I like the part were the Professor says he doesn't know how much long is the shortest pulse, but says "I can look it up if you want". Such great minds have no problem with not knowing something, while this planet is so full of know-it-all s.
And that's on a single wire. I imagine you could have 1000s of wires on a single cable, yes? Maybe you run into more physical limitations or errors at too small of a size too, though
Run lots of cables in parallel yes but also with slightly different properties allowing them to be a control for overall transmission (plus 'overhead') for further range of frequencies (handling things as a bundle). Ie the different cables extend in part the range of frequencies available resulting in more relative bandwidth per cable.
There is an energy density problem, and mutual coupling between wires. Shannon's theorem says that data rate is given by bandwidth times signal to noise ratio. (For our purposes, dynamic range is a limit on signal to noise ratio.) A one Hertz bandwidth on a single cable can transmit a terabyte per second with a mere 44 teradB of dynamic range, or equivalently 8 x 10^12 bits of resolution.
all we need is to start encoding data in sperms and transmit them through pipes. they carry terabytes of data per teaspoon, so if we want to copy entire massive databases just have a sperm truck go to the new location, then do a memory check to see if everything transmitted and use the internet to make up the tiny difference.
Ali Moeeny I am not sure. He introduced some math, but he steered clear of even naming FT. I think that if he directly went for the concept, explained it, and made it part of the vocabulary, the whole explanation would have sound less sketchy.
Michael Merrifield , I just wanted to thank you for making me passionate about topics I never knew I had passion for until I saw your videos here and on Deep Sky Videos. You have pushed me to learn and understand some of the involved math, encouraged me to purchase a telescope for skygazing purposes, and opened up new areas for me to explore. I am more grateful to you, Ed Copeland, Meghan Gray, Phil Moriarty, and Roger Bowley than I could ever express.
yeah, especially in guitars you can hear beats and harmony off diff. instruments creating non-periodic reverbs, musicians are pretty close to science compared to other non-scientific fields..
Why wasn't the term heterodyning used in this video? Bandwidth requirements become much more understandable, when the concept is explained. Two frequencies will combine to create both constructive and destructive frequencies. Heterodyning was demonstrated with the iPhone-based tone generators, but not clearly explained. If a 1000Hz tone is pulsed at 1Hz, the transmitter will occupy 2Hz from 999-1001Hz. Today's data networks use something called Dense Wave-division Multiplexing. We combine multiple frequencies often near 190THz with spacings of a mere 0.8nm, each channel representing 10Gbps. Prisms are readily available to combine up to 80 channels on a single pair of fibers, which translates to 800Gbps, or a Terabyte of raw data in 10 seconds.
@Michael Hall: What you're talking about is the actual binary information that is transmitted. The video is concerned about the physical limits in transmitting information accross a fiber-optic cable. And mind you, as the professor clearly states, a mathematically true pulse requires an infinite amount of frequencies or otherwise an infinite amount of preparation and decay times to form. Fortunately quantummechanics comes to the rescue there giving us a bounded physical world i.s.o. a purely mathematical one. I really love that part.
I have to admit, I have never heard of the range on frequencies that are able to travel down the optical fiber to be the limiting factor. I was taught that it was the Group Velocity Dispersion that limited the data rate, as due to this effect the pulses spread out temporally (That is, the pulses duration continues to increases as the pulse travels down the fibre). This increase of pulse duration makes it so that if pulses are fired too quickly after each other, sometime down the fiber the pulses would spread into each other and you will lose your nice wave packets and hence your data.
That is true. However he's talking about different types of limit. It is also strange that he talked about visible spectrum when fibre optics use only IR.
Well I too mostly understand what is explained, but not quite this time. He lost me at the point that he explained that if you switch a laser on and off, it is a superposition of waves and if you keep it on continuously, it isn't. I understand the Heisenberg uncertaintly principle, and I understand superposition, but I cannot understand what that has to do with it. I also understand the (unmentioned) Fourier analysis, but I cannot make one single image in my mind on what the difference is between switching a laser on/off and leaving it on continuously.
Ronald de Rooij The thing that he said about continuous wave (CW) is true. It is almost never really exists in real world. I do a lot of EM simulation, and it is a real problem to simulate a CW. We use different kinds of tricks to suppress high frequency signals that permeates from starting the so called 'CW'. These high frequency signals are real problems in our frequency domain analysis.
While I understand the argument being made in the video, it feels like he's contradicted himself. I was under the impression that individual photons represented a packet of energy of a single frequency... What he's saying suggests that a photon itself is a superposition of multiple frequencies?
earthworm768 - actually group velocity dispersion (group delay in the radio world) means that different wavelengths travel through the fibre at different speeds and the frequency components arrive at different times which blurs the pulse edges. it can be partially corrected. it is still a function of bandwidth
You sir explained a whole lot of wireless communication to me in minutes! Thanks prof. Your students are so lucky to be taught by you! Please consider doing more videos on wireless communication. Yours will make a great channel.
The number of "beatings" per second its equal to the difference between the two frequencies in hertz. So, 1000Hz and 1001Hz: 1Hz difference, or one beating per second. 1000Hz and 1002Hz: 2Hz difference, or two beatings per second.
When he played two tones of different frequencies, superposition of two signals took place. It will result into wave/signal of different frequency. Therefore, that resulting frequency will be LCM of the two original frequencies.
A sine wave can convey three pieces of information. Amplitude, Phase, and Frequency. The 19kHz FM pilot tone used to decode stereo information (United States) is an example. You can think of turning a sine-wave on/off as multiplying the sine-wave by a windowing function that is 1 when the sine-wave is on, and 0 when the sine-wave is off. The windowing function has a frequency spectrum of its own, and this gets convolved with the sine-wave's spectrum in the frequency domain. A pure sine-wave's spectrum looks like two impulses (sharp spikes) at +f and -f. A rectangular (perfectly "square" pulse) window function's spectrum is a sinc (sin(x)/x) function. When the sinc spectrum gets convolved with the sine-wave's impulse spectrum, it copies, shifts, and divides by 2 the sinc spectrum around +f and -f. Negative frequencies arise from converting Real valued time-domain signals into their complex-frequency domain representation. Check out the Shannon-Hartley Theorem that relates maximum channel capacity to Channel Bandwidth and Signal-to-Noise ratio.
That's not the *type* of information he is referring to. That is nothing more than the nature of the transmission. You need an oscilloscope to translate that. Hes talking about using that like morse code.
Sir, this is the most beautiful and concrete explanation of Quantum physic, Fourier Transformation and Information theory lectures in the whole universe!!! I'm a engineer and finally after all those years someone showed me how to explain connections in all of those laws using one simple example!. This really worked for me!!! Thank you!!
+Juan Bonnett the moment when someone realise or deeply interiorise that the basic sciences are actually the thing that the modern technology comes from.
Bandwidth isn't computerphile? With current technology there is a known maximum bandwidth that we could Mor's Law predict ourselves into a problem with, in a relatively short time frame. At that point laying more wire to fill an exponentially increasing demand isn't plausible, you need better technology with a higher bandwidth or better compression to send less data. This all seems pretty computer related, to me.
All he talks about is in fact rudimental Computer stuff. *You learn what bandwidth is and how it is “produced“ when you are learning any computer related job!*
As a non-physicist I was doing great for the first 4.5 mins and then it all went! However I still love watching these videos of Brady's in the hope that I will start understanding eventually. Still envious of how Brady seems to understand so much as shown by his responses. I am beginning to think he no longer represents the "ordinary" person and so would like to volunteer my services for future videos!
It's brilliantly explained. He starts with the beats that happen when two waves of different frequency superposition against each other (with a handy graph showing it beating in real time). He then draws an "infinite" sine wave and a pulse. You can clearly see that the pulse has tails, these tails that go to zero which is clear evidence that something other than the sine wave (like another frequency) is affecting the wave. A pure sine wave would just stop with no tails.
I think the lower bound is definitely the size of intergalactic dust (so perhaps IR?) I don't think there is an upper bound tbh, since space can have almost no mattet (a few atoms per m^3).
So that would be around 80 Tbps or 80.000 Gbps. Should be enough for future applications. Looks like some people even managed to almost reach the maximum. www.techspot.com/news/52066-hollow-fiber-optic-cable-tops-73tbps-promises-near-light-speeds.html We really need fiber optical cables everywhere and into every building. You'd only have to upgrade the terminals and recievers to get higher bandwidths in the future.
The reason why you don’t switch on and off and use a sum of sine waves to make 1s & 0s instead (Brady’s unanswered question) is that by modulating multiple sine waves you can actually encode a crapload more information per second into the optical fiber. The resulting non-sine signal contains many more contemporaneous 1s encoded. Professor Merrifield explained instead that you can simulate the “ON/OFF” physical phenomenon with a finite sum of many sine waves. The more numerous they are, the more squared is the resulting wave, and the easier it is to tell a 1 from a zero, hence a wider bandwidth.
2:04 a resonance. A pattern of the waves complimenting each other and getting stronger, or canceling out and making the noise more fuzzy. Like pushing your legs on a swing set
I learned something new! What a fantastic explanation. I had no idea that optical communications relied on the wave component of light to transmit information. Thank you so much Professor Merrifield. I absolutely love watching your videos. Don't be discouraged by negative comments. You can't please everyone and the people you do reach are very grateful for sharing your knowledge with us.
This reminds me of how they said CAT7 cable was going to be needed to handle the bandwidth of gigabit Ethernet and within a few years the engineers figured out how to do it over CAT5 cable with the right processing power and better noise reduction algorithms. Then wireless added new modulation schemes and multiplexing and jumped from 802.11b to 802.11n to 802.11ac. The lesson: don't tell an engineer there is a limit as they will find ways to exceed your limits.
+Stephen Furr Advancements in technology only push the speed near and near to the fundamental limits of the media. If the media is for example a CAT5 cable with a given noise floor you can only push so much information down the line (see Shannon theorem).
That's what the video is about, the shortest period you can send the light for has a limit; the shorter the period, the more frequencies of light that are required and fibre optic cable only has a limited amount of light frequencies it can transmit - which we generally call 'visible light'.
Talking about photons got me thinking. What is the smallest wavelength difference that two different photons could have? I understand that each photon has a particular energy, determined by its wavelength, so that a photon can have wavelength x, and the next highest energy photon will have wavelength x-n. But what is the value for n? Is it the Plank length?
+E “Anonymous Nerdfighter” Hernandez The smallest wavelength difference, I would assume, would be one Planck Length. The reason is because the Planck Length is the smallest discrete distance that can be traveled and measured in the universe, so it would be impossible for two wavelengths of light to differ by any less than that.
I'm sure someone has already pointed this out, but the bandwidth only basically tells you how many pulses (or symbols) per second you can send. But you can actually send much more data than the number of pulses, basically, by sending pulses of different shapes. The fundamental limit (Shannon limit) is about the ability to distinguish the different pulse shapes from each other. If quantum noise and uncertainty is your only noise in the fiber, then the fundamental limit for sending data is actually much much much more than the mere ~10^15 bits/s. The trick is that the pulses are not made of single photons, but of a number of photons. Pushing more energy increases the signal-to-noise ratio, i.e., lessens the uncertainty regarding the aggregate pulse shape. For reference, DVB-C2 (digital video broadcasting for TV cables) specifies QAM-4096 as the densest coding to pulses. This basically means that there are 2^12 = 4096 different pulse shapes, i.e., 12 bits per pulse can be transmitted. So, the fundamental limits in the video apply when you're sending only one photon per pulse. Ultimately, I guess the limit is how much energy the fiber can transmit before melting.
Dr. Merrifield please don't pay any mind to comments like this. You're my favorite professor on any of Brady's channels. There are many more people who feel the exact opposite of those who make comments like that one!
As a big fan of your videos and someone who normally has no problem understanding them, I have to admit none of this made any sense to me. I would greatly appreciate it if you could make another video explaining this better. I know you have lots of videos to make but this seemed interesting but was really not explained well in my opinion, so if you get a chance I would love an opportunity to actually understand what he is talking about here.
Yeah normally i understand everything pretty well too but this time i really struggled to understand anything of substance the first time watching. By now i only understand the first part of the video, where the delta f * delta t = 1 formula comes from.
(1) The speed of your fiber connection is how much information can travel through (2) Information is 1s and 0s (3) "1" and "0" are basically "pulse" and "no pulse") (4) The shorter in duration the pulses the more pulses can travel through your fiber within a given period of time (5) A short (duration) pulse can only be created by combining loooots of waves of different frequencies (6) There is a limitation on what range of frequencies (bandwidth) your fiber can support (7) Since the range of frequencies is limited, the duration of the supported pulses is also limited, because of (5)
Search for frequency genrator apps on play store or app store. To replicate that pulsing effect, choose a stable frequency let's say 1khz and then play 1.01 khz in other phone.
I calculated his answer a bit more so its easier to understand. In terms of data transfer the ultimate limit you can send is about 50,000,000 Mbyte/sec or 400,000,000 Mbit/sec in a fiber lika that (from blue light to red light)
I believe people like him feel the need to post such comments, not because they want to help you or themselves, but because of their own inability to understand, and on account of their character, instead of trying to be constructive, they're unhelpfully critical of the explainer. Such people are not worth paying attention to. I think you explained this, and other topics very well. I derive much learning from watching your videos.
Taking the dispersion effects into account the datarate would significantly drop as the phase velocity of all those frequencies across the visible wavelength wont be the same. On top of that different excitation modes withing the fibre would be excited causing even more dispersion. Along with all this we got non linear raman effects. The real limit would be wayyyyy lower in any practical application.
Of course that is my understanding, based on high school science some thirty plus years ago. Now everyone who really understands his question can attempt to reply correctly if I am not doing so, in terms such as I have used. Thanks.
Thank you Prof Merrifield and Sixty Symbols. This is a very complicated topic I couldn't get my head around as I find it hard to understand the structure of a wave.
You actually could encode information in an unchanging sine wave; you could encode it in the exact frequency. Let's say I wanted to encode the word "it". If you break "it" down into ASCII, that's 0x69, 0x74 (in hex), or 0110100101110100 (squished together in binary), depending on how you decide to do it. Anyway that translates to the number 26996 in decimal. So, if you wanted, you could send out a tone at 26996 Hz. If another person listened to the signal and figured out what frequency it was (easy with computers!), they could decode that number into "it"! But yeah, this is obviously a silly idea. Just having some fun!
Well, there is a problem with the "easy with computers" part. So far with the word "it" we are outside the frequency range of digital audio, so you will need special hardware to digitize the incoming 26996 Hz signal. And "it" is a very short word. I'll let you calculate the frequency of the word "information" with your system, but I guess we would in the 10^27 Hz range... By the way, an unchanging sine wave has to be infinite in time, to be unchanging. Otherwise it'll contain other frequencies. As you said, having some fun.
Different vortices of different colors of light, each carrying its own unique data stream, can all be packed together into a donut-shaped beam of light that travels along the cable. Optical devices can then filter out individual vortices at the other end of the signa
This video is 100% correct when talking about the bandwidth of the fiber by itself. In practice, we need to drive the laser with electronics which are much more limited in bandwidth than the fiber. So we use multiple lasers with different wavelengths. Lookup WDM and DWDM..
Those crazy levels of speed are more commonly found in trans-oceanic cables, where billions of people's bits are going over a single bunch of fibers (like a shared cable internet connection on steroids).
G4mm4G0bl1n I never said it was a single fiber. I said "cable" and "bunch of fibers". Those cables are massive bunches of fibers. My point was that the only real need for anywhere near theoretical single fiber bandwidth is in those transoceanic fibers. No home or business network, metro net, ISP, etc. needs that at this point.
G4mm4G0bl1n Not at all. I specifically said "at this point" not "never". As long as LANs are capped at 1 maybe 10 Gbps, WANs need not be faster. Will the need be there soon enough, of course. But not today.
Visualize the pulse as a function, with time as the X axis. Now visualize the spectrum of the pulse, with frequency as the X axis - the spectrum will have a large bump of a certain frequency width. Picture keeping the graph of the pulse the same while you scale the time axis (stretch or shrink the pulse). Similarly, your mental image of the spectrum of the pulse is not changing, but the frequency axis is being scaled. Shrink the time axis, and the frequency axis stretches, and vice versa.
Once you phase shift modulate a sine wave, it is no longer a single frequency. As we saw in the video, any modulation of a sine wave creates a signal which has bandwidth.
Modulation of the wave frequency and its amplitude, can increase the bandwidth of the fiber.In addition, it is possible to transafe the light over a diffrent angles within the fiber which.
Mammalian nerves utilize an ionic electrochemical system to transmit info, and although it can be described as electrical, it is much different than a coax cable or an optical fiber, having a maximum rate of about a kilobit per second. However, the nerve that takes info from the eye to the brain is not a single fiber, but a bundle of about 4 million that convey info in parallel. So about 4 gigabits per second within an order of magnitude, I'd guesstimate.
So that actually means that in the far future we will be trading TERABYTES instead of MEGABYTES in our computer connections? This is just awesome! Thank you for explaining the science behind this...
Look up fourier transform - when you turn a sine wave on or off, it is no longer one sine wave but composed of infinitely many sine waves whos peaks and troughs add and subtract to make the 'turning on' or 'turning off' happen. These new sine waves are new frequencies and thus being emitted by the laser, as weird as it sounds.
Current long-haul fibre transmission (DWDM) systems are up to 160 wavelengths of 100Gb each, for 16Tb/sec (or 2TB for those who prefer bytes), there's already demo systems that do 400Gb/wavelenth which would be 64Tb/sec however it's known that they won't be able to drive long distance (on the order of hundreds of kilometers at absolute most, as opposed to the thousands of kilometers from the current systems).
DC is zero frequency. A signal that is DC never changes. But the universe has only finite time, therefore the signal can only be DC if we make assumptions about its border conditions at the beginning and end of time.
Mike Page no, i didnt) DC is undestandable. never changing dc is not information. the universe and time is not understandable. if you want infinite, never changing, dc then you just need an infinite energy for this useless circuit. wtf
At about 8:04 (though I can't measure the time precisely) I thought his simple example exceeded my *mental* bandwidth... but by the end, I think I got it!
what he's saying is that when you modulate a light source, harmonic frequencies are added to achieve that modulation, which consumes more of the medium's bandwidth. using a red and green laser is essentially just a way of utilizing more bandwidth, so it's just another way of doing the same thing. google "square wave fourier transform" and you'll understand why this is the case.
Square waves consist of the fundamental, and all the odd harmonics. So the pulse is as square as the bandwidth allows, with shorter and shorter pulse rise times the more bandwidth that's available. The rise time of the pulse affects the shortest pulse you can transmit. Also see "shannons law" which relates bandwidth and signal/noise ratio to determine the maximum throughput.
Are you sure that you don't have a DSL connection? Considering the fact that the slowest speed i can get with fiber optic connection is 10 down and 5 up makes me wonder.. I can get, depending on the speed/plan you pick, anything from 10Mbit/s and all the way up to 1Gbit/s both up and down, for a fairly low cost. I currently have 100Mbit/s and pay about $45/month (USD).
The waveform produced when you turn a laser on and off CAN be represented by the sum of many, many frequencies, but the photons actually leaving the laser are all (nearly) the same frequency, always.
It seems a lot of people want/need more explanation about why a laser being switched on and off somehow produces multiple frequencies; whereas a laser switched on produces a single frequency? This one bit of this "Maximum Bandwidth"* episode would be a great reason to revisit with Professor Merrifield and reminisce about days-gone-by... while making a nice new video for those of us out here. *(which, by the way, I was hoping would be about the "Lowest Frequency" known/possible -- if you get what I'm saying)
toggle a laser on/off and you get lots of sudden frequency changes. in math you can examine this with the fourier series. you can see how a simple on/off signal is created with waves in the first image on the "fourier series" wiki page. you'll find that you need infinitely many waves at different intensities to create a perfect rectangle. but fiber cables have a limited freq range so you can't have a perfect rectangle, the edges will lean. this results in a limit how short your pulses can be.
Very cool explanation. I love everyday applications of quantum principles. The whole video I was wondering if he was going to get to the absolute limit of fiber. Glad you pinned him down in the end.
If you rephrase Enceptics' questions to be a setup like the EPR test, it makes more sense. You will have two photons of exactly opposite wavefunctions. You can measure them both separately as accurately as you want. The problem is how you bring those two measurements back together again.
Does nature take care of all of it though? I remember a lecture in linear systems where my professor was talking about semiconductor designers of old going to great lengths to generate 3rd, 5th, 7th etc. harmonics so that integrated circuits could produce a "nice" fast square edge pulse.
You CAN use red and green, but infrared is normally used and the wavelengths are usually spaced 20nM apart (e.g.1510 and 1530nM). They can be much closer, but that gets expensive.
Try imagine a swinging penduluum. while it swings it doesn't need interaction. But to make it start and immediatly stop again you need to apply force 2 times. So a continous wave swings freely while a puls is a excited and dimmed wave.Another example are clicksounds of defect audio cables or old records lighting up the whole graphic analyzer on your Hifi rack, because as a sharp pulse thy contain lots of frequencies. For more inforemation look up Delta Distribution or Dirac Puls.
You need a little wave mechanics to understand it, but I'll try to help your intuition out with a example from sound waves: A constant tone is a time-varying oscillation - same goes for light if understood classicly with Maxwells equations (periodic oscillation in the electric and magnetic field). For very short periods of time however, you dont get, lets say even a period of the tone signal, which means its "tone" is'nt really defined. Its actually a mixture of a lot of waves. As for the laser.
The 'beating' effect at the beginning of the video is due to the difference - 1 hertz difference produces 1 beat per second, 2 hertz produces 2 per second, etc.
You can do that. You can send information in morse code at some, possibly large speed. But you get to a point where you your pulses, or spaces, become shorter and shorter. If you use simple DC power, interrupted for spaces, and connected for pulses, you will have a rudimentary signaling system, fully capable of carrying anything you wish to encode into that medium. But you eventually get to a point where your pulses or spaces come so close together at the other end that they can no longer ...
Prof Merrifield on several occasions in the video confuses bandwidth with information or information rate and to see the difference he should look up Shannon's channel capacity theorem. So for example if a communication system uses pulse of different sizes then more information can be transmitted compared to a system that only uses one size of pulse even if the systems have the same bandwidth. Indeed this is used in digital data formats such Quadrature Amplitude Modulation (QAM).
I believe that is exactly what the receiver is doing at the end of a fiber optic cable, but it's the properties of the cable itself that limit the pulse length (and thus how closely pulses can be packed together), not the receiver. Because a pulse is fundamentally composed of a range of wavelengths, the smooth shape of the pulse (and therefore readability by a receiver) begins to degrade once that range extends outside the range which the cable is able to transmit.
I was very interested to learn about the limit, and was quite shocked when he said a terabyte of data in a fraction of a second! I would be interested to know if the spectrum of light that the fiber-optic cables can transfer could be broadened, and if so how much by and in which directions? Also is there a similar limit on Wi-fi?
2:15 those factors of intervals put the limits of transmission over the capacity to detect the fractions, which, by generating irrational numbers, might be quite impossible to reach a theoretical limit about how much data can be transmitted.
I guess the only other thing to keep in mind about what svampebob007 said is the fact we don't get our own direct line of fiber optic, the bandwidth is usually shared among many different homes or buildings, which is why we don't see 58.8TB/s download speed and ends up being ~100MB/s.
decay, that superposition demonstration is really really cool! (tiny mistake: "wavelength" is backwards... that value actually controls the _frequency_)
It's actually a Terabit per second to be precise. but that's only the ideal case but in practise it depends on the material the fibre optic was made of because there are always some windows (wavelengths) you cannot use to send data.
This is a somewhat abstract, mathy way to describe what shape of wave is created when you switch a laser on and off. However, lasers really do only produce a small band of light centered at a single frequency. This frequency is related to the bandgap energy of the semiconductor materials used to produce the laser, and that center frequency does NOT change just because you switch the laser on and off. (you wouldn't see the laser light change colors even if you could observe it fast enough).
I think the drawing was not perfect. It would be easier to describe it if he would start the second one not from the zero but from the top of the sin. Then you would clearly see that this fast jump in value is indeed much higher frequency. Small frequency - slow changing, big frequency - fast changing signal. In other words, everything what looks sharp and pointy on the graph will be high frequency.
There is a similar limit on Wifi, but it's artificially constrained. Over the air frequencies are tightly controlled and licensed, so the unlicensed band that Wifi operates on is smaller than would be physically possible if you had the whole spectrum to work with.
I'm not sure why you say that. If you observe a laser turning on fast enough, which is quite possible with some pretty basic equipment, you certainly do see the spectrum of the light changing. In fact--though I believe Prof. Merrifield is correct that this isn't used in communications--mode locking is a very common technique for making pulsed lasers in fields that require high-power laser light. The Wikipedia article about mode locking is pretty good if you're interested.
If I may make a brief clarification (and apology to any engineers I have annoyed along the way!): this is a simple physicist's explanation of the fundamental physical limit as to how much data can be crammed down an optical fibre. Engineers are very smart at optimizing the solution to this problem, but those solutions do not eliminate the fundamental physics limitations. For example, frequency modulation faces the same issue that a higher rate of modulation uses a broader spectrum of light.
Everyone on the internet uses fibre optics......at least sometimes. That's what is used in the under sea cables that link the continents. As far as local internet infrastructure, they are rolling out fibre optics in Australia now. Some people already have fibre to the home in Australia.
Yes. The problem is simply that since the cable has a non-zero radius, there are several paths for light to travel, and some are shorter than others. The light that goes straight down the center of the fiber (fastest) arrive before the light that bounces around the inside of the cable (slower), resulting in a gaussian shaped distribution of arrival times. Has nothing whatsoever to do with the source.
Look at the diagrams he grew, the perfect sine wave going to infinity and the pulse that has tails at each end of it that drop off to zero. To get them tails you need to have other frequencies destructively interfering with each other (which you can hear happening when he has the two phones playing tones, that pulsing sound). It isn't a problem, like he said, you need multiple frequencies to transmit data and the laser automatically produces them by turning it on and off.
Planck's Law is as follows: rho(T)df=(8(pi)h/c^3)(f^3df/(e^(hf/kT)-1), where rho(T)df is the energy density, h is Planck's constant, c is the speed of light in a vacuum, f is the frequency, k is Boltzmann's constant, and T is the absolute temperature.
Ironically, the video stopped to buffer as soon as he mentioned "bandwidth" for the first time. I chuckled.
hehehe
no it didn't
I fapped
Oskar,
Do you not know what buffering is?
I conclude that the bandwidth of your line is less than 800 terabits per second. ;-)
I like the part were the Professor says he doesn't know how much long is the shortest pulse, but says "I can look it up if you want". Such great minds have no problem with not knowing something, while this planet is so full of know-it-all s.
Humility will get you a long way in this field!
10:18
*cough*Neil Dygrasse Tyson*cough*
Its just a number, i dont think anyone is proud of remembering numbers
@@thedude951 Degrasse
@@giorgosd3624 It's*
don't*
"A terrabyte of data.... in a few hundredths of a second."
*Salivate*
Run lots of cables in parallel. Problem solved.
And that's on a single wire. I imagine you could have 1000s of wires on a single cable, yes? Maybe you run into more physical limitations or errors at too small of a size too, though
Run lots of cables in parallel yes but also with slightly different properties allowing them to be a control for overall transmission (plus 'overhead') for further range of frequencies (handling things as a bundle). Ie the different cables extend in part the range of frequencies available resulting in more relative bandwidth per cable.
There is an energy density problem, and mutual coupling between wires. Shannon's theorem says that data rate is given by bandwidth times signal to noise ratio. (For our purposes, dynamic range is a limit on signal to noise ratio.) A one Hertz bandwidth on a single cable can transmit a terabyte per second with a mere 44 teradB of dynamic range, or equivalently 8 x 10^12 bits of resolution.
all we need is to start encoding data in sperms and transmit them through pipes. they carry terabytes of data per teaspoon, so if we want to copy entire massive databases just have a sperm truck go to the new location, then do a memory check to see if everything transmitted and use the internet to make up the tiny difference.
Wow, that was the best explanation of fourier transform I've ever seen.
Ali Moeeny I am not sure. He introduced some math, but he steered clear of even naming FT. I think that if he directly went for the concept, explained it, and made it part of the vocabulary, the whole explanation would have sound less sketchy.
+Ali Moeeny What are you talking about, he didn't explain FT at all.
+Ali Moeeny I was kind of surprised that I got to the end of the video and Fourier Transform was never brought up.
I'm not a physicist, but once you get used to this type of video you start to understand them better and better, it just takes time,
Michael Merrifield , I just wanted to thank you for making me passionate about topics I never knew I had passion for until I saw your videos here and on Deep Sky Videos. You have pushed me to learn and understand some of the involved math, encouraged me to purchase a telescope for skygazing purposes, and opened up new areas for me to explore. I am more grateful to you, Ed Copeland, Meghan Gray, Phil Moriarty, and Roger Bowley than I could ever express.
The word of the day is "fourier series".
*Pee Wee and the audience go apeshit*
That is two words.
harmonics :)
I think our science has gotten to the point where a little Fourier De-synthesis might be just what the doctor ordered.
That's not one word. It's an infinite number of words with varying phases and amplitudes added together. 😉
This explains why musical instruments make that "wobbly" sound when they're out of tune on the same note. AMAZING!
yeah, especially in guitars you can hear beats and harmony off diff. instruments creating non-periodic reverbs, musicians are pretty close to science compared to other non-scientific fields..
Why wasn't the term heterodyning used in this video? Bandwidth requirements become much more understandable, when the concept is explained. Two frequencies will combine to create both constructive and destructive frequencies. Heterodyning was demonstrated with the iPhone-based tone generators, but not clearly explained. If a 1000Hz tone is pulsed at 1Hz, the transmitter will occupy 2Hz from 999-1001Hz.
Today's data networks use something called Dense Wave-division Multiplexing. We combine multiple frequencies often near 190THz with spacings of a mere 0.8nm, each channel representing 10Gbps. Prisms are readily available to combine up to 80 channels on a single pair of fibers, which translates to 800Gbps, or a Terabyte of raw data in 10 seconds.
@Michael Hall: What you're talking about is the actual binary information that is transmitted. The video is concerned about the physical limits in transmitting information accross a fiber-optic cable. And mind you, as the professor clearly states, a mathematically true pulse requires an infinite amount of frequencies or otherwise an infinite amount of preparation and decay times to form. Fortunately quantummechanics comes to the rescue there giving us a bounded physical world i.s.o. a purely mathematical one. I really love that part.
I have to admit, I have never heard of the range on frequencies that are able to travel down the optical fiber to be the limiting factor. I was taught that it was the Group Velocity Dispersion that limited the data rate, as due to this effect the pulses spread out temporally (That is, the pulses duration continues to increases as the pulse travels down the fibre). This increase of pulse duration makes it so that if pulses are fired too quickly after each other, sometime down the fiber the pulses would spread into each other and you will lose your nice wave packets and hence your data.
That is true. However he's talking about different types of limit. It is also strange that he talked about visible spectrum when fibre optics use only IR.
Well I too mostly understand what is explained, but not quite this time. He lost me at the point that he explained that if you switch a laser on and off, it is a superposition of waves and if you keep it on continuously, it isn't. I understand the Heisenberg uncertaintly principle, and I understand superposition, but I cannot understand what that has to do with it. I also understand the (unmentioned) Fourier analysis, but I cannot make one single image in my mind on what the difference is between switching a laser on/off and leaving it on continuously.
Ronald de Rooij The thing that he said about continuous wave (CW) is true. It is almost never really exists in real world. I do a lot of EM simulation, and it is a real problem to simulate a CW. We use different kinds of tricks to suppress high frequency signals that permeates from starting the so called 'CW'. These high frequency signals are real problems in our frequency domain analysis.
While I understand the argument being made in the video, it feels like he's contradicted himself. I was under the impression that individual photons represented a packet of energy of a single frequency... What he's saying suggests that a photon itself is a superposition of multiple frequencies?
earthworm768 - actually group velocity dispersion (group delay in the radio world) means that different wavelengths travel through the fibre at different speeds and the frequency components arrive at different times which blurs the pulse edges. it can be partially corrected. it is still a function of bandwidth
You sir explained a whole lot of wireless communication to me in minutes! Thanks prof. Your students are so lucky to be taught by you! Please consider doing more videos on wireless communication. Yours will make a great channel.
I want to be in his class
you say that b/c you saw the bottle of scotch on the desk behind him LOL, that would be a fun class
No one is in his class.
Brady, don't forget the sixty symbols channel. I love them all but this is my favorite and you don't t upload enough here!
The number of "beatings" per second its equal to the difference between the two frequencies in hertz. So, 1000Hz and 1001Hz: 1Hz difference, or one beating per second. 1000Hz and 1002Hz: 2Hz difference, or two beatings per second.
When he played two tones of different frequencies, superposition of two signals took place. It will result into wave/signal of different frequency. Therefore, that resulting frequency will be LCM of the two original frequencies.
This just cleared up a huge hole in my understanding of waves and bandwidth. I heart you guys lots.
A sine wave can convey three pieces of information. Amplitude, Phase, and Frequency. The 19kHz FM pilot tone used to decode stereo information (United States) is an example.
You can think of turning a sine-wave on/off as multiplying the sine-wave by a windowing function that is 1 when the sine-wave is on, and 0 when the sine-wave is off. The windowing function has a frequency spectrum of its own, and this gets convolved with the sine-wave's spectrum in the frequency domain. A pure sine-wave's spectrum looks like two impulses (sharp spikes) at +f and -f. A rectangular (perfectly "square" pulse) window function's spectrum is a sinc (sin(x)/x) function. When the sinc spectrum gets convolved with the sine-wave's impulse spectrum, it copies, shifts, and divides by 2 the sinc spectrum around +f and -f. Negative frequencies arise from converting Real valued time-domain signals into their complex-frequency domain representation.
Check out the Shannon-Hartley Theorem that relates maximum channel capacity to Channel Bandwidth and Signal-to-Noise ratio.
I think you missed the point he was trying to convey.
That's not the *type* of information he is referring to.
That is nothing more than the nature of the transmission. You need an oscilloscope to translate that.
Hes talking about using that like morse code.
Sounds a bit... _convoluted_ to me
Sir, this is the most beautiful and concrete explanation of Quantum physic, Fourier Transformation and Information theory lectures in the whole universe!!!
I'm a engineer and finally after all those years someone showed me how to explain connections in all of those laws using one simple example!. This really worked for me!!! Thank you!!
That moment when I realize this is not Computerphile o.0
+Juan Bonnett the moment when someone realise or deeply interiorise that the basic sciences are actually the thing that the modern technology comes from.
+Lazeran that moment when you realise that this is topic is grey and could be found on either one
Bandwidth isn't computerphile? With current technology there is a known maximum bandwidth that we could Mor's Law predict ourselves into a problem with, in a relatively short time frame. At that point laying more wire to fill an exponentially increasing demand isn't plausible, you need better technology with a higher bandwidth or better compression to send less data.
This all seems pretty computer related, to me.
i know.... i was like :0
All he talks about is in fact rudimental Computer stuff.
*You learn what bandwidth is and how it is “produced“ when you are learning any computer related job!*
As a non-physicist I was doing great for the first 4.5 mins and then it all went! However I still love watching these videos of Brady's in the hope that I will start understanding eventually. Still envious of how Brady seems to understand so much as shown by his responses. I am beginning to think he no longer represents the "ordinary" person and so would like to volunteer my services for future videos!
Can't wait for 10TB/s speeds...
+Engineering Nonsense I know right.
+Engineering Nonsense But that will need to be split through the whole population since we share fibre cables
Matthew Lowe its not just one cable, under water, but networks. Right?
Yeah, I was going to say there will be thousands bundled up, but I doubt there will be enough to provide each household within a city with 10TB/s
Matthew Lowe True
It's brilliantly explained. He starts with the beats that happen when two waves of different frequency superposition against each other (with a handy graph showing it beating in real time). He then draws an "infinite" sine wave and a pulse.
You can clearly see that the pulse has tails, these tails that go to zero which is clear evidence that something other than the sine wave (like another frequency) is affecting the wave.
A pure sine wave would just stop with no tails.
What is the ultimate limit for sending through the vacuum of space?
That would be whole spectrum from kilometers long radio waves to planck length sized photons.
It would be some absurdly large number
I think the lower bound is definitely the size of intergalactic dust (so perhaps IR?) I don't think there is an upper bound tbh, since space can have almost no mattet (a few atoms per m^3).
@@vaibhav1618 what about the bandwidth of a signal sent between two Casimir plates at vacuum? that'd be interesting to graph out.
@@voxelsofsorrow interesting, but what are we charting? Bandwidth vs plate separation?
@@vaibhav1618 exactly! it's kinda interesting because the plates will cut off lower frequencies while allowing higher frequencies as they get closer
This is my favorite video of all sixty symbols!! Please make a video about Fournier's equations!!
So that would be around 80 Tbps or 80.000 Gbps. Should be enough for future applications.
Looks like some people even managed to almost reach the maximum.
www.techspot.com/news/52066-hollow-fiber-optic-cable-tops-73tbps-promises-near-light-speeds.html
We really need fiber optical cables everywhere and into every building. You'd only have to upgrade the terminals and recievers to get higher bandwidths in the future.
Professor Merrifield is the best. I could listen to him all day.
The reason why you don’t switch on and off and use a sum of sine waves to make 1s & 0s instead (Brady’s unanswered question) is that by modulating multiple sine waves you can actually encode a crapload more information per second into the optical fiber. The resulting non-sine signal contains many more contemporaneous 1s encoded.
Professor Merrifield explained instead that you can simulate the “ON/OFF” physical phenomenon with a finite sum of many sine waves. The more numerous they are, the more squared is the resulting wave, and the easier it is to tell a 1 from a zero, hence a wider bandwidth.
2:04 a resonance. A pattern of the waves complimenting each other and getting stronger, or canceling out and making the noise more fuzzy. Like pushing your legs on a swing set
I learned something new! What a fantastic explanation. I had no idea that optical communications relied on the wave component of light to transmit information. Thank you so much Professor Merrifield. I absolutely love watching your videos. Don't be discouraged by negative comments. You can't please everyone and the people you do reach are very grateful for sharing your knowledge with us.
This reminds me of how they said CAT7 cable was going to be needed to handle the bandwidth of gigabit Ethernet and within a few years the engineers figured out how to do it over CAT5 cable with the right processing power and better noise reduction algorithms. Then wireless added new modulation schemes and multiplexing and jumped from 802.11b to 802.11n to 802.11ac. The lesson: don't tell an engineer there is a limit as they will find ways to exceed your limits.
Stephen Furr I absolutely didn't get a clue of what you're talking about except the last sentence that I like very much and do agree with :)
allen mathew Thanks :)
+Stephen Furr Advancements in technology only push the speed near and near to the fundamental limits of the media. If the media is for example a CAT5 cable with a given noise floor you can only push so much information down the line (see Shannon theorem).
here here
Stephen Furr there's a limit ;)
That's what the video is about, the shortest period you can send the light for has a limit; the shorter the period, the more frequencies of light that are required and fibre optic cable only has a limited amount of light frequencies it can transmit - which we generally call 'visible light'.
Talking about photons got me thinking. What is the smallest wavelength difference that two different photons could have? I understand that each photon has a particular energy, determined by its wavelength, so that a photon can have wavelength x, and the next highest energy photon will have wavelength x-n. But what is the value for n? Is it the Plank length?
+E “Anonymous Nerdfighter” Hernandez The smallest wavelength difference, I would assume, would be one Planck Length. The reason is because the Planck Length is the smallest discrete distance that can be traveled and measured in the universe, so it would be impossible for two wavelengths of light to differ by any less than that.
I'm sure someone has already pointed this out, but the bandwidth only basically tells you how many pulses (or symbols) per second you can send. But you can actually send much more data than the number of pulses, basically, by sending pulses of different shapes. The fundamental limit (Shannon limit) is about the ability to distinguish the different pulse shapes from each other.
If quantum noise and uncertainty is your only noise in the fiber, then the fundamental limit for sending data is actually much much much more than the mere ~10^15 bits/s. The trick is that the pulses are not made of single photons, but of a number of photons. Pushing more energy increases the signal-to-noise ratio, i.e., lessens the uncertainty regarding the aggregate pulse shape.
For reference, DVB-C2 (digital video broadcasting for TV cables) specifies QAM-4096 as the densest coding to pulses. This basically means that there are 2^12 = 4096 different pulse shapes, i.e., 12 bits per pulse can be transmitted.
So, the fundamental limits in the video apply when you're sending only one photon per pulse. Ultimately, I guess the limit is how much energy the fiber can transmit before melting.
Dr. Merrifield please don't pay any mind to comments like this. You're my favorite professor on any of Brady's channels. There are many more people who feel the exact opposite of those who make comments like that one!
As a big fan of your videos and someone who normally has no problem understanding them, I have to admit none of this made any sense to me. I would greatly appreciate it if you could make another video explaining this better. I know you have lots of videos to make but this seemed interesting but was really not explained well in my opinion, so if you get a chance I would love an opportunity to actually understand what he is talking about here.
Yeah normally i understand everything pretty well too but this time i really struggled to understand anything of substance the first time watching. By now i only understand the first part of the video, where the delta f * delta t = 1 formula comes from.
(1) The speed of your fiber connection is how much information can travel through
(2) Information is 1s and 0s
(3) "1" and "0" are basically "pulse" and "no pulse")
(4) The shorter in duration the pulses the more pulses can travel through your fiber within a given period of time
(5) A short (duration) pulse can only be created by combining loooots of waves of different frequencies
(6) There is a limitation on what range of frequencies (bandwidth) your fiber can support
(7) Since the range of frequencies is limited, the duration of the supported pulses is also limited, because of (5)
"One of the things people know about quantum mechanics is this wonderful thing called uncertainty principle." This part got me chuckled.
Terabytes in milliseconds you say? That's amazing.
KaneLongTroy If we assume 1tb in a millisecond, that will still be done in 0,002 seconds. That's pretty fast.
+GamePhysics Regular computers won't be able to process that much of speed
A brilliant explanation, and really helpful for my A-Level Physics! Thanks.
sir what was that app for different frequency generation ? i need it.
Search for frequency genrator apps on play store or app store. To replicate that pulsing effect, choose a stable frequency let's say 1khz and then play 1.01 khz in other phone.
What a masterful description of the uncertainty principle
I calculated his answer a bit more so its easier to understand.
In terms of data transfer the ultimate limit you can send is about 50,000,000 Mbyte/sec or 400,000,000 Mbit/sec in a fiber lika that (from blue light to red light)
Let's see the math.
Google Shannon's Theorem.
I believe people like him feel the need to post such comments, not because they want to help you or themselves, but because of their own inability to understand, and on account of their character, instead of trying to be constructive, they're unhelpfully critical of the explainer. Such people are not worth paying attention to. I think you explained this, and other topics very well. I derive much learning from watching your videos.
Taking the dispersion effects into account the datarate would significantly drop as the phase velocity of all those frequencies across the visible wavelength wont be the same. On top of that different excitation modes withing the fibre would be excited causing even more dispersion. Along with all this we got non linear raman effects. The real limit would be wayyyyy lower in any practical application.
Of course that is my understanding, based on high school science some thirty plus years ago. Now everyone who really understands his question can attempt to reply correctly if I am not doing so, in terms such as I have used.
Thanks.
Found this video really helpful, appreciate you guys taking the time to film and upload it
Thank you Prof Merrifield and Sixty Symbols. This is a very complicated topic I couldn't get my head around as I find it hard to understand the structure of a wave.
You actually could encode information in an unchanging sine wave; you could encode it in the exact frequency. Let's say I wanted to encode the word "it". If you break "it" down into ASCII, that's 0x69, 0x74 (in hex), or 0110100101110100 (squished together in binary), depending on how you decide to do it. Anyway that translates to the number 26996 in decimal. So, if you wanted, you could send out a tone at 26996 Hz. If another person listened to the signal and figured out what frequency it was (easy with computers!), they could decode that number into "it"! But yeah, this is obviously a silly idea. Just having some fun!
Well, there is a problem with the "easy with computers" part. So far with the word "it" we are outside the frequency range of digital audio, so you will need special hardware to digitize the incoming 26996 Hz signal.
And "it" is a very short word.
I'll let you calculate the frequency of the word "information" with your system, but I guess we would in the 10^27 Hz range...
By the way, an unchanging sine wave has to be infinite in time, to be unchanging. Otherwise it'll contain other frequencies.
As you said, having some fun.
There is also the problem of having the different peaks and valleys of different frequencies smearing together and also losing the information.
Different vortices of different colors of light, each carrying its own unique data stream, can all be packed together into a donut-shaped beam of light that travels along the cable. Optical devices can then filter out individual vortices at the other end of the signa
Thanks for freaking out my kitten.
This video is 100% correct when talking about the bandwidth of the fiber by itself. In practice, we need to drive the laser with electronics which are much more limited in bandwidth than the fiber. So we use multiple lasers with different wavelengths. Lookup WDM and DWDM..
11:22 = O_o
I need that internet speed!
your computer won't be able to handle it.
i wonder if it would melt the fiber
Those crazy levels of speed are more commonly found in trans-oceanic cables, where billions of people's bits are going over a single bunch of fibers (like a shared cable internet connection on steroids).
G4mm4G0bl1n I never said it was a single fiber. I said "cable" and "bunch of fibers". Those cables are massive bunches of fibers.
My point was that the only real need for anywhere near theoretical single fiber bandwidth is in those transoceanic fibers. No home or business network, metro net, ISP, etc. needs that at this point.
G4mm4G0bl1n Not at all. I specifically said "at this point" not "never". As long as LANs are capped at 1 maybe 10 Gbps, WANs need not be faster. Will the need be there soon enough, of course. But not today.
Visualize the pulse as a function, with time as the X axis. Now visualize the spectrum of the pulse, with frequency as the X axis - the spectrum will have a large bump of a certain frequency width.
Picture keeping the graph of the pulse the same while you scale the time axis (stretch or shrink the pulse). Similarly, your mental image of the spectrum of the pulse is not changing, but the frequency axis is being scaled. Shrink the time axis, and the frequency axis stretches, and vice versa.
Thank you Professor - never really understood how Bandwidth limits were arrived at - until now that is!
and im stick with 5 down and 0.85 up :(
Once you phase shift modulate a sine wave, it is no longer a single frequency. As we saw in the video, any modulation of a sine wave creates a signal which has bandwidth.
Modulation of the wave frequency and its amplitude, can increase the bandwidth of the fiber.In addition, it is possible to transafe the light over a diffrent angles within the fiber which.
Mammalian nerves utilize an ionic electrochemical system to transmit info, and although it can be described as electrical, it is much different than a coax cable or an optical fiber, having a maximum rate of about a kilobit per second. However, the nerve that takes info from the eye to the brain is not a single fiber, but a bundle of about 4 million that convey info in parallel. So about 4 gigabits per second within an order of magnitude, I'd guesstimate.
So that actually means that in the far future we will be trading TERABYTES instead of MEGABYTES in our computer connections? This is just awesome! Thank you for explaining the science behind this...
Look up fourier transform - when you turn a sine wave on or off, it is no longer one sine wave but composed of infinitely many sine waves whos peaks and troughs add and subtract to make the 'turning on' or 'turning off' happen. These new sine waves are new frequencies and thus being emitted by the laser, as weird as it sounds.
He also has the benefit of constantly interviewing very smart people who are really good at explaining things.
Current long-haul fibre transmission (DWDM) systems are up to 160 wavelengths of 100Gb each, for 16Tb/sec (or 2TB for those who prefer bytes), there's already demo systems that do 400Gb/wavelenth which would be 64Tb/sec however it's known that they won't be able to drive long distance (on the order of hundreds of kilometers at absolute most, as opposed to the thousands of kilometers from the current systems).
There is no such thing as DC.
If we only had infinite time.
Correct!
very interesting. Can you tell more about it?
Do you mean a light-speed electrons movement?
a simple wire faster than optical? then why? whaaaaaa ????
DC is zero frequency. A signal that is DC never changes. But the universe has only finite time, therefore the signal can only be DC if we make assumptions about its border conditions at the beginning and end of time.
Mike Page
no, i didnt)
DC is undestandable.
never changing dc is not information.
the universe and time is not understandable.
if you want infinite, never changing, dc
then you just need an infinite energy for this useless circuit.
wtf
At about 8:04 (though I can't measure the time precisely) I thought his simple example exceeded my *mental* bandwidth... but by the end, I think I got it!
what he's saying is that when you modulate a light source, harmonic frequencies are added to achieve that modulation, which consumes more of the medium's bandwidth. using a red and green laser is essentially just a way of utilizing more bandwidth, so it's just another way of doing the same thing.
google "square wave fourier transform" and you'll understand why this is the case.
Square waves consist of the fundamental, and all the odd harmonics. So the pulse is as square as the bandwidth allows, with shorter and shorter pulse rise times the more bandwidth that's available. The rise time of the pulse affects the shortest pulse you can transmit. Also see "shannons law" which relates bandwidth and signal/noise ratio to determine the maximum throughput.
I get 5 mbps -__-
Are you sure that you don't have a DSL connection?
Considering the fact that the slowest speed i can get with fiber optic connection is 10 down and 5 up makes me wonder..
I can get, depending on the speed/plan you pick, anything from 10Mbit/s and all the way up to 1Gbit/s both up and down, for a fairly low cost.
I currently have 100Mbit/s and pay about $45/month (USD).
Gnagarn
100 mbit/s in USA or the world, dl and ul???
750Kb/s keep whining
The waveform produced when you turn a laser on and off CAN be represented by the sum of many, many frequencies, but the photons actually leaving the laser are all (nearly) the same frequency, always.
It seems a lot of people want/need more explanation about why a laser being switched on and off somehow produces multiple frequencies; whereas a laser switched on produces a single frequency?
This one bit of this "Maximum Bandwidth"* episode would be a great reason to revisit with Professor Merrifield and reminisce about days-gone-by... while making a nice new video for those of us out here.
*(which, by the way, I was hoping would be about the "Lowest Frequency" known/possible -- if you get what I'm saying)
toggle a laser on/off and you get lots of sudden frequency changes. in math you can examine this with the fourier series. you can see how a simple on/off signal is created with waves in the first image on the "fourier series" wiki page. you'll find that you need infinitely many waves at different intensities to create a perfect rectangle. but fiber cables have a limited freq range so you can't have a perfect rectangle, the edges will lean. this results in a limit how short your pulses can be.
Very cool explanation. I love everyday applications of quantum principles. The whole video I was wondering if he was going to get to the absolute limit of fiber. Glad you pinned him down in the end.
A short lecture on various kinds of modulation that look as if they beat the Nyquist limit would be a nice addition to this.
If you rephrase Enceptics' questions to be a setup like the EPR test, it makes more sense. You will have two photons of exactly opposite wavefunctions. You can measure them both separately as accurately as you want. The problem is how you bring those two measurements back together again.
Thanks! These technicalities are why I love science.
Does nature take care of all of it though? I remember a lecture in linear systems where my professor was talking about semiconductor designers of old going to great lengths to generate 3rd, 5th, 7th etc. harmonics so that integrated circuits could produce a "nice" fast square edge pulse.
You CAN use red and green, but infrared is normally used and the wavelengths are usually spaced 20nM apart (e.g.1510 and 1530nM). They can be much closer, but that gets expensive.
Try imagine a swinging penduluum. while it swings it doesn't need interaction.
But to make it start and immediatly stop again you need to apply force 2 times. So a continous wave swings freely while a puls is a excited and dimmed wave.Another example are clicksounds of defect audio cables or old records lighting up the whole graphic analyzer on your Hifi rack, because as a sharp pulse thy contain lots of frequencies. For more inforemation look up Delta Distribution or Dirac Puls.
You need a little wave mechanics to understand it, but I'll try to help your intuition out with a example from sound waves: A constant tone is a time-varying oscillation - same goes for light if understood classicly with Maxwells equations (periodic oscillation in the electric and magnetic field). For very short periods of time however, you dont get, lets say even a period of the tone signal, which means its "tone" is'nt really defined. Its actually a mixture of a lot of waves. As for the laser.
I do radio as a hobby and this is a great video.
The 'beating' effect at the beginning of the video is due to the difference - 1 hertz difference produces 1 beat per second, 2 hertz produces 2 per second, etc.
Brilliant - Of all the explanations, I've seen, this is the best explanation of bandwidth.
You can do that. You can send information in morse code at some, possibly large speed. But you get to a point where you your pulses, or spaces, become shorter and shorter. If you use simple DC power, interrupted for spaces, and connected for pulses, you will have a rudimentary signaling system, fully capable of carrying anything you wish to encode into that medium. But you eventually get to a point where your pulses or spaces come so close together at the other end that they can no longer ...
I feel for the dude as he is trying to explain in layman terms workings of F-transform, without actually touching on the math.
2020 Rewatch & I still wish I had enough money to be in this guys classes.
Prof Merrifield on several occasions in the video confuses bandwidth with information or information rate and to see the difference he should look up Shannon's channel capacity theorem. So for example if a communication system uses pulse of different sizes then more information can be transmitted compared to a system that only uses one size of pulse even if the systems have the same bandwidth. Indeed this is used in digital data formats such Quadrature Amplitude Modulation (QAM).
I believe that is exactly what the receiver is doing at the end of a fiber optic cable, but it's the properties of the cable itself that limit the pulse length (and thus how closely pulses can be packed together), not the receiver. Because a pulse is fundamentally composed of a range of wavelengths, the smooth shape of the pulse (and therefore readability by a receiver) begins to degrade once that range extends outside the range which the cable is able to transmit.
I was very interested to learn about the limit, and was quite shocked when he said a terabyte of data in a fraction of a second! I would be interested to know if the spectrum of light that the fiber-optic cables can transfer could be broadened, and if so how much by and in which directions? Also is there a similar limit on Wi-fi?
2:15 those factors of intervals put the limits of transmission over the capacity to detect the fractions, which, by generating irrational numbers, might be quite impossible to reach a theoretical limit about how much data can be transmitted.
I guess the only other thing to keep in mind about what svampebob007 said is the fact we don't get our own direct line of fiber optic, the bandwidth is usually shared among many different homes or buildings, which is why we don't see 58.8TB/s download speed and ends up being ~100MB/s.
decay, that superposition demonstration is really really cool! (tiny mistake: "wavelength" is backwards... that value actually controls the _frequency_)
It's actually a Terabit per second to be precise. but that's only the ideal case but in practise it depends on the material the fibre optic was made of because there are always some windows (wavelengths) you cannot use to send data.
Mike Merrifield! Best teacher!
This is a somewhat abstract, mathy way to describe what shape of wave is created when you switch a laser on and off. However, lasers really do only produce a small band of light centered at a single frequency. This frequency is related to the bandgap energy of the semiconductor materials used to produce the laser, and that center frequency does NOT change just because you switch the laser on and off. (you wouldn't see the laser light change colors even if you could observe it fast enough).
I think the drawing was not perfect. It would be easier to describe it if he would start the second one not from the zero but from the top of the sin. Then you would clearly see that this fast jump in value is indeed much higher frequency. Small frequency - slow changing, big frequency - fast changing signal. In other words, everything what looks sharp and pointy on the graph will be high frequency.
This is like the only sixty symbols video wherein i fully understand every word and concept adressed XD
There is a similar limit on Wifi, but it's artificially constrained. Over the air frequencies are tightly controlled and licensed, so the unlicensed band that Wifi operates on is smaller than would be physically possible if you had the whole spectrum to work with.
I'm not sure why you say that. If you observe a laser turning on fast enough, which is quite possible with some pretty basic equipment, you certainly do see the spectrum of the light changing. In fact--though I believe Prof. Merrifield is correct that this isn't used in communications--mode locking is a very common technique for making pulsed lasers in fields that require high-power laser light. The Wikipedia article about mode locking is pretty good if you're interested.
If I may make a brief clarification (and apology to any engineers I have annoyed along the way!): this is a simple physicist's explanation of the fundamental physical limit as to how much data can be crammed down an optical fibre. Engineers are very smart at optimizing the solution to this problem, but those solutions do not eliminate the fundamental physics limitations. For example, frequency modulation faces the same issue that a higher rate of modulation uses a broader spectrum of light.
Everyone on the internet uses fibre optics......at least sometimes. That's what is used in the under sea cables that link the continents. As far as local internet infrastructure, they are rolling out fibre optics in Australia now. Some people already have fibre to the home in Australia.
Yes.
The problem is simply that since the cable has a non-zero radius, there are several paths for light to travel, and some are shorter than others.
The light that goes straight down the center of the fiber (fastest) arrive before the light that bounces around the inside of the cable (slower), resulting in a gaussian shaped distribution of arrival times.
Has nothing whatsoever to do with the source.
Look at the diagrams he grew, the perfect sine wave going to infinity and the pulse that has tails at each end of it that drop off to zero.
To get them tails you need to have other frequencies destructively interfering with each other (which you can hear happening when he has the two phones playing tones, that pulsing sound).
It isn't a problem, like he said, you need multiple frequencies to transmit data and the laser automatically produces them by turning it on and off.
Planck's Law is as follows: rho(T)df=(8(pi)h/c^3)(f^3df/(e^(hf/kT)-1), where rho(T)df is the energy density, h is Planck's constant, c is the speed of light in a vacuum, f is the frequency, k is Boltzmann's constant, and T is the absolute temperature.