DARPA have created first solid state receiver with gain at 0.85THz | Hacker News Reader

DARPA have created first solid state receiver with gain at 0.85THz

darpa.mil

83 pointsp4bl033 comments

For reference, at 0.85THz light travels only 350um (micrometers) in one clock cycle.

Unbelievable.

Can you explain why that is of any significance? I understand the relationship between frequency and wavelength I'm just not certain how that makes the process more complicated. (I'm sure THAT it does because DARPA are researching it, I'd appreciate some explanation of HOW)

I am not a hardware guy, so I'll shy away from ignorantly answering the question about hardware significance. My limited understanding, however, is that equipment that operates in the higher frequency ranges is exotic and currently quite expensive. That's why you primarily only see satellites utilizing the >6 GHz range (RADAR and microwave backhaul links being two other applications in the higher frequency range [1]). One of the problems in the higher-frequency range is increased attenuation due to oxygen and water absorption [2].

Why would the military interested in higher-frequency communications? I can think of a couple of reasons of the top of my head:

1) There has been significant pressure over the last couple of years from Congress (lobbied by commercial carriers) to reallocate spectrum from the federal government and to auction it to commercial carriers [3]. With increasingly advanced communications systems and waveforms being deployed in the military which require significantly more bandwidth, that means they have to do more with less. Shannon's law tells us that higher frequencies are capable of more bandwidth than the lower frequencies, which means high-frequency RF could transfer a lot more information than low-frequency RF systems [4]. This is important given the military's push towards buzzwords like "sensor fusion".

2) High-frequency communications systems are point-to-point rather than broadcasting over a wide area. Think of how light propagates from a laser pointer versus a light bulb. Low-frequency broadcasts, such as TV stations, propagate over a wide area of land from a single antenna. High-frequency microwave communications, such as the white cylindrical drums you see on cell towers, are directional and require the communication antennas to more or less be pointing directly at each other. There are three advantages of these point-to-point antennas: A) higher bandwidth, as discussed earlier, B) much lower likelihood of detection by your enemy, since your communications are targeted rather than broadcasted, and C) lower probability of interception by your enemy, because they would have to place an antenna directly in the path between your two links in order to capture your signal.

3) RADAR systems can be significantly more accurate in the higher frequencies, allowing more precise targeting and identification of targets. This has major applications both within and outside the military. The Doppler Radars are an example of a non-military application that provides weather information about the US [5]. The more accurate they can make the RADARS, the better they can distinguish between cloud formations and therefore provide more accurate weather predictions.

(Disclosure: I was previously involved in the battles between commercial carriers and US Federal Government regarding reallocation of spectrum. The statements in this comment are my own beliefs and should not be construed as the beliefs of my current or previous employers.)

[1] http://www.ntia.doc.gov/files/ntia/publications/2003-allochr... [2] http://www.mike-willis.com/Tutorial/PF5.htm [3] http://broadband.about.com/od/wireless/a/Ntia-Continues-To-I... [4] https://en.wikipedia.org/wiki/Shannon%E2%80%93Hartley_theore... [5] http://radar.weather.gov/

> Shannon's law tells us that higher frequencies are capable of more bandwidth than the lower frequencies, which means high-frequency RF could transfer a lot more information than low-frequency RF systems [4].

Almost: Shannon's law sets a limit on the amount of information that can be transmitted through a channel with a given signal-to-noise ratio and a given BANDWIDTH (it's bandwidth * log(s/n) ). Now, it happens to be the case that systems operating at higher frequencies often (nay, usually) do have higher bandwidth for a number of reasons, but Shannon's law doesn't care what carrier frequency your channel uses.

You're right. Thank you for the clarification!

Wouldn't an 850GHz signal be lost if an insect flew between sender and receiver? Isn't that going to make it pretty impractical? Honest questions. I could easily be missing something.

High-res radar seems viable, though.

The term "line-of-sight" typically refers to the obstacle presented by the curvature of the Earth or large metallic objects[1]. An insect would not block a THz EM wave, just as the walls of your home (probably) don't block your cell phone's radio.

[1] http://en.wikipedia.org/wiki/Line-of-sight_propagation

It's significant, because historically its been difficult for electronic devices to deal with electromagnetic radiation in the THz range. The announced device is a significant step in fixing this.

As devices have improved over the years, the radio frequencies devices can process have increased. DC to 60GHz is the current practical limit in consumer electronics. Above that is the domain of specialised (and expensive) transceivers.

At the other end, devices can process optical signal relatively easily. Red light is 400THz and the useful range of optical devices might extend down to 10THz or so.

In between there is a gap, in the region about 1THz: The THz gap. These are frequencies whose wavelength is too long for optical devices, but whose frequency is too high for RF devices. Being able to easily access these frequencies offers the promise of all sorts of useful imaging and communications devices. Think of the THz band as being the virgin "wild west" of the EM spectrum.

Lump circuits are defined to be circuits whose size is smaller than the wavelength of their highest frequency. In lay terms that means that every signal can get to any part of the circuit in less than one 'cycle.' That makes the math a whole lot easier and it means you can safely ignore 'transmission line' effects.

When your circuit is larger than the wavelength of your highest frequency, your propagation delays make it more like a distributed system than a single circuit.

Digital circuits like microprocessors work in part by creating 'synchronization points' with global clocks. Synchronous circuits avoid the issues of wave front propagation by pausing until the signal has reached all over the circuit.

It means that the geometry of your device has an increasingly bigger impact on your delay times. It's also going to cause more geometry features of your device to behave like antennae, sapping power (and if you're the NSA, leaking information).

This turns a digital logic problem into a digital logic problem AND a antenna design problem. This is part of the reason why we've been stuck at 3 ghz for so long. We can go faster, it just takes more power; something we've been moving away from.

Information cannot travel faster than the speed of light: it's the theoretical upper-bound. If in one clock cycle information can only travel 350 micrometers, that puts serious limitations on the engineers designing this device.

The fastest information can ever travel is the speed of light (possibly ignoring quantum entanglement effects---I'm not up to scratch on that) and this is just a theoretical maximum, so it takes multiple clock cycles for information to travel around on a piece of hardware, and thus optimising its layout is immensely important for performance.

>possibly ignoring quantum entanglement effects---I'm not up to scratch on that

Including quantum entanglement effects. If you know otherwise, publish a paper ;).

http://en.wikipedia.org/wiki/No-communication_theorem

If you think of it as a CPU running at that speed and what is needed to achive that then that might help get an idea about this in terms of achievement. As for radio the higher the frequency (think sin wave) the shorter the wavelength (distance between sin waves). This also means smaller recievering antentna and etc. One other use this does have is one day your computers chips and memory wont need loads and loads of intricate expensive wires to connect them together as they will talk wirelessly.

That all said I'm sure others will explain this news in far better terms and alot more acurately.

It's also a big increase in the bandwidth-delay product, the amount of data you have an opportunity to send down the pipeline before any responses could arrive.

Very true, for asemetricaly processing unlike simplex processing. Though for inter device communications for currently technology it could be used to save alot of wire space - especialy between memory and CPU's. When it becomes cost effective to do for things like that today, well that can be a opertunity as the delay is more than accomodated by the slower parts and buffers, just to save the use of parralel wires is worth it and can be faster as less hassels to syncronise channels. Look at disc interfaces over time with SATA replaceing PATA for speed. Just easier to design and process for.

Yes, the wavelength is on the same order as the circuit elements. You can tell from the layout of the IC that this is a traveling wave amplifier:

http://en.wikipedia.org/wiki/Distributed_amplifier

Does this mean there have been non-solid-state receivers at this frequency? Or is this a use of 'solid state' I'm not familiar with?

I don't know much about this area. I know that there are vacuum tubes that operate in this frequency range.

http://en.wikipedia.org/wiki/Backward_wave_oscillator http://apl.aip.org/applab/v93/i14/p141108_s1?isAuthorized=no http://www.sron.nl/index.php?option=com_content&view=art...

It looks like the existing receivers at this frequency range are of the heterodyne type, possibly without any RF gain. A heterodyne receiver mixes the RF signal with a local RF oscillator, which results in a lower frequency IF signal which is easier to work with.

At these frequencies, it looks like the local oscillator is a "backward wave oscillator" or a "quantum cascade laser", and the mixer is some kind of esoteric non-linear device like a "hot electron bolometric mixer", whatever that means.

The highest frequency amplifier I know of is the "traveling-wave tube" amplifier, which goes up to something like 50 GHz. It's a crazy vacuum tube device and it's found on satellite transmitters. Making these things solid-state would be awesome because the tubes take up so much space and power on the satellites.

Summary: There are vacuum tubes that operate in this range, and there are receivers without gain in this range.

"backward wave oscillator"

"quantum cascade laser"

"hot electron bolometric mixer"

"traveling-wave tube"

Computer scientists, take note: start naming your ideas like this and you will get more grant money.

Friend once said the difference between solid-state and non, was that solid-state didn't use magic crystals.

But personaly I don't like the term as it is not as clear-cut in all peoples perspectives. You could say one solid-state is digital and non solid state is not 100% digital. Though some would also argue that digital is mearly a perception of measurement in a analogue World.

    I'll let you read the wiki on it and draw your own conclusions:

 http://en.wikipedia.org/wiki/Solid-state_(electronics)

Crystals are actually a major portion of the study of solid-state physics:

http://en.wikipedia.org/wiki/Solid-state_physics#Crystal_str...

Thats physics and not electronics but that is the whole problem with the phrase solid-state and in that it is not a good way to refer to things without being qualified in a way that negates its use or even mentioning it in the first place.

Could you create a communications system with these solid state receivers ? I'm thinking about a communication network with these things? Or is that impossible.

Yes, if you have a circuit that can operate at .85Thz you can modulate it between say .84 and .845 Thz. It is fairly straight forward to build a circuit to detect the frequency of a signal by sampling 10 cycles of that signal. Further the difference between .84Thz and .845Thz is 5Ghz. That is a pretty big difference so easy to detect. 10 cycles per 'baud' would yield a baudrate of 84Gbs or with 8b/10b encoding 8.4GB/s. That's 120 DVDs per minute :-) The Shannon limit is 10x that.

It would have to be line of sight of course, its not going to reflect off much.

Well, right now, the article only mentions a receiver. So you're half-way there, you just need to build a transmitter.

An oscillator with a loose wire [1] is a transmitter :-)

[1] http://sci-toys.com/scitoys/scitoys/radio/am_transmitter.htm...

And do please show how to build an 0.85 THz oscillator...

Connect a capacitor to be base of a .85Thz transistor. My favorite design for the simplest form of relaxation oscillator [1].

[1] http://hyperphysics.phy-astr.gsu.edu/hbase/electronic/relaxo... can't draw the one from the Art of Electronics here but this site captures it pretty well. The 'trick' is having a .85Thz transistor.

AFAIK, an oscillator is necessary for building a receiver. In other words, this is apparently a solved problem.

Well, you do get vacuum tubes that operate up to a terahertz - they don't count as solid-state devices, though.

This changes everything.

I hope the TSA doesn't hear about this.

Reply on news.ycombinator.com