POTS line quality help

Apr 20, 2010 87 Replies

d
e

ng

e
6k

ts

to

hly

I'm baffled why anyone would use dial up. Somewhere in my paper documents I should have tossed (and might already have), I have the frequency responses from the old AEA phone line simulator. I had a tech run the spectrum analyzer on every line prior to designing a proprietary filter for (gasp) 2400bps. I don't recall the designators, but it should be documented in some dusty corner of an old Bell manual.

The problem you have with a 56k modem is the fallback modes are intentionally crappy. What you would do in a modem design is throw the kitchen sink at doing the best possible performance on the top speed of the modem, and then do a minimal job on the fallback modes because they were just to have a check off on the datasheet.

I was at Excess Solutions (Milpitas CA) today and they have two Rockwell 56k modems on the shelf for some crazy price like $15 each. In Silicon Valley, they are worthless. But if you had to use dial up in the boonies, I'd be using Rockwell gear. They always made the best modems. [I worked for the competition, so trust me on this one. We never could meet their specs.] The fallback modes on Rockwell gear were great. You paid for such performance dearly.

I remember when we bought the AEA phone line simulator. It was about $20k in the early 80s. In the 90's, I saw it at RA Surplus for $200.

Purportedly 1200bps and 2400bps modems are still worth something because they're still so commonly used by ATM machines, credit card machines, etc. -- since there's only some hundredsof bytes to be exchanged, the total call time is shorter than with a 56kbps model since there's no fancy training sequence/synchronization/etc.

I guess that will likely go away within the next decade or so as well, as direct Internet access become nearly ubiquitous even in the boonies.

I remember the old 9600bps Telebit models (OMG THAT WAS FAST), and then the Netblazer which had a *built-in router!* as well... all for something in the low 4-digit price range. Now a DSL modem with built-in router and wireless and NAS-like features via USB is under $200...

Amazing that the entire history of Telebit spanned no more than a decade, really.

---Joel

You may be thinking of 3002 leased line conditioning C2 For further info you can find a table describing it at:

formatting link

It has been a long time since I have had to think of those conditioned lines for modems. bye, Jim

Robert Baer wrote:

They won't give you one. Real copper circuits are now reserved for their profitable ADSL customers. Ordinary voice is bandwidth limited to get two (or more) customer voice channels down one physical circuit. Most of the worlds telecoms companies are doing exactly the same trick.

If you have been DACSd (or whatever acronym the US gives it) then you need to find a US based service provider that will do a frequency response line test and feed you back the results.

If you are DACS then forcing a non-56k connection will almost certainly get you a better connection. DACS and the other ways of multiplexing voice onto copper violates the foundations of the way 56k works. Namely that there is nothing altering the signal between the exchange and you so that accurate ADC conversion yields meaningful values.

You might find a prehistoric early 56k UK FAQ helpful in trying to configure your modem to work retro style.

formatting link

(it is out of date and UK based but still has good hints)

Regards, Martin Brown

To the best of my knowledge, USWEST/QWEST did not offer a guaranteed dialup data rate. The rate you got depended on the facility type serving your premises; it might be non-loaded cable (up to about 9kft as I recall), or loaded cable (for a length of cable having a 1 kHz loss not exceeding 8 dB, or a pair gain multiplexer channel. Also, if you believe the circuit is noisy or the transmission amplitude is week, they should measure those characteristics, but if they find no trouble they may charge you for their efforts.

If you can't exceed 26.4 at all, I suspect line "trouble." Beyond that, there is a possibility they moved you to a multiplexer that uses only 32kbit/s coding instead of the normal 64 kbit/s coding. Since you are having a service problem they should tell you what the facility is. If it IS 32 kbit/s coded channel (ADPCM) tell them to put you on a non-ADPCM channel.

The Public Utility Commission can be your friend.

I may be able to help if needed.

Sorry for be> JosephKK wrote:

d
e

ng

e
6k

ts

to

hly

I have US Robotics modem and it seems to have the Rockwell chip; does that sound correct?

Thanks; rather interesting and informative.

  • They did (what they said) "an extensive test" and found no problems; very low noise was mentioned. "Voice quality" is all they "guarantee" - which to me means they can crap the piss out of the bandwidth so much one would be lucky to get
2400 baud.

  • The PUC here does not care; i get better service talking to a wall because at least i can hear an echo.

FYI, RE: JB's post.. leased line parameters are in no way applicable to switched services. Forget about 3002 channels, etc.

Since you had better data service that has become degraded, something is clearly wrong somewhere.

There are some defects the average CO Tech probably will never find without the help of a Data Tech. For example, A terminal multiplexer sync option might be mis-optioned. In such a case, voice grade tests can be in limits, but analog modem data rate will be hobbled. (Been there, fixed that before).

Have you tried another modem? What is your estimate of distance to the CO? What city (town?)? Can you measure the off-hook phone line loop current? How long have you been trying to get help from QWEST?

f

an

and

one

CO

king

ave

56k

ompts

e
45K to

onthly

...

Something changed, certainly. Wrong? Unlikely.

Could be, but they're not likely to spend any time on it since he is getting what he pays for. There is no incentive for them to spend money on the line.

Not wrong. Different. He should force a V34+ connection and see if that behaves any better - but he doesn't listen to advice.

Most likely one of his neighbours has moved to ADSL and the engineers have pinched his real copper circuit for that. Multiplexed local lines do not carry 56k signals at all well. The older analogue technology modems are a bit more robust in that environment.

Regards, Martin Brown

There is a major interconnect box about 2 blocks away and from ther it is maybe 10miles more or less to a CO (have no idea as towhere that may be and if there is more than one so 15 miles is possible if downtown Portland OR is location). Have a newer (at least it looks newer) US Robotics modem; i guess i should try that. Will try to measure off-hook loop current; what should i expect? The latest round with Qwest regarding DSL they indicated that data rates should be better than last year's farce (250K quoted, 200K at max, typical seen 125K WHEN IT WORKED and ranDUMB dropout of carrier for indeterminable time - - all made it useless as even a 10baud rate continuous would be better). They said that DSL would be better because they (using their words as i do not understand WTH they meant) "added a stinger" in that nearby box. Cannot say that my data rate went down from 48+K to 28.8 at or near that time.

INCENTIVE????? For Qwest, that DOES NOT exist, except to extract as much money as possible from their captive customers.

Well,now if i KNEW how to force V34+ i would try.

You should pay more attention to the postings in this thread and also ask on comp.modems and/or us.telecom or whatever it is called. They will know where you can dial into for a DIY line bandwidth test. Assuming that end users are allowed to do such things in "the land of the free".

You might find a prehistoric early 56k UK FAQ helpful in trying to configure your modem to work retro style.

formatting link

(it is well out of date and UK based but still has good hints)

Regards, Martin Brown

Your 56k comment is correct, but the point here is that he can't even to 28.8, which narrows the likely problem/solution. Nothing is gained by considering your point.

D

if

y can

s

s and

, one

LLCO

making

gave

de 56k

re

prompts

y"

to

ine

t 45K to

K

monthly

gs...

r
i

as

r

That info helps a bunch. The Stinger is a multiplexer which has switched voice channel ports and DSL ports. The device handles the switched voice as packets. I've no doubt you are on a port of the Stinger. I suspect the coding of the voice channel before it is packetized, is 32kbit/s or less, but I'm not sure - I'll see what I can find. This would hobble your modem data rate.

If they did not warn you in advance that they were going to put you on a mux that would degrade modem speed, you may get some sympathy from the PUC.

What city are you in?

D

OK; i measured line current at 35mA and am using my "new" US Robotics

56K FaxModem "V.92" (old / original did not have the V.92 designation on it) and i still get 26.4K data rate.

I have that UK reference and it is interesting and a bit informative. I have TerraTerm Pro but i cannot use it with my ISP to do any testing. If i ATDTetc in, that works and see the prompt for username but +++ does not escape to modem; if i am online like now, TerraTerm barfs. So _that_ type of testing is out unless there is another number or alternate to try. One of the suggested tests was with

formatting link
and jitter was

610mSec, packet loss of 0%, MOS score (whatever that is) of 2.9, download speed 26.2K, upload speed of 21.5k and QOS at 92%. Measured loop current 35mA, different US Robotics modem acts the same.

I am "in" Portland Oregon but Lake Oswego is physically closer. And i get as much s*it from the PUC as from Qwest.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required