Why 56Kbps for dial-up connections?

I’ve never seen any dial-up modem faster than 56Kbps. Is this the limit for dial-up modems?

I thought that 64kbps was the limit but I can’t for the life of me remember why.

I remember byting a modem in 1990 and it was 33.6 because 56k hadn’t been invented yet.

In 1998 I had ISDN and then ADSL in '99, so maybe it’d possible to go faster but it’s not being researched?


In reality the max speed for dial-up modems is less than 56K. Sure the box says 56k and technically the modem is capable of that but it is restricted by the FCC to ~52k speeds (the FCC does not want more power than that required to push 52k along on the lines…technical reasons I do not claim to understand). The reason they do not market, say, 128k modems (not counting dual link modems if they are even still around) is that copper wire cannot manage better than 56k. Again a technical limitation and saying your modem could do some crazy high number in theory starts to push truth in advertising I would think. If the FCC removed regulations 56k is the best a modem will do on copper wire.

With compression and other tricks you may get a rate of over 52k but the actual data speed is still 52k or less.

ISDN used multiple channels to get its higher speed and is digital.

DSL is also digital and works with the added bandwidth unused by analog telephones.

Modems are restricted because they are an analog device.


I think 56k cheats by being a half digital/analog conection.

Digital was the key. Single channel ISDN was still 64K data rate. I remember my first broadband fondly!

Strictly speaking, the limit for an analog modem is 33.6–that one’s due to Claude Shannon. How 56k modems exceed that limit is very interesting. You can read about it here.

I’ll explain it as I do to my sales reps: 56K is all the data that can be transmitted at analog voice frequencies over a copper line.

Actually, it’s really 64K. Where’d that extra 8K go, you ask? The 8K is used for signaling. That’s where your phone (actually the local switch which generates your dial tone) tells the receiving switch, “I’m phone number 212-555-1234, and I’m calling 404-555-1234.”

By using higher frequencies, ISDN can move the signaling out of the channel you’re using (inline signaling) to a “D-channel”.

I could go into more (and more accurate) detail, but that would involve pulling out my old text books, and I’m already having to down coffee to stay awake this afternoon.

ADSL has been around for about 20 years.

There is a way to go faster on copper (that’s the important phrase here), VDSL.

Text book time: Crap - Someone “borrowed” them. Sorry.

Going from memory: VDSL has bandwidth of about 100M, but the signal could only travel 1000 feet on 24 gauge copper wire before the resistance in the lines degraded the signal. Perhaps one day, after they invent room temperature superconductors…

56K bits per second is the speed settled on in the USA as the minimum data rate required for toll quality speach. In Europe, 64K bps was the widely agreed requirement. In both cases, only a very simple form of compression is used, called companding. One can get toll quality speach with much lower data rates by using more sophisticated compression as used for mobile phones.

Telephone companies were (and are) under great pressure to get as much revenue from their infrastructure as possible and to this end phone calls would be digitised at some point and transmitted on high capacity data links along with a whole bunch of other calls.

When ISDN came along - I don’t have a date for you, but the specification is quite old - it was designed to be as compatible as possible with the existing telco infrastructure i.e. channels of 56Kbps in the USA and 64Kbps elsewhere.

Where do modems come in to this? A 56Kbps modem is working at the limits of a US telephone connection that is digitised at some point. Obviously it is just impossible to get more data down a line if the telephone company is going to digitise the signal and send it on at a mear 56Kbps.

Also, the protocol for the 56K modem was specifically designed to get the very best out of the by then well standarised method employed to digitise an analogue phone call. It is in this sense that Cat is right to suspect that a 56K modem cheats.

In fact, there was a brief time when it was possible to by analogue modems faster than 56Kbps, there were intended for use over short private wires and could, for example, be employed to join computer networks in separate buildings.

This is all very interesting and its amazing that it’s not dawned on me before! Am I summarizing this right?

The 56kbps limit is a result of the telco digital circuits’ 64kbps (less handshaking and protocol management) single circuit digital limit on the back end of things rather than any appreciable limit to the data carried in the audible spectrum via the modulated digital data?

I’d always thought that the problem was that you’re only counting on the 300hz to 3khz audible bandwidth of the POTS line, and that there was just no way to audibly modulate an uncompressed signal more than 56kbps given that bandwidth.

So, what’s the maximum pure digital data rate given a 300 to 3,000 hz band (a 2,700hz bandwidth)? Assuming no noise and perfect discrimination, something like !2700 bits per second? That’s a bigger number than Excel wants to give me :slight_smile:

Basically: Yes.

Woo-hoo! I was actually helpful in a GQ thread!

I don’t know where you got your information, but in both the US and Europe, phone calles are digitized with 8-bit resolution, 8000 times per second, resulting in a 64000 bit/second data rate for a single line.

They’re aggregated differently - DS1 (T1) in the US is 24 of these, for 1.544 Mb/s including some framing bits, and E1 in the rest of the world is 32 calls, for 2.048 Mb/s, but the individual calls are all 64 kb/s.

I suggest you read ultrafilter’s link for an explanation of where 56 kb/s came from.