Hacker News new | past | comments | ask | show | jobs | submit login
Low-cost 60GHZ phased array antenna (eenewseurope.com)
208 points by tiagod on March 4, 2020 | hide | past | favorite | 75 comments



You can buy a set of pre-paired (configured) devices from Microtik for $195 that are 60Ghz 1Gbit full duplex that don't require any dish. I've used them, and they work well. (https://www.amazon.com/dp/B077992GG3)

The difference here seems to be a substantial increase in distance. 1.5km is a lot more than the 100-200m promised by Microtik.

Also, yes, any 60Ghz device is going to be line-of-sight only.


The "wireless wire" are too short range to be useful for many things. A better example of an 802.11ad product shipping right now is this:

https://mikrotik.com/product/lhg_60g

Or this, which is $299 per end:

https://store.ui.com/products/af60-us

If you're going over 300-400 meters with these you need to keep in mind that the 60 GHz band sees EXTREME rain fade. People try to build links that are too long with these and have no fade margin. RSL should be something like -40 to -42 when properly aimed, which gives you about 20dB of fade margin.

Sure you can get a radio set to link up at 1.5km+ at -60 on a clear sunny day, but the slightest rain shower will have adverse affects on it.


With 60GHz it would take a storm with a rate of rain approaching 55mm (2") per hour to exceed a 20dB fade margin. That is quite a severe rain. A slight rain shower should not significantly impact a 1.5km link at 60GHz.


In real world use if you want five nines availability of a 60 GHz link in Portland, OR, you need to keep the distance to about 700 meters or less. And it doesn't see really heavy rain showers (as Miami or Houston do), just constant drizzle. 300-400 meters max in Miami.


That's why lots of options are shipping with 5ghz failover.

Still not the best case to reduce bandwidth during heavy rain but at least it still works.


In central Kansas we would see rainfall exceeding that rate multiple times a year. I've seen it as high as 12"/hour for a few minutes.


Oh. My. At this point, I suppose you'd need Wifi for your boat? :-)


12"/hour is 0.2"/minute, so that for a couple of minutes doesn't make for ridiculous overall rainfall. Signal quality is affected by the current momentary rate, and having the signal drop for seconds or minutes is problematic.


It's not so much the rain fade as the oxygen absorption that really gets you with 60Ghz[1]. You can get better distance and reliability with 70/80Ghz. They're expensive though.

[1] https://www.rfglobalnet.com/doc/fixed-wireless-communication...


brisbane has 80-90% humidity and temps around 30 a lot of the time. what would persisting moist hot air do, to your rainfade?


The wireless wire is intended for indoor use, where you can move it around. Here automatic beam forming is more appropriate than a dish which means fixed installation requiring precision alignment and very narrow main lobe.


They are waterproof and use PoE for power. They appear to be designed for indoor or outdoor use. I'm using them outside, as are many other people.


Beam forming is still good for outdoor use if it is cheaper. Dish is a significant part of the installation cost. In some places it is not easy or even possible to install a dish (for example, it might be prohibited).


Yeah, but both of those devices are dishes. What we're discussing is what's possible without a dish.


1.5KM makes 'wiring' a village possible from one relatively high structure, such as a church's bell tower.


Line of sight truly means line of sight with 60GHz. As a human standing in the way is enough to mess with the signal so would a flock of birds I guess.


Do flocks of birds regularly hover in one place for an extended period of time? Coming up with increasingly bizarre limitations here.


They could fly down the beam ;)


Or better yet, oscillate along the path of the beam. And they would naturally have another flock to swap out when they tire plus another flock available as a hot swap in case something unexpected happens. :P


A village that small and unwired, rural, third world, would be happy with that line of sight limitation.


I'm thinking this is primarily point to point, not suitable for showering an area in sweet bits.


It is likely MIMO, I don't see any obstacle why it couldn't do multiple spatial channels simultaneously.


It also requires clear weather, so won't be enough to "wire" a village with reasonable expectation of service quality. At which point there is little reason to go with anything, but a passive optical network in a village.


It depends on what you mean by reasonable. In some places anything that faster than the non-existing 4G-connection they were promised 5 years ago will be reasonable.

These high frequency links might not be suited for bad weather or flock of birds, but they do enable low latency, high throughput most of the time. You can always downgrade to a lower frequency, much like how 5G operates, completely transparent for the clients.


People are not isolated from the rest of the world into specific areas, they know what broadband internet is like and this sets expectation of quality. I know of some organizations that use links on that frequency and not working in rainy, snowy or foggy weather always comes up, but they can deal with that, while consumers can't.


In good weather...


Well for PTP have a fallback 2.4 Ghz with a pair of polarised Yaggi's and jack the power up.

What I mean by having a pair is have one channel vertically polarised and one channel horizontal double bandwidth


Not line of sight only. Many recent 60Ghz devices with better beam forming also work acceptably well using reflection off solid surfaces.


That's... actually impressive. I've built a point to point WiFi link before, with 802.11g.

I was under the impression 60GHz is for extremely short range applications, like wireless connectivity in a living room and such.

Really surprised it works that far, even if it needs LoS and perfect weather.



Using 60 GHz for communication at these distances is pretty strange because 60 GHz is strongly attenuated by the atmosphere. Not only that, but atmospheric attenuation peaks at 60 GHz because that's what oxygen absorbs[0]. Some military satellites even use 60 GHz for intersatellite links because it's highly attenuated by the atmosphere, making eavesdropping more difficult

[0]https://en.wikipedia.org/wiki/Extremely_high_frequency#Propa...


The next generation of RF communications always clusters around RF spectrum that the previous generation considered undesirable. From shortwave radio to 2.4 GHz, this has always been the case. It's not a huge stretch to suspect that it might be the case for 60 GHz as well.


So what would the most desirable frequencies be for long range ground based communication?


Probably VHF.


The high attenuation at 60 GHz is actually an advantage since it lets you put lots of highly directional antennas in close proximity while attenuating their side lobes to lower interference. If your attenuation is 20 dB/km but you have a 22 dBi gain antenna, a 1.5 km range seems pretty feasible.


Yeah, this is also why 60 GHz FCC power limits are higher because they don't expect you will interfere with anyone. I thought it was usually thought of as an "in room" transmission medium.


Agreed. 60GHz appears to be allocated for fixed inter-satellite communications (probably for this reason):

https://transition.fcc.gov/oet/spectrum/table/fcctable.pdf


I read some of the above PDF... what's amazing to me is that a whole bunch of names of foreign countries pop-up over and over again in this document, specifically with assigned frequency ranges, as if the FCC is regulating frequency ranges in foreign countries... Before reading this, it seemed self-evident (to me) that the FCC's regulatory power does not extend to those foreign countries... Although, perhaps the FCC's regulatory powers do in fact extend to those countries via international agreements and/or treaties?

Does someone know?

How does that work?


If you re-read the document you will see that there is a note regarding the international frequency ranges indicating that they are taken from the 2016 edition of the ITU Radio Regulations.


Interesting! So ITU is sort of the international standards organization for radio frequency bands... makes sense!


Since it’s in the 60GHz range, that means it would basically require line-of-sight to function, right?

Not downplaying the significance, but I’m not sure we’ll get WiFi at 1.5km just yet because of it.


These types of devices have existed for a while with a dish at least. You're right, at this frequency you need LOS and there's little forgiveness there.

Current products don't really intend (to my knowledge) for you to connect a client directly to this antenna, rather they're meant for both to be in a fixed location properly aligned to each other, then they go to the access points for the clients.

Like the other's mention, this is useful for many environments with multiple buildings, easy example is a shed out in your backyard.

Obviously if you could run a copper/fiber connection out directly that'd be better, but these are for situations where you can't.


Could be useful for backhaul though for mesh networks outdoors


Something something rain and fog.


Well 2.4 is effected by rain - I think with 60Ghz you would cut the trees down


You could use it to extend range to a large campus, stadium, race track, farm or even between two office builds in close proximity. It’s somewhat niche but there is a use case for short line of sight connection.


Says low-cost but anyone knows what exactly is "low cost" here?


With the process technologies they're using, I'd guess 10s to 100s dollars.


For a phased array antenna?


Sure, why not? It depends on the number of elements, of course, but at these frequencies, you can fabricate a LOT of elements for low cost because they're small. Perhaps even chip-scale integration is possible. $1/element is feasible for phased arrays.


The antennas are practically free; the cost comes from phase shifters, amplifiers, etc, which don't see much miniaturization benefit at high frequencies. One day $1 per element is feasible, but definitely not today.


For the antenna yes. Doesn't included the associated electronics to drive it.


I'm betting the front-end chip is all CMOS, as opposed to SiGe, so it will be quite cheap (under $10 in volume). 16 TX elements at a total of 20 dBm transmit power is 8 dBm/element, which is feasible in CMOS. State of the art in CMOS is 9 dBm at 80 GHz.

The antenna will be cheap too as there are number of spread spread glass laminates out now; I have used Rogers RO1200 for mmWave work. The only thing that may make it expensive is the air-gap construction, which they have probably done to increase the bandwidth. I would assume two laminates are needed.


It can achieve that without a dish because (correct my math if I'm wrong) a full wavelength antenna for 60GHz is 0.1968 inches. So I'm pretty sure the silica includes the antenna.


The antenna is built on a separate substrate, I'm guessing a spread glass such as Rogers RO1200. You can do the aperture calculation here, which for a square aperture at 100% efficiency is 18x18 mm. It's probably more like 75% efficiency.

https://www.everythingrf.com/rf-calculators/effective-antenn...


This isn't all that new - I've lost the link to the press release, but this thing will also do 1Gbps+ at 1km: https://www.siversima.com/wp-content/uploads/product-brief-t...


How does that work when no Silicon runs at 60ghz? What can process the signal?


You are thinking of big CPU clocks, this is not the speed of a single transistor switching, it is determined by whatever was accepted as the longest path of components for electricity to propagate through (think of a ripple adder as a simple example). You can have a much smaller separate clock and thus higher frequency for single purpose silicon when the path is significantly shorter.

60GHz here is talking about signal, you only need to create an oscillation at 60GHz not do fp multiply. There are various ways of doing so, but consider that it only takes three inverters to create a ring oscillator. As for the data - you don't need to fill the buffer at the speed which bits are sent, which is how you can interface with a lower frequency microcontroller, i.e the buffer is filled at a lower frequency but in large chunks i.e words.

Someone with proper EE and signal processing knowledge will explain more accurately but that's the crude idea (components frequency vs clock).


Not sure if this is what you are getting at but high speed comms modulate a signal up from a lower frequency to a much higher carrier frequency for transmission. Then the receiver modulates back down to baseband (the original frequency before modulation). This can be done with discrete hardware, no need to do the DSP with a CPU that can't reach those clock speeds. It's one of the reasons your 5GHz WiFi doesn't actually have a 2.5Gbps bitrate (satisfying the Nyquist criterion), but something closer to 1.3Gbps.

Source: graduate with MS EECE in Comms, Control, and Signal Processing in a couple months


> It's one of the reasons your 5GHz WiFi doesn't actually have a 2.5Gbps bitrate (satisfying the Nyquist criterion), but something closer to 1.3Gbps.

Ah, this is not what I meant but just as relevant. I think you are essentially saying that you don't have to send as many bits per second that the signal frequency is capable of? I suppose that would satisfy any arguments that focus on bit throughput.


At those kind of frequencies (and way higher), devices like oscillators, mixers and amplifiers (FETs) are typically fabricated on Gallium Nitride (GaN) or Gallium Arsenide (GaAs) substrates.


Is this just a passive phased array? I can’t electronically steer it, can I?


No, it's active. Assuming they are using their 16 channel chip:

https://perasotech.com/x-series-products/

I'll bet it's two, 32 element arrays (one each TX and RX) composed of 2-element sub arrays.


So where it says support for 16 antennas, that means that there are 16 individual rf amplifiers, and the phase and amplitude of each channel is being set to do the beamforming?

Can a user just put in 16 rf feeds and do the beam steering themself?

If used in the standard form, can it do digital beamforming (send different info on different beams)?


Yes. Each PA probably has no more that 10 mW (assuming CMOS), but coherently added will yield over 100 mW EIRP. For RX the signal coherently adds, but the LNA for each channel has uncorrelated noise, so the RX SNR is not as good as a passive array (assuming zero feed loss). Probably other circuitry to keep all the channels calibrated.

No, not in that chip. It has differential I & Q baseband I/O which are split and converted using a LO coherent to all the channels, then amplitude and phase shifted.

No, that would require phase and amplitude weighting to be done at digital baseband. There are some other ICs with 4 channels to drive 4 element sub-arrays. These large MIMO arrays will have sub-arrays with RF amplitude and phase weighting, then each sub-array is fed with I & Q baseband with separate weighting.


With 22dBi they're quoting, it'd have to be all Tx or all Rx. Unless they're conflating amplifier gain in there, which sometimes happens.


The block diagram didn’t show a switch, and the NF is 6 dB, so I figured it was separate. I have used the HMC-6301, which has about the same NF. The switches get quite lossy at mmWave.


Is 60GHz band unlicensed in US and EU?


Ish in the UK, see http://static.ofcom.org.uk/static/spectrum/fat.html

Some ranges are usable for Industrial/Scientific/Medical (ISM), and there are exempt low power devices although I leave it as an exercise to the reader to find out which:

https://www.ofcom.org.uk/__data/assets/pdf_file/0028/84970/i...

For example:

IR2030/7/2, wideband transmission systems that are not fixed can be used up to 40dBm between 57-71GHz. Fixed systems are allowed, but not within 6km of certain sites. You can also use it for tank probing radar under various situations. One of them is MOD Hebrides which is a large closed area for weapon system testing, another is Castlemartin training area, and another near Luffenham (no idea about that one). Presumably it interferes with military radios and at least two of those sites are used with live ammunition, so probably it's a safety thing.


61-61.5GHz is an ISM band, so it's unlicensed pretty much everwhere. This frequency range is heavily attenuated by oxygen, so it has inherently limited range and low risk of interference even at high emission levels.



Not in the US at least.


60 GHz is unlicensed in the US. Bridgewave and others have been shipping 60 GHz band point to point outdoor radios since 2008.

71-86 GHz ("E band") is light licensed, which is easier and less costly to coordinate than the traditional 6/11/18/23 GHz FCC part 101 microwave bands.


I believe that's what he meant, not [licensed] in the US at least.


That’s Hz, isn’t it?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: