Multi-Gigabit Wireless Link using W-Band Phased Array Chipset

[Hugoneus]

Hi everyone,

I have been busy with work and the MTT conference. Since I haven't had a chance to release a video, I thought I would let you know what I have been busy doing.

We had a live demo of a wireless link setup at MTT during our presentation which is capable of establishing a link (beyond) 32Gb/s at a carrier of 92GHz using two polarization. This uses the phased array chipset that I designed at Bell Labs. Each chipset has 16RX/4TX or 16TX/4RX phased array elements with up- and down-converters, integrated PLL, various calibration, self-test, etc. Two dies are flip-chipped directly onto PCB (die-on-PCB) with the phased array antennas integrated onto the PCB. This is no easy task at 92GHz! The EIRP at 90GHz is about 34dBm which is a record in itself at W-band.

I can try answering some questions as long as it is within released information about this project.

[R_G_B_]

Can you combine this technique with other diversity techniques to improve the speed much  further than 32Gb/s

R_G_B

[ConKbot]

Sounds very interesting! What sort of stimulation software is being used with this? I'm assuming this is to the point where even on die interconnects are controlled impedance? Any laser trimming or is the stimulation and fabrication good enough that it's ready to go after dicing? could on die antennas (patch?) Be a possiblity in the future? Hows the 3rd harmonic on the transmitter? 

[Lukas]

Some questions:
How do you get data in/out of the chips?
What material is the chip made out of?
What modulation does it use / what's the modulation bandwidth?
What's the phased array stuff used for? Spatial multiplexing/MIMO or just to get the radiation pattern right?
Was the the chip first time right?
Is there a paper/publication?

[Hugoneus]

Some answers to your questions:

The IC is designed in Cadence. It is a SiGe BiCMOS process with fT/fMAX of about 240GHz/270GHz.
The interface to the IC is baseband I/Q with up to 7.5GHz of bandwidth per side. (15GHz of RF bandwidth).
The modulation is adaptive and depends on the link quality. This is controlled by the modem ranging using various QAM.
Phased array is used for spatial beam forming.
There is a paper, it is published at RFIC 2015 which was held last month.

[Howardlong]

Am I right to assume that this is for point to point links and that the phased arrays are fixed, i.e. no clever dynamic beam steering?

I had enough trouble building WR42 waveguide filters connected to LO, mixer and 30dBm amp 15 years ago for narrowband K band, let alone the sort of stuff you're doing. I guess waveguide's a bit old fashioned these days.

[Dago]

Sounds tremendously interesting! I've always wondered what you do for work. I've also recently been very interested in microwave stuff (currently building a FMCW radar).

Do you have any pictures of the system/PCB and could you share them?

[Hugoneus]

Am I right to assume that this is for point to point links and that the phased arrays are fixed, i.e. no clever dynamic beam steering?
I had enough trouble building WR42 waveguide filters connected to LO, mixer and 30dBm amp 15 years ago for narrowband K band, let alone the sort of stuff you're doing. I guess waveguide's a bit old fashioned these days.

Beam steering algorithms are handled by the modem and often at the firmware level.

[Hugoneus]

Sounds tremendously interesting! I've always wondered what you do for work. I've also recently been very interested in microwave stuff (currently building a FMCW radar).
Do you have any pictures of the system/PCB and could you share them?

FMCW at what frequency?

Unfortunately the PCB photo isn't released yet, although I did show it during the conference presentation which is not distributed.

[coppice]

Sounds tremendously interesting! I've always wondered what you do for work. I've also recently been very interested in microwave stuff (currently building a FMCW radar).

Do you have any pictures of the system/PCB and could you share them?

Is FMCW (or FMICW) still an interesting kind of radar to build these days?

[Dago]

Sounds tremendously interesting! I've always wondered what you do for work. I've also recently been very interested in microwave stuff (currently building a FMCW radar).
Do you have any pictures of the system/PCB and could you share them?

FMCW at what frequency?

Unfortunately the PCB photo isn't released yet, although I did show it during the conference presentation which is not distributed.

It is for 4.2-4.4 GHz altimeter band. By changing the VCO it would be easy to use it for 2.45 or 5.8 GHz too (just need to redesign the bandpass filter and swap the VCO) but the noise levels are very high and bandwidth is very low on those ISM bands. Also have lots of stuff for a 10 GHz YIG-oscillator based one.

A pic of the PLL+VCO board (PLL not locked yet): http://www.dgkelectronics.com/storage/electronics/fmcw_radar/vco_synth/pll_vco_working.jpg
Hairpin bandpass filter: http://www.dgkelectronics.com/storage/electronics/fmcw_radar/hairpin_filter/hairpin_filter_v2.jpg

Sounds tremendously interesting! I've always wondered what you do for work. I've also recently been very interested in microwave stuff (currently building a FMCW radar).

Do you have any pictures of the system/PCB and could you share them?

Is FMCW (or FMICW) still an interesting kind of radar to build these days?

I've found it fascinating. Involves lots of system level design, RF design, analog design, digital (embedded) stuff, software work (embedded and desktop)...

Question for Shahriar: How long did this project last? Maybe even split to design phase, prototyping phase etc.

[daqq]

This area of electronics is a completely amazing and alien realm for me...

What kind of spacial/length/time tolerance can you afford in such systems?

[nfmax]

Are you using any sort of adaptation in the beam-steering, to cope with mechanical tolerances or temperature-dependent changes in phase shift, or is it a one-time set-up?

[Hugoneus]

Question for Shahriar: How long did this project last? Maybe even split to design phase, prototyping phase etc.

Great work!

The project took about 1 year from transistor circuit design to demonstration.

[Marco]

How much useful bandwidth remains if it has to use a bounce off another surface?

[Hugoneus]

How much useful bandwidth remains if it has to use a bounce off another surface?

A couple of issues arise. The coherent bandwidth may be affected as well as multipath may cause fading at certain frequencies. There is also the issue of polarization scrambling due to reflection. All of these effects would be addressed at the modem layer.

[vaualbus]

How much useful bandwidth remains if it has to use a bounce off another surface?

A couple of issues arise. The coherent bandwidth may be affected as well as multipath may cause fading at certain frequencies. There is also the issue of polarization scrambling due to reflection. All of these effects would be addressed at the modem layer.

So your team/you have also work on some algorithm for the modem layer?
The project is similar to the one you post a photo a while back on the blog?

[rfdave#gmail.com]

I saw you there talking to the Rohde & Schwartz guys!

[Hugoneus]

I saw you there talking to the Rohde & Schwartz guys!

I wish you had come and said hello!

[R_G_B_]

What type of forward error correction are you using?

R_G_B

[rfdave#gmail.com]

I saw you there talking to the Rohde & Schwartz guys!

I wish you had come and said hello!

You looked like you were having a good discussion. Maybe next conference, I'll introduce myself.