The phone system is magic. The ingenious crazyness that (before DSL and ISDN) takes a digital computer, a digital phone system and joins the two together using some analog technology that tries to mash digital data down into frequences used by human voices. The truly clever part is that my broadband phone line still supports an old GPO rotary dial phone from the 80s I have that contains a proper mechanical bell and pulse dialling done through a mechanical dial.
It's pretty cool that, with everyone happy to replace our current physical wiring with fiber, a smart engineer can still squeeze more bandwidth out of copper that was designed to have a fraction of what is has now. I think it's a testament to the fact that we can always improve what we have, even if we are certain it's at its peak.
4:03 Well, the Nyquist theorem doesn't state that you _should_ sample at twice the maximum frequency in your signal, it states that you _have to_ sample at twice that frequency or you will not only not get those higher frequencies in your discretized signal, you will in fact get false frequencies instead (mirrored at the Nyquist frequency). I love the Brits for their polite language but sometimes you really can't replace "must" by "should".
I usually explain it as how a radio can pick up one frequency out of dozens of radio stations. Now imagine all the stations working together to transmit the same piece of music, not 5.1 surround sound channels or even 7.1 but 76.5 channels. The receiver tunes *all* stations at once and splits the sound out to different speakers. Thus you use the whole length of the radio dial to transmit one piece of music, so DSL uses a *broadband* of frequencies (cut up into bins or cells) to transmit data.
A lot of 'computer people' get brainfarts when thrown into analogue signal processing. I have issues with it, but can follow the modulation stuff okay.
It wasn't the best was it? I have some understanding of how this stuff works, and he really didn't explain the basic or advanced concepts as well as some other presenters have with their content.
Kabitu1 Lol, I got to the bit were they played the fax sound and I was like 'Hey, I recognise that' and the rest of the time I was kind of fooling myself into believing that I still remember how AM/FM works.
Kabitu1 I'm not in computers at all, and this one I understood more than most of the other computerphile vids because I'm super interested in audio. I think audio knowledge would help here more than computer knowledge. When he started talking about phase shifting, I was like "man, nobody who doesn't get sound and audio is going to understand this."
80mbps over FTTC (Fibre to the cabinet)! You're pretty fortunate as most of the residents around here struggle to even get 20 over FTTC due to poor quality aluminium cabling and long distances. Vectoring looks like it might be able to increase speeds but, at the end of the day, FTTC, FTTdp (fibre to the distribution point - mini cabinets closer to people's houses) and FTTrN (Fibre to the Remote node, pretty much the same as FTTdp) are just dragging on the life of the copper network, which should be being replaced with pure fibre!
The copper network also supplies phones that work during a local power failure, fibre would require a local power supply and need to be battery backed to supply phone service. And as I found during a large area power cut the other week, the mobile phone system simply drops dead during a power failure, apparently there's no $()(!@#$) backup batteries at the cell site
Yikes! Aluminum as an electrical conductor? I mean sure it's not the worst conductor in the world, but the only reason I can see a company making the decision to use aluminum as an electrical conductor that is meant to carry many frequencies with little resistance is because they're just too cheap to spend the extra dollars on copper. This must just be a UK thing. It's not uncommon in the US to see downlink speeds in excess of 100Mbps. I pull 300Mbps from my ISP over copper wiring, and the only thing preventing me from going faster is the cost of the faster service. You could easily pull close to 1000Mbps over a well shielded copper wire.
@@TheChipmunk2008 Must just be your mobile phone provider. Power was knocked out over a large region of the state that I live in and despite my entire town not having power for close to 24 hours (including street lights and signals), my cell phone still worked just fine. Internet, voice, and all.
@@gfifer1 it sounds like you are talking about coaxial cables, there is no shielding in twisted pair phone cables, even those made of copper. Those speeds are almost impossible to obtain for 90% of people with copper lines thanks to attenuation and crosstalk (and overselling). Thankfully fiber is becoming more and more easily available.
This would have been immensely helpful for my computer networks exam last year. I can't wait until you do videos for the higher layers of the OSI layer.
I must say ADSL is one of the greatest inventions since the first home modems. Certainly a big deal for someone who grew up in rural New Brunswick! Finally! Decently fast internet without a separate line!
How does the dsl 'modem' know what the reference (Zero degree) phase is? for the PSK? Does it have a reference oscillator that's synced up occasionally, or does it measure relative to the previous 'bit' and could thus get out of sync?
How does the Bee Orchid that attracts Bees to come and mate with it and thus dupes said Bee into carrying its pollen know or came to know what a Bee looked like in the first place.
Really really good explanation to the issue, but it feels like the video is heavily edited to fit as much information into the 09:35. Personally I am a ham radio operator so I have fair bit of knowledge with analogue radio, but I still have to concentrate a lot to understand how QAM modulate, which I didn't before. I'd suggest breaking it down further to explain what modulation / carrier signal is, then going into how PSK differs from AM/FM, then how PSK is combined with FM to form QAM.
9:04 Is that a new trend, to have a slightly opaque overlay along with subtitles? The CERNPeople channel does it as well. I can't tell if someone is overwhelmed with their editing tool, or if it's a deliberate mistake, like dubsteb.
At about 6:40 he uses the phrase "tap it out". I hear that as "put the signal through". Right? Like "tap it out over this line". Is that what that meant? And if so, that's got to be a holdover phrase from the days of telegraphy. Like, tapping out a telegraph over the telegraph lines. Am I way off here??
Wait, that 1983 War Games movie... was that little scrub using a modem? and was he connected to the internet.. in 1983??? where did he get the modem from?
Ok, I'll try. If you consider a standard Sine wave, it starts at 0 goes up to 1, comes back down through 0 to -1 and then heads back up to start over again. It does this again and again at a certain frequency. Let's say that that standard Sine wave is 0 degrees out of phase. In other words, it's not out of phase at all. If we were to look at a Sine wave that is 180 degrees out of phase, it would be exactly opposite. We say it's 180 degrees out of phase because you can model the Sine function on a circle. So our first Sine wave started on the right-most point of the circle and went counter-clockwise. If we start on the left-most point of the circle, it's 180 degrees around the circle at that point. Ok, so a Sine wave 180 degrees out of phase is the exact opposite of one at 0 degrees. This means it starts at 0, goes down to -1, comes back up through 0 to 1, and then heads back down to start all over again. You could then build a machine to detect which phase you were in, since all of the peaks are now valleys and vice versa. I'm not entirely sure how it works but I could think of one way to detect the phase of the a signal: Since you know the amplitude and frequency of the signal you're supposed to be detecting, you could build a machine that takes that signal as well as generates it's own signal at the same amplitude and frequency and combines them. If the signal coming in is at 0 degrees, the peaks and valleys match up and they add together. You're left with a signal that has twice the amplitude. If the signal coming in is at 180 degrees the peaks and valleys are opposite each other and therefore cancel out and the resulting signal is completely flat. The detection machine then looks if there is a signal there at all. If there is, it sends out a 1, if there is not, it sends out a 0.
Without getting too technical, I am going to give it a try. AM: Changing the Amplitude of the carrier wave FM: Changing the Frequency of the carrier wave PM: Changing the Phase of the carrier wave AM and FM are fairly easy to understand, I'm going to skip that. PM (or PSK) is basically changing the "starting point" of one signal. Imaging a clean AM signal, it would start from 0, going to +1, 0, -1 then back to 0. This completes the cycle. For PSK, let's assume 0, +1, 0, -1, 0 signal is the "starting position" of this signal. The "opposite" of this signal would be 0, -1, 0, +1, 0. The transmitter can choose to start the next symbol by going from 0 to +1, or from 0 to -1. This represent 2 values, 0 and 1. This change in ramp direction is a change in phase. In real life, PSK used are often of more than just 2 phases, in order to make it "faster". In the latest WiFi technology, 256QAM is used, 256 possibilities (or bits) could be squeezed into just 1 symbol transmitted. The trade off of using higher rate QAM is losing resilient to interference, therefore most digital devices would auto negotiate and slow down if required.
At 0:56: "Phase modulation hasn't been particularly used in the analog space". I beg to differ. For decades analog Police, Amateur, commercial and consumer 2 way radio all used PM. FM receivers can pick up PM because shifting the phase shifts the frequency (and vice versa).
Well, this video don´t really explain how ADSL works, rather digital encoding over analog lines work. Also, ADSL is mostly not usesed any more, XDSL, IDSL and other later was renamed ADSL2, ADSL2+ and so on. Firstly ISDN broke throw the 56kbit/s limit a long time before it even was reached. Already in the early 90-tys ISDN use 2 carrier channels of 64bit/s just using analog filtering and basic bit inversion keying. So when the modem was pushing 28kbit/s, ISDN already reached 128kbit/s. The problem was that it have to use a ISDN to telecom network converter in the other side. And then in the telecom server park a bypass conection to internet. What was done with ADSL was that a special server was introduced that converted ISDN streams directly to internet traffic before the line reached the telecom switches. In this case there was no more any limitation of only 2 channels per telefonwire. Then they could basically use a unlimited number. Later ISDN B-chanel introduces a additional 4 channels limiting the maximum to 384kbit/s... This is also happening to be the maximum limit of the 2G GSM data channels (128kb/s för GPRS and 384kbit/s for edge, this is due to design). With ADSL they basically keeped the old telecom system on the server side, but swiched out the old 6 times 64kHz side to a 256 times 4kHz, increasing the bandwith needed by 3 times, but the number of channels by 40 times. Every channel still carry 64kbit/s, but using a combination of PSK and ASK (supposedly 8 bit PSK and 7 bit ASK? total of 15bit?), total of 15Mbit/s of witch 8 is usable....1998 specifiation Later version 2001 use less overhead increasing the capacity to 12Mbit/s with over 500 chanels also using more robust coding, later versions (2008) increased capacity to 24Mbit/s decreasing overhead and others stuff. VDSL and G.fast does not use ADSL backbone. It uses simular carrier system with 4kHz wide carriers, but they are shifted to a higher frequencies, using up to 4000 carriers but more robust coding, limiting the bandwidth to a maximum of 100Mbit/s even tow the number of carriers is a lot higher. The road side box in this case is connected directly to the internet, not routed to the telecom provider. This enable just about any company to install it. (in theory) Also VDSL and ADSL is designed so it could run simulations on the same cable
What is the rationale behind the Nyquist Theorem? Is doubling the sampling rate necessary to properly replicate a sound digitally or is it done to provide an ample amount of headroom or buffer zone? For instance, since the human ear can only perceive up to 20 khz, would a digital sampling of 20 khz be virtually indistinguishable from 44 khz or is that doubled rate necessary to overcome audible anomalies introduced by the digitization process?
If you sample a 20kHz tone at 20000 samples/second, each sample will be at the same point at the graph, for instance the point of the positive-goint zero crossing. Each sample will be at the same voltage, i.e. there will be no tone. If you sample at double(+a little) frequency, you at least get each sample at opposite sides of zero, and you will get out a tone of the right frequency. If you sample 20.010 or 19.990kHz, at 20k samples/sec, you will get a 10Hz output.
As an EE I don't like this video very much. What is the picture at 3:15 supposed to show? There are no axis titles and somehow there time domain waveforms as well as constellations drawn on the same diagram? This neither does make sense, nor is it in any way correct. But the worst is, it confuses more than it explains :-(
Talking about PSK/QAM the axes (time and amplitude) are implied since I really can't see them being anything else when you are talking about using carrier phase angle to encode data.
Really interesting, I'd love to hear more about these thing :)
10 ปีที่แล้ว
I'm completely ignorant about this topic. I mostly got the IDEA, but I can't say I now understand what the modem sounds exactly are. Are they like calibration sounds? Why were they only hearable during the first part of the connection but then it would only make much dimmer "whiter" noises? I didn't really get that part, and I didn't fully understand how the divided frequency bands worked either. :/
it'd all be scrambled, it'd require a lot to decode any data (you'd not know what frequency slot each bit was going in). As for the analogue POTS, yes, you can often overhear on a line especially if there are earth faults or a lack of balance on the line
Well yeah i'd would be a werry bad signal but it would be interest to know how much data is theoretically preserved. And how obtainable that information would be. The banding of frequencys would probably help a lot. As for any important information it would probably be encrypted, im just curious an ee sort of way.
Intercepting facilities for law enforcement/etc. are usually required to be provided by the Telcos by law - tapping a DSL line would be a much harder way.. :) On the other hand - a technique called vectoring takes crosstalk and line interaction into account on VDSL2 lines to guarantee higher data rates..
When I was a young kid, we still had a dial up modem and whenever someone called us or made an out-going call, the internet would drop out. If you wanted to download anything over 10 megs or so, you were dreaming. Kinda puts my current "slow" 6mbps into perspective... though I still wish I could get FTTP.
My question is if both the phone line and USB2.0 use 2 conductors for data transfer, then why is ADSL so much slower than USB2's theoretical 480Mbps even if we don't have a phone? At least be 10-20Mbps or something decent.
usb 2.0 is less than 3m cable length. telephone lines to your exchange are usually 1km-6km in length and are subject to line noise and bad conditions. does this answer your question?
Well, maybe because DICSIS is a set of standards for data transfer over coaxial cable, especially cable tv distribution systems, not twisted pair phone lines.
DOCSIS 3.0 is standard in many countries, including Australia where I live, and we can achieve speeds of over 100mbit/s, and with splitters on coaxial cable you can run things such as Foxtel (cable TV) and many other things like having a phone line converter on your house all on the one coax cable. I'm guessing it's just less popular because it's a tad more expensive than twisted pair, and far more expensive than fiber, and as it is in Australia, it's cheaper just to take the leap from the old twisted pairs to direct fiber to the home. As for why most of the Australian grid is doing fiber to the cabinet like the UK, I honestly have no idea, but perhaps it just makes servicing the fiber terminations easier for technicians in a cabinet than having to gain access to a residential property, and in most cases they can use the existing twisted pairs/coax that's dug into the house.
Marcello Zucchi That's a great article from a data/programming viewpoint but from an electronic engineering viewpoint the baud rate at the exchange was determined by how may signal switches the pots switch gear could perform per second. Every time the there is a phase change then exchange gear sees a break in the signal and that is equal to 1 baud. It is pretty much the same process your switching hub does for every data packet that passes through it. For instance, my current switch at home can manage about 12 gigs-packets per second as it's data packet switching speed. Most POTS exchanges were limited to 9600 baud during the 80's and 90's and since the telecoms companies would not make any assurances about line quality they would generally not officially support more than 1 bit per baud. I could be a bit fuxxy about some of the details since it has been about 20 years since I studied all this but the concepts are sound. I suggest you read more about the communications switches that handle all the data and communications traffic across every network in the world.
Surma Sampo Baud is signal rate. It's used to specify how often the signal changes. The bit rate is usually higher then the baud rate. Remember that the signal on the phone line is analog and can have other values then just 0 and 1.
BersekViking Correct but what qualifies as a signal in the per second rate? It is a wave form cycle or phase cycle. Every time a wave form cycle, which in huger bit rate modems is a phase change, that is 1 baud. The number of bits in that baud is defined by the combination on the AM + FM + PM which basically defines an analog data packet size. Modems up to and including generally used just amplitude and frequency modulation on a full sine wave. Phase modulation allowed the modem to create more permutations per cycle (or sine wave) but changing which segment of the wave form was sent per cycle. Line quality greatly effected how granular the modulation control could be as noise and attenuation interfere with AM and FM. We bascially used really complex sounds wave fragments to carry data from one digital device to another. Turns out you can use the same principles with 2 laptops or smart-phones using the mics, speakers and dsp's to transmit data or commands between the machines. This even can operate outside out 20khz hearing range for hacking purposes when normal network interconnects are not available.
Regardless, early 300 baud modems were indeed 300 bits/sec, because they used only two symbols. It wasn't until the mid-80's that higher data rates became common.
great video but if you really want to impress me (and you know you do) do a computerphile on why one can pass a turing test 10 times out of 10 but fail a captcha 9 times out of 10 :-/
is not that your title show , : you put just an introduction of the signal type ... please can you said exactely the adsl-protocol comunication ? so translate to wan ?!
Fibre. My DSL is connected to my phone line which is a single cable stretching from my building ( containg 6 flats) to a telephone pole. This telephone pole looks like a maypole with a cable stretching to each building in our Close. If everybody went on line together the bandwidth would go down to 2Mb. 1 8th of its capacity. So if I am watching a video it lags or stops completely. I now pay more for 4G which always gives me 2MB per sec and when 5G comes to Europe I will be using that.
We only had to know modulation with 8 phases and then we had to manually draw an example byte on paper. With the knowledge that you not simply can use the same phase in sequence and I think Huffman correction code. Was a nasty thing I am no good at drawing with pencil. Although then I even though found it funny. Wish I had done music like Conservatorium or so, but I am no good at music.
I'm Jello, as I cannot get FTTC, I have no cabinet. :( Direct Exchange connection. So perfect line, but no ability to "wire in" the fibre, as no one is allowed to mess with the Exchange. :(
dr34m3r2k 99% of places are not allowed due to the risk of breaking equipment. Just the next street has it. I'm literally too close to get the upgrade. :(
dr34m3r2k No. Literally, there is no solution they provide in the UK. They are afraid to drill holes/change the exchange racks etc for direct lines into the exchange. They are worried the fibre would interfere with the other equipment. I can get a WHOLE NEW LINE INSTALLED. But as that includes a cabinet, a line, the digging, the planning permission, and a new socket in my wall, you can guess the price would be astronomical. All because my line has no cabinet for them to wire up as fibre, because it's too short a line. :P
I guess I'm not a computerphile. I took College Physics I and II back in my college days and I work in the biomedical sciences, but I got totally lost in all that information he explained after the different phases and the sinusoidal wave circle diagrams.
we seem to forget that docsis over coaxial cable is like a jillion times better. fiber is the winner but coax and seemingly the hardware involved is just way more crazy. docsis for life. but i dont live in a place with docsis :) so i use radio in the 5.6ghz band (licensed)
i remember the 300 baud, 1200, 2400. omg 14.4 and 28k were dreamy! well going CoAxial Cable is the way to go now and well i hope they start getting Fiber to Home in Canada soon.. plzz :).. big 3 telecomms are "fighting" for $$ n control (rogers, bell, telus)..
ADSL? How quaint. Wait, people still use that? Oh yeah in rural Africa right? What, in "developed" countries? Surely not? ;) Always nice to hear about antique technology like this =P "No one will ever need more than 9600 baud because you can't read faster than that!"
Worst keywordspamming ever. He only talks modem technology and doesnt even mention PPPoE or anything DSL-specific. And there is no infographics either.
I think this video in particular defeats the purpose of YT "casual" educational channels, it's too technical, with no clear simple explanation of the ideas.
Too bad they're going to artificially limit all that stuff soon. All that hard work to cheat like a boss! With luck, we either find a way around, or it never happens to begin with.
The one and only true explanation of how a modem really works.
As an amateur radio operator this was the easiest explanation I could've ever asked for.
Yeah I agree too! An radio amateur and some knowledge of computers can understand pretty much everything! Nice video!
ok?
The phone system is magic. The ingenious crazyness that (before DSL and ISDN) takes a digital computer, a digital phone system and joins the two together using some analog technology that tries to mash digital data down into frequences used by human voices.
The truly clever part is that my broadband phone line still supports an old GPO rotary dial phone from the 80s I have that contains a proper mechanical bell and pulse dialling done through a mechanical dial.
I would turn that phone over and look at its production date. I bet it is older than the 80's. I have one from '62.
It's pretty cool that, with everyone happy to replace our current physical wiring with fiber, a smart engineer can still squeeze more bandwidth out of copper that was designed to have a fraction of what is has now. I think it's a testament to the fact that we can always improve what we have, even if we are certain it's at its peak.
I absolutely LOVE these technology history lessons explaining how we got to where we are today.
4:03 Well, the Nyquist theorem doesn't state that you _should_ sample at twice the maximum frequency in your signal, it states that you _have to_ sample at twice that frequency or you will not only not get those higher frequencies in your discretized signal, you will in fact get false frequencies instead (mirrored at the Nyquist frequency).
I love the Brits for their polite language but sometimes you really can't replace "must" by "should".
That was really interesting! I'd love to see a series on signal theory!
Electrical Engineer -Phile. This is all stuff I learned in class.
I usually explain it as how a radio can pick up one frequency out of dozens of radio stations.
Now imagine all the stations working together to transmit the same piece of music, not 5.1 surround sound channels or even 7.1 but 76.5 channels. The receiver tunes *all* stations at once and splits the sound out to different speakers.
Thus you use the whole length of the radio dial to transmit one piece of music, so DSL uses a *broadband* of frequencies (cut up into bins or cells) to transmit data.
My DSL cable whose only job is to communicate to the splitter and has just two copper wires is broken.
I think this is the first Computerphile video where I understood jack shit.
A lot of 'computer people' get brainfarts when thrown into analogue signal processing. I have issues with it, but can follow the modulation stuff okay.
It wasn't the best was it? I have some understanding of how this stuff works, and he really didn't explain the basic or advanced concepts as well as some other presenters have with their content.
Kabitu1 Lol, I got to the bit were they played the fax sound and I was like 'Hey, I recognise that' and the rest of the time I was kind of fooling myself into believing that I still remember how AM/FM works.
Kabitu1 I'm not in computers at all, and this one I understood more than most of the other computerphile vids because I'm super interested in audio. I think audio knowledge would help here more than computer knowledge. When he started talking about phase shifting, I was like "man, nobody who doesn't get sound and audio is going to understand this."
Well, if you've studied the right parts of maths, you'd get it too
Nice explanation of QAM, its used in a lot of applications nowadays so good to have an appreciation of how these things work at a low level.
80mbps over FTTC (Fibre to the cabinet)! You're pretty fortunate as most of the residents around here struggle to even get 20 over FTTC due to poor quality aluminium cabling and long distances. Vectoring looks like it might be able to increase speeds but, at the end of the day, FTTC, FTTdp (fibre to the distribution point - mini cabinets closer to people's houses) and FTTrN (Fibre to the Remote node, pretty much the same as FTTdp) are just dragging on the life of the copper network, which should be being replaced with pure fibre!
The copper network also supplies phones that work during a local power failure, fibre would require a local power supply and need to be battery backed to supply phone service. And as I found during a large area power cut the other week, the mobile phone system simply drops dead during a power failure, apparently there's no $()(!@#$) backup batteries at the cell site
@@TheChipmunk2008 And now nearly no one uses analog lines, most people have cellphones and buissnes have voip or cloud telephony for their deskphones.
Yikes! Aluminum as an electrical conductor? I mean sure it's not the worst conductor in the world, but the only reason I can see a company making the decision to use aluminum as an electrical conductor that is meant to carry many frequencies with little resistance is because they're just too cheap to spend the extra dollars on copper.
This must just be a UK thing. It's not uncommon in the US to see downlink speeds in excess of 100Mbps. I pull 300Mbps from my ISP over copper wiring, and the only thing preventing me from going faster is the cost of the faster service. You could easily pull close to 1000Mbps over a well shielded copper wire.
@@TheChipmunk2008 Must just be your mobile phone provider. Power was knocked out over a large region of the state that I live in and despite my entire town not having power for close to 24 hours (including street lights and signals), my cell phone still worked just fine. Internet, voice, and all.
@@gfifer1 it sounds like you are talking about coaxial cables, there is no shielding in twisted pair phone cables, even those made of copper. Those speeds are almost impossible to obtain for 90% of people with copper lines thanks to attenuation and crosstalk (and overselling).
Thankfully fiber is becoming more and more easily available.
This would have been immensely helpful for my computer networks exam last year. I can't wait until you do videos for the higher layers of the OSI layer.
I was looking for such a video since forever, thank you!
I must say ADSL is one of the greatest inventions since the first home modems. Certainly a big deal for someone who grew up in rural New Brunswick! Finally! Decently fast internet without a separate line!
How does the dsl 'modem' know what the reference (Zero degree) phase is? for the PSK? Does it have a reference oscillator that's synced up occasionally, or does it measure relative to the previous 'bit' and could thus get out of sync?
Marcello Zucchi Thank you, always wondered since QPSK was used in the 80s for digital TV sound (NICAM) :)
How does the Bee Orchid that attracts Bees to come and mate with it and thus dupes said Bee into carrying its pollen know or came to know what a Bee looked like in the first place.
Thanks for explaining QAM so well!
Really really good explanation to the issue, but it feels like the video is heavily edited to fit as much information into the 09:35.
Personally I am a ham radio operator so I have fair bit of knowledge with analogue radio, but I still have to concentrate a lot to understand how QAM modulate, which I didn't before.
I'd suggest breaking it down further to explain what modulation / carrier signal is, then going into how PSK differs from AM/FM, then how PSK is combined with FM to form QAM.
9:04 Is that a new trend, to have a slightly opaque overlay along with subtitles? The CERNPeople channel does it as well. I can't tell if someone is overwhelmed with their editing tool, or if it's a deliberate mistake, like dubsteb.
Holy crap, that looks like Jubilee Campus to me. Hope you guys weren't too badly affected by the fire last night.
what a great watch! i have always learned more about a concept or technology by going back to its origins, thanks heaps!
SO AM, FM and PM are all linked to radios? To achieve weird sounds ( Massive, Operator and Gladiator) we use Fm or PM which is fairly new!
At about 6:40 he uses the phrase "tap it out". I hear that as "put the signal through". Right? Like "tap it out over this line". Is that what that meant? And if so, that's got to be a holdover phrase from the days of telegraphy. Like, tapping out a telegraph over the telegraph lines. Am I way off here??
+floofytown It means we tap into the line before or after the filters.
Ala 'phone tap'
Wait, that 1983 War Games movie... was that little scrub using a modem? and was he connected to the internet.. in 1983??? where did he get the modem from?
Brady are you using a new camera? or shooing raw? because the dynamic range of that shot is awesome
Great informational video! Guy speaks very clearly too! Loved it....
the amplitude and frequency modulations are easy to understand, but I completely missed the phasing modulation. can anyone clarify that for me? :)
/watch?v=QFi16s4RXXY
Signals are waves, so they have phases. You change the phase to encode data.
Ok, I'll try.
If you consider a standard Sine wave, it starts at 0 goes up to 1, comes back down through 0 to -1 and then heads back up to start over again. It does this again and again at a certain frequency. Let's say that that standard Sine wave is 0 degrees out of phase. In other words, it's not out of phase at all.
If we were to look at a Sine wave that is 180 degrees out of phase, it would be exactly opposite. We say it's 180 degrees out of phase because you can model the Sine function on a circle. So our first Sine wave started on the right-most point of the circle and went counter-clockwise. If we start on the left-most point of the circle, it's 180 degrees around the circle at that point.
Ok, so a Sine wave 180 degrees out of phase is the exact opposite of one at 0 degrees. This means it starts at 0, goes down to -1, comes back up through 0 to 1, and then heads back down to start all over again. You could then build a machine to detect which phase you were in, since all of the peaks are now valleys and vice versa.
I'm not entirely sure how it works but I could think of one way to detect the phase of the a signal:
Since you know the amplitude and frequency of the signal you're supposed to be detecting, you could build a machine that takes that signal as well as generates it's own signal at the same amplitude and frequency and combines them. If the signal coming in is at 0 degrees, the peaks and valleys match up and they add together. You're left with a signal that has twice the amplitude. If the signal coming in is at 180 degrees the peaks and valleys are opposite each other and therefore cancel out and the resulting signal is completely flat. The detection machine then looks if there is a signal there at all. If there is, it sends out a 1, if there is not, it sends out a 0.
APaleDot so you don't modulate a wave, you start another wave with it's phase offset from the previous one. it that right?
Without getting too technical, I am going to give it a try.
AM: Changing the Amplitude of the carrier wave
FM: Changing the Frequency of the carrier wave
PM: Changing the Phase of the carrier wave
AM and FM are fairly easy to understand, I'm going to skip that.
PM (or PSK) is basically changing the "starting point" of one signal.
Imaging a clean AM signal, it would start from 0, going to +1, 0, -1 then back to 0. This completes the cycle.
For PSK, let's assume 0, +1, 0, -1, 0 signal is the "starting position" of this signal.
The "opposite" of this signal would be 0, -1, 0, +1, 0.
The transmitter can choose to start the next symbol by going from 0 to +1, or from 0 to -1. This represent 2 values, 0 and 1.
This change in ramp direction is a change in phase.
In real life, PSK used are often of more than just 2 phases, in order to make it "faster".
In the latest WiFi technology, 256QAM is used, 256 possibilities (or bits) could be squeezed into just 1 symbol transmitted.
The trade off of using higher rate QAM is losing resilient to interference, therefore most digital devices would auto negotiate and slow down if required.
Like feeding my brain a steak dinner! I enjoyed that very much. Quite informative even if it was a little hard to follow at times.
Would love to see one of these about VDSL, VDSL2, and VDSL2+.
The "Fiber To The Cabinet" he mentioned at the end uses VDSL
Can you put subtitles on your videos? Thanks.
that would help me
At 0:56: "Phase modulation hasn't been particularly used in the analog space". I beg to differ. For decades analog Police, Amateur, commercial and consumer 2 way radio all used PM. FM receivers can pick up PM because shifting the phase shifts the frequency (and vice versa).
Well, this video don´t really explain how ADSL works, rather digital encoding over analog lines work. Also, ADSL is mostly not usesed any more, XDSL, IDSL and other later was renamed ADSL2, ADSL2+ and so on.
Firstly ISDN broke throw the 56kbit/s limit a long time before it even was reached. Already in the early 90-tys ISDN use 2 carrier channels of 64bit/s just using analog filtering and basic bit inversion keying. So when the modem was pushing 28kbit/s, ISDN already reached 128kbit/s. The problem was that it have to use a ISDN to telecom network converter in the other side. And then in the telecom server park a bypass conection to internet.
What was done with ADSL was that a special server was introduced that converted ISDN streams directly to internet traffic before the line reached the telecom switches. In this case there was no more any limitation of only 2 channels per telefonwire. Then they could basically use a unlimited number. Later ISDN B-chanel introduces a additional 4 channels limiting the maximum to 384kbit/s... This is also happening to be the maximum limit of the 2G GSM data channels (128kb/s för GPRS and 384kbit/s for edge, this is due to design).
With ADSL they basically keeped the old telecom system on the server side, but swiched out the old 6 times 64kHz side to a 256 times 4kHz, increasing the bandwith needed by 3 times, but the number of channels by 40 times. Every channel still carry 64kbit/s, but using a combination of PSK and ASK (supposedly 8 bit PSK and 7 bit ASK? total of 15bit?), total of 15Mbit/s of witch 8 is usable....1998 specifiation
Later version 2001 use less overhead increasing the capacity to 12Mbit/s with over 500 chanels also using more robust coding, later versions (2008) increased capacity to 24Mbit/s decreasing overhead and others stuff.
VDSL and G.fast does not use ADSL backbone. It uses simular carrier system with 4kHz wide carriers, but they are shifted to a higher frequencies, using up to 4000 carriers but more robust coding, limiting the bandwidth to a maximum of 100Mbit/s even tow the number of carriers is a lot higher. The road side box in this case is connected directly to the internet, not routed to the telecom provider. This enable just about any company to install it. (in theory)
Also VDSL and ADSL is designed so it could run simulations on the same cable
Thank you
What is the rationale behind the Nyquist Theorem? Is doubling the sampling rate necessary to properly replicate a sound digitally or is it done to provide an ample amount of headroom or buffer zone? For instance, since the human ear can only perceive up to 20 khz, would a digital sampling of 20 khz be virtually indistinguishable from 44 khz or is that doubled rate necessary to overcome audible anomalies introduced by the digitization process?
If you sample a 20kHz tone at 20000 samples/second, each sample will be at the same point at the graph, for instance the point of the positive-goint zero crossing. Each sample will be at the same voltage, i.e. there will be no tone. If you sample at double(+a little) frequency, you at least get each sample at opposite sides of zero, and you will get out a tone of the right frequency.
If you sample 20.010 or 19.990kHz, at 20k samples/sec, you will get a 10Hz output.
i have a one ADSL connetion but i have two modems . sir can i conected both modem at one connetion with two diffrent placees at same time
As an EE I don't like this video very much. What is the picture at 3:15 supposed to show? There are no axis titles and somehow there time domain waveforms as well as constellations drawn on the same diagram? This neither does make sense, nor is it in any way correct. But the worst is, it confuses more than it explains :-(
Talking about PSK/QAM the axes (time and amplitude) are implied since I really can't see them being anything else when you are talking about using carrier phase angle to encode data.
In constellation diagrams the axes are Q and I parts of the signal, not time and amplitude.
Could anyone tell me the type of "rule" the paper he is using it called.?
Loved the video!
Can you tell us more about how frequency modulation works?
I like your video very much. It's really great. I'll keep an eye on your channel. I am your fan and I will support you.
The telephone system went all digital in the 80's, not the 60's
I love the graphics again.
In Australia its fibre yo the node?
so witchcraft then..........
mightyfinejonboy hahaha! best comment ever
I like that. 56K to (many years later) to 2000K. And here comes 5G. We all know the rudimentary incantations. Witchcraft. .
Really interesting, I'd love to hear more about these thing :)
I'm completely ignorant about this topic. I mostly got the IDEA, but I can't say I now understand what the modem sounds exactly are. Are they like calibration sounds? Why were they only hearable during the first part of the connection but then it would only make much dimmer "whiter" noises? I didn't really get that part, and I didn't fully understand how the divided frequency bands worked either. :/
***** Thanks! :D
It would be interesting to know how the asynchronous part of ADSL works and why it is commonly so
Did he say he gets 80Mbps with his fibre to the cabinet service? (right at the end of the video)
can we have an emergency update video on the fire please.
Does this mean that someone potentially could intercept the crosstalk from other lines in other frequency bands from other subscribers?
it'd all be scrambled, it'd require a lot to decode any data (you'd not know what frequency slot each bit was going in). As for the analogue POTS, yes, you can often overhear on a line especially if there are earth faults or a lack of balance on the line
Well yeah i'd would be a werry bad signal but it would be interest to know how much data is theoretically preserved. And how obtainable that information would be. The banding of frequencys would probably help a lot. As for any important information it would probably be encrypted, im just curious an ee sort of way.
Intercepting facilities for law enforcement/etc. are usually required to be provided by the Telcos by law - tapping a DSL line would be a much harder way.. :)
On the other hand - a technique called vectoring takes crosstalk and line interaction into account on VDSL2 lines to guarantee higher data rates..
I struggle seeing how the receiver could differentiate between a shifting frequency, and a shifting phase.
My thought exactly!
When I was a young kid, we still had a dial up modem and whenever someone called us or made an out-going call, the internet would drop out. If you wanted to download anything over 10 megs or so, you were dreaming. Kinda puts my current "slow" 6mbps into perspective... though I still wish I could get FTTP.
I don't like how you close tag, but not title tag, my OCD kicks in
So how does fibre optic broadband work???
My question is if both the phone line and USB2.0 use 2 conductors for data transfer, then why is ADSL so much slower than USB2's theoretical 480Mbps even if we don't have a phone? At least be 10-20Mbps or something decent.
usb 2.0 is less than 3m cable length. telephone lines to your exchange are usually 1km-6km in length and are subject to line noise and bad conditions. does this answer your question?
Great video! Sadly my ADSL only goes up to 1.8MBps, not 80 MBps.
Ah well, good enough for Computerphile videos!
(at 360p...)
Awesome video.
so what about DOCSIS? all the numbers say that it is better so why isn't BT using it?
Well, maybe because DICSIS is a set of standards for data transfer over coaxial cable, especially cable tv distribution systems, not twisted pair phone lines.
DOCSIS 3.0 is standard in many countries, including Australia where I live, and we can achieve speeds of over 100mbit/s, and with splitters on coaxial cable you can run things such as Foxtel (cable TV) and many other things like having a phone line converter on your house all on the one coax cable. I'm guessing it's just less popular because it's a tad more expensive than twisted pair, and far more expensive than fiber, and as it is in Australia, it's cheaper just to take the leap from the old twisted pairs to direct fiber to the home.
As for why most of the Australian grid is doing fiber to the cabinet like the UK, I honestly have no idea, but perhaps it just makes servicing the fiber terminations easier for technicians in a cabinet than having to gain access to a residential property, and in most cases they can use the existing twisted pairs/coax that's dug into the house.
My First Modem - 300 baud.
300 bits per second... Woohoo!
Baud is switching speed not a data rate.
Marcello Zucchi That's a great article from a data/programming viewpoint but from an electronic engineering viewpoint the baud rate at the exchange was determined by how may signal switches the pots switch gear could perform per second. Every time the there is a phase change then exchange gear sees a break in the signal and that is equal to 1 baud.
It is pretty much the same process your switching hub does for every data packet that passes through it. For instance, my current switch at home can manage about 12 gigs-packets per second as it's data packet switching speed. Most POTS exchanges were limited to 9600 baud during the 80's and 90's and since the telecoms companies would not make any assurances about line quality they would generally not officially support more than 1 bit per baud.
I could be a bit fuxxy about some of the details since it has been about 20 years since I studied all this but the concepts are sound.
I suggest you read more about the communications switches that handle all the data and communications traffic across every network in the world.
Surma Sampo Baud is signal rate. It's used to specify how often the signal changes. The bit rate is usually higher then the baud rate. Remember that the signal on the phone line is analog and can have other values then just 0 and 1.
BersekViking Correct but what qualifies as a signal in the per second rate? It is a wave form cycle or phase cycle. Every time a wave form cycle, which in huger bit rate modems is a phase change, that is 1 baud. The number of bits in that baud is defined by the combination on the AM + FM + PM which basically defines an analog data packet size.
Modems up to and including generally used just amplitude and frequency modulation on a full sine wave. Phase modulation allowed the modem to create more permutations per cycle (or sine wave) but changing which segment of the wave form was sent per cycle.
Line quality greatly effected how granular the modulation control could be as noise and attenuation interfere with AM and FM. We bascially used really complex sounds wave fragments to carry data from one digital device to another.
Turns out you can use the same principles with 2 laptops or smart-phones using the mics, speakers and dsp's to transmit data or commands between the machines. This even can operate outside out 20khz hearing range for hacking purposes when normal network interconnects are not available.
Regardless, early 300 baud modems were indeed 300 bits/sec, because they used only two symbols. It wasn't until the mid-80's that higher data rates became common.
great video but if you really want to impress me (and you know you do) do a computerphile on why one can pass a turing test 10 times out of 10 but fail a captcha 9 times out of 10 :-/
pheers modulation?
is not that your title show , : you put just an introduction of the signal type ...
please can you said exactely the adsl-protocol comunication ? so translate to wan ?!
So this is where Trigonometry class becomes useful.
Technically the wargames modem was an acoustic coupler: en.wikipedia.org/wiki/Acoustic_coupler
Fibre. My DSL is connected to my phone line which is a single cable stretching from my building ( containg 6 flats) to a telephone pole. This telephone pole looks like a maypole with a cable stretching to each building in our Close. If everybody went on line together the bandwidth would go down to 2Mb. 1 8th of its capacity. So if I am watching a video it lags or stops completely. I now pay more for 4G which always gives me 2MB per sec and when 5G comes to Europe I will be using that.
Who is the man who is speaking?
So what about vectoring?
Can you make a video explaining how OpenGL works?
We only had to know modulation with 8 phases and then we had to manually draw an example byte on paper. With the knowledge that you not simply can use the same phase in sequence and I think Huffman correction code. Was a nasty thing I am no good at drawing with pencil. Although then I even though found it funny. Wish I had done music like Conservatorium or so, but I am no good at music.
"That's crazy man... you ever done discrete multitone?"
***** Can you remake this video? His explanation wasn't very clear and if you can add more images and graphics that would be helpful too.
Not even a mention of Fourier in this? D: All my DSP friends would be saddened.
We must fix the IPv4 problem though.
Well we’ve got IPv6 on it’s way
@@samuelhulme8347 Nice reply about six years later
I'm Jello, as I cannot get FTTC, I have no cabinet. :( Direct Exchange connection. So perfect line, but no ability to "wire in" the fibre, as no one is allowed to mess with the Exchange. :(
Not true. FTTC for exchange-only lines is a reality.
dr34m3r2k
99% of places are not allowed due to the risk of breaking equipment. Just the next street has it. I'm literally too close to get the upgrade. :(
Breaking equipment? lol. It is not installed by amateurs. Find your local BDUK project if you're not covered by BT's national roll-out.
***** EO lines can be converted by placing the PCP and DSLAM outside of the exchange.
dr34m3r2k
No. Literally, there is no solution they provide in the UK. They are afraid to drill holes/change the exchange racks etc for direct lines into the exchange. They are worried the fibre would interfere with the other equipment.
I can get a WHOLE NEW LINE INSTALLED. But as that includes a cabinet, a line, the digging, the planning permission, and a new socket in my wall, you can guess the price would be astronomical. All because my line has no cabinet for them to wire up as fibre, because it's too short a line. :P
I guess I'm not a computerphile. I took College Physics I and II back in my college days and I work in the biomedical sciences, but I got totally lost in all that information he explained after the different phases and the sinusoidal wave circle diagrams.
yay 90's kids!
Oh dear, I'm afraid you lost me towards the end. Still very interesting, though; I may want to watch it again when I'm a bit more focused.
we seem to forget that docsis over coaxial cable is like a jillion times better. fiber is the winner but coax and seemingly the hardware involved is just way more crazy. docsis for life. but i dont live in a place with docsis :) so i use radio in the 5.6ghz band (licensed)
oh i just realized there is in fact cable docsis near me but its 5mbit so i pay triple the money for 12mbit radio
fiber to the prem or bust!
Wireless telecommunication is way more complex. CDMA and OFDM are mind blowing.
And today, I'm getting over 50 mbits on ADSL.
And 6 years later, I'm getting under 25 mbits on VDSL
Wouldn't it be nice to have 4 megabits per second download on ADSL2+ when the best I have ever seen was 3.1 and it's usually 2.6 or less.
I didn't get it. ._.
i always wondered why they don't just put fiber strait into your house instead of cable or twisted pair. i guess its way too expensive.
By and large because the twisted pair is already there.
Thank you!
80 mbps? Never move to the US, the internet speed here will probably kill you, with our superior 10 mbps service that blacks out at 8pm every Tuesday.
I'm glad with my low-end connection of 50Mbps.... The high end one is 300Mbps.
if i could understand this explenation I would not even be here.
I still have no idea what any of this means....
oh....so thats how im watching this video................
So... why can't you just move onto ethernet? In my country, we moved to optic fibre already. I mean, we have optic fibre going into the apartments.
i remember the 300 baud, 1200, 2400. omg 14.4 and 28k were dreamy! well going CoAxial Cable is the way to go now and well i hope they start getting Fiber to Home in Canada soon.. plzz :).. big 3 telecomms are "fighting" for $$ n control (rogers, bell, telus)..
ADSL? How quaint. Wait, people still use that? Oh yeah in rural Africa right? What, in "developed" countries? Surely not? ;)
Always nice to hear about antique technology like this =P
"No one will ever need more than 9600 baud because you can't read faster than that!"
Worst keywordspamming ever. He only talks modem technology and doesnt even mention PPPoE or anything DSL-specific. And there is no infographics either.
I think this video in particular defeats the purpose of YT "casual" educational channels, it's too technical, with no clear simple explanation of the ideas.
I am sorry to say, but as someone who has no background in signal processing, I got nothing out of this video. Not even the intuition.
Its over now
Too bad they're going to artificially limit all that stuff soon. All that hard work to cheat like a boss!
With luck, we either find a way around, or it never happens to begin with.
So complicated.
Who knew
First
6 seconds late >-
9:17
I have like 1bph internet it seems.
Alex Madison how so?
ntwede wait, you or I was late?
eeeee hoooo
haha
dmt
lol I agree adsl is scary complicated
Signal detection is such a complex field that it overlaps direcly with statistics.
Given a data set what are the salient figures?