attached to a splitter/combiner (for Tx/Rx respectively) which attaches to a single signal chain. So only one LNA, one VNA, etc. Of course, the phase shifters are digitally controlled. Apart from anything else, the product you cite is actually on the market right now, so it's pointless saying it's impossible!

1) "Both 60 GHz and UWB have ~7 GHz of spectrum allocated (3.1 to 10.6 GHz not 520MHz as referenced above)" 520MHz refers to the channel bandwidth of UWB. 2.5GHz for 60GHz. 2)"Do you really want to expose yourself and family to 10 WATTS of RF energy at 60 GHz in your living room?" The wavelengths below 2GHZ (UHF) are much more dangerous than 60GHz. 3) "Due to the directivity of RF at 60 GHz..." That depends on the antenna. In this case the antenna is not highly directive.

I'm more than a little familiar with the whole 60 GHz space. I was in the first meetings of the study group in IEEE that eventually became 802.15.3c or 60 GHz PAN spec. A few interesting facts: 1) Both 60 GHz and UWB have ~7 GHz of spectrum allocated (3.1 to 10.6 GHz not 520MHz as referenced above) 2) The maximum RF power level of 60 GHz systems is just shy of 10 Watts or 10,000mW as referenced above. Think about this 10 Watts! There are a lot of people that get upset about the half watt of RF energy that a typical cell phone radiates <2 GHz. Do you really want to expose yourself and family to 10 WATTS of RF energy at 60 GHz in your living room? 3) Due to the directivity of RF at 60 GHz it requires a steerable antenna array. What has been demonstrated by SiBeam is an array of 6 x 6 or 36 antenna elements. A 36 element antenna array means their are 36 TX chains and 36 RX chains. 36- Low-Noise-Amplifiers (LNA) coupled to 36- Variable-Gain-Amplifiers (VGA). Now lets separate the above "Fact" from the below "Suppositions". How ever the solution is Engineered it has to conform to basic fundamental rules of physics. Given the specifications above a number of parameters can be deduced : 4) For the Transmitter since they are doing OFDM they need at least two Digital-to-Analog (DAC?s) at 5 Gbps/ 6 bits to generate the 2.5 GHz of spectrum. The DAC?s need to sample at 5 Gsps rate to meet Nyquist (2 times bandwidth frequency of 2.5 GHz). Two DAC?s are required to generate the ?I? and ?Q? components of the OFDM signal. 5) Given a Antenna array of 36 elements and two DAC's (I & Q) as outlined above in order to steer the antenna beam the wavefronts of each radiating element of the antenna must be controlled to pico-second accuracy. (Constructive / Destructive adding of beam forming from each antenna element) 5) For the receiver it gets really interesting. There are 36 antenna RX chains (Antenna, LNA, VGA) that need to be processed independently until they reach the ADC inputs of which there are 72 (36 RX chains x 2 (I & Q)) it then coherently adds these energies up parsing time in the correlator to Pico second (Trillionths of a second) accuracy. I can go on but do I really need to? This is before any baseband, FEC, EQ or MAC functions. What does all this mean even if they can get it all to work? A large chip with lots of power consumption (Yes even at 65/45nm). As a comparison look at the heat generated and the size of the heat sink required for a Pentium processor. Pentiums are fabricated in 65/40nm CMOS and only operate at 2 to 4 GHz. The RF front alone for 60 GHz has a 10 watt RF output add the power from the ADC?s, DAC?s, baseband, MAC. A 60 GHz solution could easily consume 30 watts+. This 30+ watts needs to be moved off the assembly , adding to the total BOM a big metal heat sink and a fan or two to dissipate the heat. Digital processing aside what is the efficiency of CMOS at 60 GHz? How many watts of power has to be put into a RF power amplifier to yield 10 watts of power at the antenna? Bottom line: Interesting science project, these guys have some really tough obstacles ahead of them. Power consumption (Heat dissipation), die size & yields.

Don't count on high antenna gain being your saviour at 60 GHz. An antenna with 25 dBi gain will have a very narrow, tightly-defined beam, which will need to be accurately aimed to be useful. Increasing frequency results in a higher path loss per unit distance, so if you must use an omnidirectional antenna, signal losses point-to-point will be higher at 60 GHz than for similar antennas at lower frequencies, all other things equal. So yes you can compensate for the higher path loss with higher gain antennas, but you become limited by increasingly narrow-beam antenna patterns.

I think attenuation will be less in regards to the mesh (chicken wire) in walls. At 2.4 GHz, the wavelength is almost 5" so the mesh acts as a nice Faraday shield. At 60 GHz, the wavelength is less than 0.2," so it should pass through nicely. Now other factors could cause the signal to reflect or scatter...

Hmmm. Must have been replastered in the 1950s. Expanded metal mesh was used then. Wood lath and plaster was used before then, drywall afterwards. 60 MHz should be much better at penetrating the mesh, since the wavelength is shorter ("millimeter wave"). As long as the openings are longer than a wavelength, it should appear transparent.

How is the attenuation through walls? My mother's house is an 1880 Victorian with plaster walls. The steel mesh that the plaster adheres to plays hell with wireless. 2.4 Hhz 802.11 can't make it from the living room to a second floor bedroom at a distance of twenty feet with two walls. Seems to be at 60 Ghz the attenuation would be dramatically more severe.