Wired telephone bandwidth

It's roughly between 300 to 3400 hz, thats standard for either the old FDM or digital transport 4 khz channel systems for everything except trans-ocianic lines which, used to be at any rate, limited to 3000 hz to fit more calls onto the cables/satelite links. With the prevalence of ATM these days they've probably probably gone up to 4 too.

All bets are off for IP telephony.

H.

Reply to
Howard Eisenhauer
Loading thread data ...

Wrong, AlwaysWrong. When v.90 modems first came out I was regularly doing

48-52K on one of my POTS lines (the other was stuck at 28.8 because it went five miles, all the way back to the CO).

Wrong again, Dimmie. One pipe, one stream, PCM encoded.

Wrong, Nymbecile. Nothing is *ever* noise-free.

A lie. You're lying a lot more these days, DimBulb.

AlwaysWrong is once again, wrong.

Wrong again, AlwaysWrong. It's limited by the integrity between you and the first DAC and the NUMBER of DACs (

Reply to
krw

PCM, I think. QAM is used on the, slower, uplink.

Doesn't he always.

...and that snotty attitude has lost Ma' bundles of $$.

Reply to
krw

You're an idiot. It is 100% pure ISDN from the first link forward for POTS today.

Reply to
MakeNoAttemptToAdjustYourSet

"MakeNoAttemptToAdjustYourSet" wrote in message news: snipped-for-privacy@4ax.com...

I used to get ~4.5kB/s downloads with the modem reporting connecting speeds to

46-52kbps similar to what Keith reported. It wasn't twice the speed of 28.8kbps modems, but *very* noticeably faster.
Reply to
Joel Koltner

AlwaysWrong does it again.

Reply to
krw

At least in Europe, the "56k" only worked in the downlink direction, i.e. the ISP had a typically a E1 (2.048 Mbit/s) connection for 30 subscribers and a bit by bit copy at 64 kbit/s 8000 samples/s was delivered to the DAC in the telephone exchange of the subscriber.

Theoretically, the 8 bits/sample can represent 256 analog values, since the floating point A-law representations use different step sizes depending of the "amplitude", the smallest steps are about the same size as a the LSB of a 12 bit DAC.

The cabling from the local exchange to the subscriber modem will distort the waveform to the modem and the equalizer in the modem tries to compensate for the distorted waveform and tries to determine which actual analog voltage was generated by the DAC. In practice, about

50-100 discrete voltage levels can be reliably detected, thus about 5.5-6.5 bits/sample can be transferred, corresponding to 44-52 kbit/s.

Of course, any noise in the subscriber cable (such as crosstalk from ADSL connections in other pairs) will degrade the SNR and hence reduce the throughput.

In the US, the T1 connection and u-law compression may have different constraints. Are they still using in-band signaling ? This at least caused a lot of problems in early ISDN and pre-ISDN connections across the Pond.

If the telco would deliberate want to force the use of 28k even on downlink, then they either would have to add some LSB noise to the samples, but only when the sample "amplitude" is low, which would appear as a slightly added noise in a voice contact. Alternatively, some analog noise would have to be added after each DAC in each telephone exchange.

Reply to
Paul Keinanen

Most definitely NOT Qwest, which cares only about money: (1) make a call, and after about 5 rings you get a recording asking for $1 "for later delivery"; (2) try *57 on a harassing call 2 or more times ($1 each) then call their so-called caller id bureau(or whatever they call it) and find that they have NO RECORD of the *57 and refuse to do anything; (3) their ads included with your bill promise extremely fast DSL and 985 of "customers" are lucky to get half that rate, other ads promise a very low rate ($12/mo) and say zilch concerning gotchas, addons, etc. This is a SHORT list...

Reply to
Robert Baer

second,

is,

Check to see if there is wiring from phonejack outlets or places for them. Then the problem would be at the entry box where incoming lines (which i guarantee still physically exist) are nominally connected to each service line pair. If you can look there, perhaps you can see the disconnect; put a short on YOUR pair and use ohmmeter inside to verify; remove short and call telco for service. Lettuce (or tomato) know results.

Reply to
Robert Baer

  • Sorry; they tested the line and said noise free, so noise is not the limiting factor.
  • Not the point; they supported 48+Kbps for ages (20+ YEARS) and all of a sudden crapped it down to 28.8K ON PURPOSE to get dial-up users to get pissed off enough to $pend more $$$$$$$$ for their DSL (which BTW is LESS reliable).
Reply to
Robert Baer

second,

is,

Telcos have not been doing party lines since the 50's. And if i was rich, i would not pay thru the nose for Blue Streak (2 COPPER pairs DIRECT to CO); i would put an antenna on a 100 foot pole for WiNet.

Reply to
Robert Baer

Wish Qwest would go back to that...

Reply to
Robert Baer

Hmm..Not filtering - just added low-level noise that their analog line tester would report a "good" line?

Reply to
Robert Baer

They are doing the US equivalent of DACSing to share your line because you are an unprofitable cheapskate is standard practice.

56kFlex modems were only around since late 90's and only really stable and interoperable with ISP kit after the V90 standardisation ~1998.

RTFM and force your modem to do a V34+ 33k6 connection in old world technology and you should be able to make the best of a bad job. Or switch to ISDN then they have to give you real copper or better for that.

ADSL is rock solid in most of the civilised world.

Regards, Martin Brown

Reply to
Martin Brown

If you can use an ISDN connection at each end the bandwidth is often dc to just under 4kHz. I have successfully transmitted dc-coupled audio across Europe and between UK and USA over the public telephone network using an ISDN-equipped computer at each end of the connection using the standard A-law/u-law coding at 8 kHz sampling rate.

In case you were wondering why I bothered to do this, it was related to debugging problems with an IVR system where DC level changes (caused by silences between prompts being encoded using u-law when the rest of the system was using A-law) were causing mobile phone network echo-cancellers to misbehave.

John

Reply to
John Walliker

Robert Baer wrote in news:MY2dnS5_esNSRA3WnZ2dnUVZ snipped-for-privacy@posted.localnet:

my downloads thru Opera 9.64 are usually around 5.3 Kbps. at first read,it goes as high as 12Kbps,but rapidly decays to ~5K. I figure it's the browser overhead that makes the difference.

--
Jim Yanik
jyanik
 Click to see the full signature
Reply to
Jim Yanik

Not quite. The initial load is of the main HTML text which is typically extremely compressible (more so than ordinary text) and so the initial characters per second peaks at ~12kb/s then the images start to load and they are almost always uncompressible JPEG or GIFs so the download then proceeds at the basic rate of the connection for the bulky binaries.

Regards, Martin Brown

Reply to
Martin Brown

can

that=20

amplitude

2400 symbols per second and about 12 bits per symbol maximum. Trellis = coded=20 QAM both directions. Down channel was typically limited to 50 kbps and = up=20 channel was typically around 10 kbps to reduce up/down interference. = Requires=20 real fancy adaptive equalization.

Don't know if such is done in my area i use DSL. If it is done they = manage it=20 in the digital domain.

And i remember the "voice quality" wars from one to two decades ago.

Reply to
JosephKK

I have no idea what are you babbling about, however V.34 provides symbol rates of up to 3.429k, with 960-point constellation.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

formatting link

Reply to
Vladimir Vassilevsky

You've probably hit the nail squarely here. The various methods are collectively known as "pair gain", on this side of the pond.

X2 and Vflex were usable. I remember choosing ISPs based on which modem banks they used. I remember when my ISP switched from X2 to Vflex. They denied it but I gave them the make and series of the old and new modems (I had development code for my modem that would tell everything about the modem at the other end). They didn't appreciate being caught in a lie.

ISDN has always been stupidly expensive in the US. DSL is far cheaper.

If you're close enough to the DACs. The US is a big place and there is a

*lot* outside the range of DSL.
Reply to
krw

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.