Passive Optical Diode Created At Purdue University 92
wbr1 writes "Researchers at Purdue University have managed to create a silicon device that acts as a passive diode for infrared optical signals. From the Purdue news release: 'The diode is capable of "nonreciprocal transmission," meaning it transmits signals in only one direction, making it capable of information processing, said Minghao Qi (pronounced Chee), an associate professor of electrical and computer engineering at Purdue University. "This one-way transmission is the most fundamental part of a logic circuit, so our diodes open the door to optical information processing," said Qi.' One of the same researchers had already (using similar technology) created a way to convert laser pulses to RF."
Re: (Score:2)
Is there a better article on this somewhere? (Score:4, Insightful)
Both the summary and TFA are devoid of anything concrete on how this is actually done. It basically says what the title does, they created a diode. Telling me that light entering the opposite side doesn't make it through really doesn't tell me anything the word "diode" in the title doesn't. I'm sure the science behind this particular device is both clever and interesting but you'd never be able to tell since that information is completely missing. Reporting on stories is nice, but shouldn't journalists actually strive to make their articles contain actual information on what they are covering? You'd think a story about a new discovery would actually contain information about how it actually works (since that's the actual "new" part anyway).
Re: (Score:1)
Both the summary and TFA are devoid of anything concrete on how this is actually done. It basically says what the title does, they created a diode. Telling me that light entering the opposite side doesn't make it through really doesn't tell me anything the word "diode" in the title doesn't. I'm sure the science behind this particular device is both clever and interesting but you'd never be able to tell since that information is completely missing. Reporting on stories is nice, but shouldn't journalists actually strive to make their articles contain actual information on what they are covering? You'd think a story about a new discovery would actually contain information about how it actually works (since that's the actual "new" part anyway).
Completely missed the second link to the university's release. Still not a very informative article however.
Re: (Score:2)
what hz are the radio transmissions done at? because the article.. well.. the article justifies that it can transfer more data simply because it's optical at one point before transmission.
Re: (Score:1)
Having read the summary and TFA and understood what they were saying backs up your comments re paucity of information. IE if I can understand it there must be more! Actually.
Re:Is there a better article on this somewhere? (Score:5, Informative)
Re: (Score:1)
Subscription required....
Re: (Score:2)
Re:Is there a better article on this somewhere? (Score:4, Informative)
Holy Entropy (Score:3, Interesting)
---
Third Magic [blogspot.com] @ Blogspot
Re:Holy Entropy (Score:5, Funny)
Re:Holy Entropy (Score:4, Interesting)
It lets heat go in one direction only! That the biggest change in Thermodynamical Law since Claude Shannon.
Now convert it a transistor, and with a maxwell daemon, (Quantum Weak Measurement, + Quantum Computer, + Classical Prediction Logic)
and we have (possibly) a free entropy device, capable of turning waste (heat) energy back to useful energy.
From TFA:
Depending on which ring the light enters first, it will either pass in the forward direction or be dissipated backward, which creates one-way transmission.
It seems to more just act as a nonlinear lens, dissipating (or more likely scattering) preferentially in the backwards direction.
Re: (Score:2)
It lets heat go in one direction only! That the biggest change in Thermodynamical Law since Claude Shannon.
Are you thinking that infrared mirrors are not possible? Does the article claim that his device does work without increasing entropy?
Re:Holy Entropy (Score:5, Informative)
If you're going to go around with a name like physburn, please ensure you understand what you're talking about
Re:Holy Entropy (Score:4, Insightful)
I think it comes from a basic lack of understanding of how heat is given off by fires. If all you know is something about photons and light... and that light absorption causes heat, you fill in the blanks and reason out that fire must release most of it's energy in the non-visible light spectrum. Also infrared cameras show hot and cold, therefore many may reason that infrared = heat.
In my memory of high school physics we didn't go in depth into heat transfer nor radiation. If that was the standard curriculum of the time then many people of gen X could believe that heat transfer is due to infrared light. It's interesting that the wikipedia page on Heat [wikipedia.org] shows that many science textbooks use the term in confusing ways. Also
So one could be a competent scientist and still use the term in a semantically incorrect way, unknowlingly passing on disinformation.
It would be interesting to do a little informal polling of what heat is and how it transfers. What percentage of people know how it really works? What percentage of scientists?
Re: (Score:2)
True, it is a fairly subtle concept. I am much more concerned about the lay confusion between temperature and heat than between light and heat. However, it should be fairly well reinforced that light does not interact with something unless it interacts: windows heat up less than asphalt in the presence of energetic radiation in the optical regime.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
This device is pretty big to make a transistor.
So was this [wikipedia.org] but look how that turned out.
Does not "open the door"... (Score:4, Interesting)
Why do so many researchers lie so shamelessly to the press? This may be a step in that direction, but it is a rather small one. Key components are missing and a lot depends on the actual characteristics of this device.
Re: (Score:3)
And even if major components like a light transistor are developed I don't see optical computing taking off in the foreseeable future. This "diode" is kind of weak since it stops reverse light by dissipating the energy. So any non trivial computing function made with it will consume a non trivial amount of energy in the form of input light. Compare this to electronics where a switch in the on or off state doesn't use energy apart from leakage. The major thing consuming energy there is switching states, and
Re: (Score:3)
Every time I read about an R&D achievement at an American university, the lead (and often associate) researchers are Chinese nationals.
Do Americans no longer conduct advanced R&D at American universities?
Seems foolishly short-sighted, if so.
Nothing like a Qi spell to give technology a boost...
And maybe he is an American, or is working on his citizenship. This country used to take pride in being a melting pot.
Namely (Score:1)
Finally, we got the Pixel Qi
Great news for the slashdot smart people (Score:2)
Bad news for us who have no idea what this means or implies. What could this do if this thing were real? I haven't seen a laymen explanation.
We're not all smart guys you know.
Re:Great news for the slashdot smart people (Score:4, Informative)
http://simple.wikipedia.org/wiki/Logic_gate [wikipedia.org]
All those degrees in looking things up online finally paid off. You're welcome, citizen!
Re:Great news for the slashdot smart people (Score:4, Insightful)
Re: (Score:2)
I'll give you the answer that I think you're looking for, but I will rant a bit at the bottom about what I believe is a flaw in your attitude. If the rant is something you will ignore because you're not interested in constructive criticism, feel free to not bother reading anything after the explanation.
The goal is to build circuits that are not electronic, but optical in nature. This could result in faster computers (if we can make fast enough optical switches), but it will almost certainly result in more
Re:Great news for the slashdot smart people (Score:5, Informative)
A transistor acts as an electrical switch. Basically, it means that when an input is provided, the rest of the circuit can do something, but at the same time the input is electrically isolated from the rest of the circuit.
Before transistors, there were relays and tubes which accomplished the same thing. They were slower, larger, hotter, and used a lot more electricity. And they were prone to burning out. As a result, computers were hopelessly complicated, the size of small rooms, and were programmed with a screwdriver. And "bugs" in the computer program were sometimes, literally, bugs. A moth, in at least one story. And an IBM chairman famously stated that he saw a world-wide market for about 5 computers.
Then the digital transistor came along and revolutionized that. They were smaller, faster, and required much less power. And they were cheaper, too. The integrated circuit - millions of transistors etched onto a single silicon die - revolutionized that further.
A transistor also acts like a diode, in that it only lets current flow in one direction. But note that this isn't really even necessary: relays, for instance, didn't prevent current from flowing backward. The main thing is that the input is electrically isolated from the output, not that it allows current to flow in only one direction.
So basically, they have everything required to build an optical transistor, except for the switching part. Which is kind of the more important part anyway.
Re: (Score:2)
That doesn't follow. Computers were hopelessly simple back then. Or are you claiming the latest quad core CPU with a billion transistors is simple compared to a relay adder?
Re: (Score:2)
To be fair, you can build logic gates using diodes. The only drawback is that the devices operate on the principal of voltage drops across diodes and level shifting, bad for complex circuits.
Re: (Score:2)
It does not operate on the principle of voltage drops, it's like saying that cars operate on the principle of heating up their environment (65%+ of your gasoline heats up the air around you). Diode logic operates, obviously enough, utilizing the large ratio of forward-to-backward current flow across an isolated semiconductor junction. Voltage polarity across a diode turns it on/off, but what is turned on is the current flow. I don't know where the idea about level shifting came from, it's not necessary at a
Re: (Score:2)
Re: (Score:3)
We're on the threshold of a lot of things, for a long time, before they become reality.
Newton wasn't a lone genius (parallel development notwithstanding) he was just the right man at the right time, standing on the shoulders of the right giants.
I suspect, once ITER is complete, or maybe it's successor, and practical fusion power becomes reality, then we'll just as equally forget that "we were on the threshold" for over 50 years.
Re: (Score:2)
The paper is out online only.
Bull. They're halfway, the easy half at that. (Score:3)
The "most fundamental part" of logic isn't one-way transmission, it's the ability to control that transmission by applying a voltage to the transistor's gate. The fact that current will only flow one direction between the emitter and the collector is really not that important by comparison.
You can't build logic from diodes.
Re: (Score:2)
Actually, I'm not an expert on photonics, the analogue of voltage could be wavelength, in which case upconversion can provide the same function. There's a lot of fluff on the Purdue site but this is actually a fairly interesting device and the research isn't bad.
Re:Bull. They're halfway, the easy half at that. (Score:4, Interesting)
Actually it's really really hard. The optical transistor is the current holy grail of photonics and optical computing. The person who invents it will be incontinently rich.
Re: (Score:2)
Re: (Score:2)
An optically controlled switch is easy. Any medium where you get optical saturation lets you implement a light-controlled switch. You can also have media that, say, change opacity when exposed to light of one wavelength, so you can control transmission of say red light using UV. You can also have a light-controlled-gain device, like, say, a light-pumped laser -- turn off the pumping light and gain drops below 1. Turn on pumping light and you have positive gain. The power levels involved are not very practic
Re: (Score:2)
Re: (Score:1)
Not voltage, signal. Maybe there's a way for the signal to be light (creating an all-light computer) or maybe it's a voltage - or maybe it is something else. But it has to be something we can control.
Re:Bull. They're halfway, the easy half at that. (Score:4, Informative)
My point was that the signal is the important part. They have no way of switching this optical diode.
The diode effect is fairly irrelevant and unimportant. As a matter of fact, digital logic doesn't require something to act like diode at all: relay logic doesn't use diodes.
Re:Bull. They're halfway, the easy half at that. (Score:4, Insightful)
Your point being, I take it, that you can create certain gates with diodes (AND and OR).
There are some truth tables which can be achieved by nothing but AND and OR. There are some that cannot. All truth tables can be achieved by solely the use of either NAND or NOR, but you can't create those gates using just diodes.
Digital logic requires the ability to do something on the 0 state. Without an inverting gate, you can't do that, and diodes can't invert.
Re: (Score:2)
Those diodes perform an auxiliary function and are not doing any sort of logic by themselves (as they would were it diode, not relay, logic). If you're careful about it, you can do without diodes, using a snubber instead. Heck, there are applications where you don't want a diode across a coil at all -- that's when you want to control broadband emissions. A snubber-on-coil circuit is tuned and has high damping, you can tune it to have relatively low frequency oscillations. With a diode you have fast switchin
Re: (Score:2)
You can build some logic using diodes.
Not all logical functions can be implemented in diode logic alone; only the non-inverting logical AND and logical OR functions can be realized by diode gates.
You can build some logic using a plain old switch, too: IF the switch is closed, THEN light the bulb. Not terribly useful. Neither is diode logic.
Re: (Score:2)
Here's an abstract for ya. And this is nothing new: it's from work done in late 50s [ieee.org]:
The realization of switching functions using current-operated diode logic gates, which employ the direction of current as the binary variable, is presented. Two realization methods are discussed. The dual method consists of a set of rules which are analogous to the procedures employed for the realization of voltage-operated circuits. The lattice method is based on the transmission characteristics of a lattice network of four series-connected diodes. Either method is capable of realizing any arbitrary transmission function, although the number of diodes required may be different. Hybrid networks, which combine the advantages of both methods, are described to illustrate the versatility of current-operated techniques. Practical design procedures are included to guide the designer in the application of the realization rules. A static multiplier, capable of simultaneously multiplying two 4-bit binary numbers, was designed and built to demonstrate the practicability of current-operated diode logic gates. Approximately 270 diodes were required to construct the twenty gates used in the static multiplier. Dynamic tests of the assembled system indicated a reliable capability of 333,000 multiplications per second.
If you can make a 300kHz 4x4 multiplier using diodes, then I'd say this is hardly some logic -- in my book, at least. The wikipedia article on diode logic, and most other intro-level articles I could find, really miss on how diode logic was (and is) used, and provide only some lame, almost useless examples.
Re:Bull. They're halfway, the easy half at that. (Score:4, Informative)
You can't build logic for any given truth table using nothing but diodes. You can only build logic for some truth tables, which doesn't give you programmable digital logic. Happy now?
Re: (Score:2)
I think you're wrong. You can build logic for any given truth table, using only "plain old" diodes, nothing exotic. This was state of the art in late 50s/early 60s. I'm pretty sure you could do programmable logic as long as you'd use an active device like an SCR to retain the state of the programmable "fuse". LUTs and multiplexers, the staples of modern programmable logic, are fairly easy to do with nothing but diodes. And they aren't terribly slow either, you could probably build a current-driven 10 LUT "P
Re: (Score:2)
Re: (Score:2)
Mr Henry Reinecke [ieee.org] seems to disagree. I've downloaded the paper and they do use current drivers and current mirrors to drive inputs and couple stages (including feedback from outputs to inputs), but the logic itself is all diodes. I've done a quick simulation of the most complex network he used in the multiplier, and with modern diodes and well laid out board you could probably run it at 10MHz. In the paper he shows that diode logic has dual representation -- one uses voltages as signals, another uses curren
Re: (Score:2)
Re: (Score:2)
It's not the voltage drops that kill it, it's voltage levels and the fact that a current source needs to have infinite impedance. It's hard to approximate that without active devices.
Re: (Score:2)
I meant obviously without any active devices, if one considers a linear transformer to be a passive device (as opposed to a magnetic amplifier that would be considered active).
Re: (Score:2)
You can't build logic from diodes.
You most definitely can, even complex logic like multipliers [ieee.org]:
A static multiplier, capable of simultaneously multiplying two 4-bit binary numbers, was designed and built to demonstrate the practicability of current-operated diode logic gates. Approximately 270 diodes were required to construct the twenty gates used in the static multiplier. Dynamic tests of the assembled system indicated a reliable capability of 333,000 multiplications per second.
Just google for some ideas. You can even build voltage-controlled multiplexers using nothing but diodes -- quite good ones, even. Your basic high-frequency sampler (going up to tens of GHz) is nothing but a sampling capacitor in a sampling diode bridge. Such bridges maintain symmetry that preserves integrity of the sampled differential signal and are the simplest way to quite accurately sample a quickly changing signal. They have incredible bandw
Re: (Score:2)
That's a pretty interesting concept, but it uses an entirely different binary representation (current polarity switch for 0/1, rather than voltage potential). In essence, it means that every signal is 2 leads rather than 1, and switching them creates a NOT gate.
You might be right, but I'd want to see a working flip-flop before I'd accept that diodes could be used for digital logic.
Re: (Score:2)
The paper shows that the current representation is merely a faster dual of the voltage representation. The circuits be mechanically transformed from one form to another. Current-driven diodes do work as flip-flops, but this logic of course needs external active components to inject input currents and to propagate currents between combinatorial stages. That's how you get flip-flops or any other function that requires feedback: you put a current mirror between output and the input.
If you want purely diode-bas
Re: (Score:2)
Oh, and I should mention... optical signals only have 1 lead (fiber), not 2. My brain is too tired to try to figure out what that would mean as far as trying to use optical diodes to accomplish anything similar to the current-direction-based logic used for that multiplier.
Re: (Score:2)
With two fibers you can encode information in phase difference, and I bet that would "polarize" a two-input optical diode OK.
Optical CPU's ? (Score:1)
So.... (Score:1)
Correct me if I'm wrong but... (Score:2)
I thought any diode only allowed transmission one way.
Re: (Score:1)
Re: (Score:2)