casey@gauss.llnl.gov (Casey Leedom) (01/09/91)
I just moved into a house in the Berkeley hills. I plan on
tele-commuting to work several days a week using an X terminal and V.32
over Telebit T2500s (GE7.00 PROMs.)
Unfortunately, my V.32 connections keep on dropping after anywhere from
45 minutes to an hour and a half. When I examine the line quality
register, S78, it's 100 both before and after the connection drops. It
appears that the T2500s can maintain a PEP connection forever, but the
``instantaneous'' transmit and receive bit rates, S70 and S72, are very
low, in the 12K-15K range. When I look at the bit assignments per PEP
sub-carrier, I see a curve that looks like the following:
6 | ------ - -----
4 | - -------------------------
2 | ---------
0 |-- - - -----
whereas the typical PEP bit assignment curve I've observed looks more
like:
6 | ------ - -------------------------------
4 | - -------
2 | --
0 |-- - - ----
(Sorry I can't give you exact figures -- I need to hook up a PC of some
kind to capture the data from the modem but I haven't had to time to
borrow one yet.)
I've been told that the two zero drop outs are for the PEP retraining
signals. The big issue is that the mid and high frequency sub-carriers
don't seem to be up to snuff.
I asked PacBell to come out and test the frequency response of the line
outside and inside the house to determine whether it was ``their
problem'' or my house wiring, but they don't seem to be able to do that.
Amazing. Apparently there used to be a PREFIX-00XX number that the
service technicians could call that would provide a 0DB frequency sweep,
but all they've got available now is PREFIX-0020 which provides a 1004Hz
tone at 0DB. I can't tell whether the central office was jiving the
technician who came out (he made a real and very valiant effort to help
me -- this is my second great experience with PacBell service technicians
by the way) or whether they really don't have any method of doing a
frequency response test. I suppose their attitude could be that there's
really nothing simple (read "cheap") that can be done if there is a
frequency line impairment, so why bother test -- besides, it would give
the customers something solid to bitch about (corresponding computer
programming maxim: ``Don't test for bugs you either can't fix or don't
want to deal with.'')
The service technician who came out also mentioned that my house is
just about as far from the central office as I could get without being
assigned to a different CO. However, he thought that all the street
wiring was fairly new. By the way, the drop to my house is twisted
6-pair.
The other possible source of these problems is the house wiring. The
wiring in the house is very old, untwisted aluminum 3-pair. Yes, I said
aluminum! Don't ask me -- everyone I've told of and showed the wiring to
says they've never heard of aluminum being used for telephone wiring ...
and some of them have been in the telephone business for over twenty
years!
I'm running a voice circuit on line one (green/red -- tied to the
white-blue/blue-white pair in the 3-pair) and the modem on line two
(black/yellow -- tied to the white-orange/orange-white pair of the
3-pair.) The white-green/green-white pair of the 3-pair is unattached
and unterminated.
There's about forty feet of the wire strung between the drop box and
the telephone jack I'm trying to use. It's also wired serially through a
jack about ten feet from the drop box. That earlier jack has a telephone
set on line one, but nothing on line two. The far jack also has a
telephone set on line one in addition to the modem on line two.
While in PEP mode, I can hear a very small amount of cross talk when
both the modem and voice lines are idling (very low level regular
clicking.) As soon as the modems start up it becomes nearly impossibly
to hear the cross talk. I don't think I can hear any cross talk when
using V.32.
In any case, because I wasn't able to get anywhere with PacBell, I'm
left to simply replace the wiring in my house an hope that that clears up
the problems.
So, the point of this article:
1. I welcome any comments about frequency response testing and
getting PacBell to fix their wiring if it's the problem.
And just what are the nominal levels of service that they
do promise to provide?
2. I welcome any comments about the potential problems that very
old, untwisted aluminum wiring might generate and in particular,
does anyone think it could be responsible for my frequency
response loss?
3. I think I remember hearing, perhaps in this group, that twisted
pair wiring can actually *degrade* frequency response because of
capacitance coupling. Am I dreaming that up? Will I be doing
more harm than good by running copper twisted pair?
4. We're thinking of running twisted copper 6-pair throughout the
house to accommodate future expansion with an Ethernet, AppleTalk
net, and up to three phone lines. Does anyone see any problem
with cross talk doing this?
Thanks for any help you may be able to offer. Since this is such a
broad question, may spark a lot of discussion, and be of interest to a
lot of people, I think that posting followups would be best rather than
sending me mail. Thanks again!
Casey
P.S. I just learned about the ATJ6J0 command. Here's the output of that
command after one of my line drops. Unfortunately I have no idea
what the output means ... Is the interpretation of this output part
of the ``Undocumented Features'' document by Telebit?
ATJ6J0
R000000 R000 N003 EC000
T000000 F000000 R000000 M000000 E000000
F000000 000000
000 000 000
001
000 000 000 000 L L T000
M000 L001 C000
OKfloyd@ims.alaska.edu (Floyd Davidson) (01/09/91)
In article <88953@lll-winken.LLNL.GOV> casey@gauss.llnl.gov (Casey Leedom) writes: > > [...] > I've been told that the two zero drop outs are for the PEP retraining >signals. The big issue is that the mid and high frequency sub-carriers >don't seem to be up to snuff. Note that high frequency roll-off is normal for long cable runs. The real problem is signal to noise ratio, which frequency response certainly does affect. [...] >me -- this is my second great experience with PacBell service technicians >by the way) or whether they really don't have any method of doing a >frequency response test. I suppose their attitude could be that there's >really nothing simple (read "cheap") that can be done if there is a >frequency line impairment, so why bother test -- besides, it would give They might not be able to do it. But what you say may also be true, and in fact they may not be able to do anything about it. If you are that far from the switch... well you are still going to be that far from the switch, no matter which cable they use. [...] > The service technician who came out also mentioned that my house is >just about as far from the central office as I could get without being >assigned to a different CO. However, he thought that all the street [...] > The other possible source of these problems is the house wiring. The >wiring in the house is very old, untwisted aluminum 3-pair. Yes, I said >aluminum! Don't ask me -- everyone I've told of and showed the wiring to >says they've never heard of aluminum being used for telephone wiring ... >and some of them have been in the telephone business for over twenty >years! Never, ever, heard of it. Is it really untwisted too? Replace it. > While in PEP mode, I can hear a very small amount of cross talk when >both the modem and voice lines are idling (very low level regular >clicking.) As soon as the modems start up it becomes nearly impossibly >to hear the cross talk. I don't think I can hear any cross talk when >using V.32. Disregard this as a meaningful test. Any noise you can actually distinguish from the modem tones will blow the modem away. But what you are hearing could be coming from the modem, you just can't tell with your ear. (I can to some degree with some kinds of modems and I know people who are better than I am at it. But we don't trust our ears, and we have been at it for a long long time.) [...] > So, the point of this article: > > 1. I welcome any comments about frequency response testing and > getting PacBell to fix their wiring if it's the problem. > And just what are the nominal levels of service that they > do promise to provide? I'll pass on this. I don't know what PacBell sells and I'd have to dig up what a "standard" line specs are, which might be different. > 2. I welcome any comments about the potential problems that very > old, untwisted aluminum wiring might generate and in particular, > does anyone think it could be responsible for my frequency > response loss? If it really is untwisted, it sure isn't helping any. > > 3. I think I remember hearing, perhaps in this group, that twisted > pair wiring can actually *degrade* frequency response because of > capacitance coupling. Am I dreaming that up? Will I be doing > more harm than good by running copper twisted pair? It is mo' betta, by fer. (That telco cable is twisted pair all the way from the switch to you, can't be too bad.) > 4. We're thinking of running twisted copper 6-pair throughout the > house to accommodate future expansion with an Ethernet, AppleTalk > net, and up to three phone lines. Does anyone see any problem > with cross talk doing this? Depends on how you do it. A cable should be terminated with the proper load at both ends. If you run a wire from point A to point B to point C with a jack at each point there will be only one time when the cable is properly terminated: when point A and C are. If A is your demarcation box where you connect to the telco line, and B is a phone and C is a modem, then the modem going off hook is a proper load. When the phone is off hook it has this stub line going out to point C that is going to cause a problem, either on or off hook. If the line is short it isn't a problem when off hook, and just double terminates the line when on hook (reduces the volume). If the line is long it may cause things like high frequency roll-off. The important thing is it won't much bother you if B is a phone. It might if B is the modem and C is the phone. But really the best thing for modems is one line to the modem with nothing else on it. You mentioned that you are getting knocked completely off after 45-60 minutes. That is the real problem. Having a slightly slower connect speed because of a poor line is one thing, getting knocked off is another. Another problem too. It may be related to your being able to hear "crosstalk" on the lines. If in fact it is noise or crosstalk that you can hear over the modem tones, that is what is knocking you off. If it is coming from your house wiring it should relate to something local, like when the other line goes off hook, when the refigerator starts up, or something else that generates noise. Floyd -- Floyd L. Davidson floyd@ims.alaska.edu Salcha, AK 99714 paycheck connection to Alascom, Inc. When I speak for them, one of us will be *out* of business in a hurry.
casey@gauss.llnl.gov (Casey Leedom) (01/10/91)
| From: floyd@ims.alaska.edu (Floyd Davidson) | | > 4. We're thinking of running twisted copper 6-pair throughout the | > house to accommodate future expansion with an Ethernet, AppleTalk | > net, and up to three phone lines. Does anyone see any problem | > with cross talk doing this? | | Depends on how you do it. A cable should be terminated with the | proper load at both ends. If you run a wire from point A to point | B to point C with a jack at each point there will be only one time | when the cable is properly terminated: when point A and C are. Ah, I knew there was something I forgot to mention. I'm thinking of running the telco wires into the house to a punchdown block and then wiring the house radially from there to each room. I'll only connect those room cables with instruments to the punch down block, leaving the other room cables unconnected. Thus, the circuit diagram will look something like the following: G ------+-----------+--------- ... | | R --------+-----------+------- ... | | | | B -----------+-----------+---- ... | | | | | | Y -------------+-----------+-- ... | | | | | | | | room 1 room 2 Thus, I'm going to get rid of the currently serially connected line arrangement. This should have been question 5: is this a sensible thing to do? Casey
grr@cbmvax.commodore.com (George Robbins) (01/10/91)
In article <88953@lll-winken.LLNL.GOV> casey@gauss.llnl.gov (Casey Leedom) writes: > > I just moved into a house in the Berkeley hills. I plan on > tele-commuting to work several days a week using an X terminal and V.32 > over Telebit T2500s (GE7.00 PROMs.) ... > The other possible source of these problems is the house wiring. The > wiring in the house is very old, untwisted aluminum 3-pair. Yes, I said > aluminum! Don't ask me -- everyone I've told of and showed the wiring to > says they've never heard of aluminum being used for telephone wiring ... > and some of them have been in the telephone business for over twenty > years! I'd find it really wierd if it was aluminum and not tinned copper... > > 2. I welcome any comments about the potential problems that very > old, untwisted aluminum wiring might generate and in particular, > does anyone think it could be responsible for my frequency > response loss? I wouldn't be too quick to blame the house wiring, the amount of wire there is small compared to the local loop. Usually house wiring problems are due to loose or corroded conections, which cause a static noise, especially when the weather is damp. Make sure the wire and connectors are clean and screwed down tight. Check for obvious crip type splices where shorter pieces of wire were joined or extensions added. You can replace the wire if you want, but I wouldn't be too optimistic about improvments. > 3. I think I remember hearing, perhaps in this group, that twisted > pair wiring can actually *degrade* frequency response because of > capacitance coupling. Am I dreaming that up? Will I be doing > more harm than good by running copper twisted pair? Not likely to be a significant effect either way. > > 4. We're thinking of running twisted copper 6-pair throughout the > house to accommodate future expansion with an Ethernet, AppleTalk > net, and up to three phone lines. Does anyone see any problem > with cross talk doing this? Shouldn't be too much a problem, although you might hear some "digital" noise in your phone due to pick up of "RF" signals from the network stuff. If this happens you can treat it the same way you would CB or Ham Radio pickup. I'd suggest complaining to the telco in terms they can understand. Tell them that the volume is too low or there's ugly static anytime it rains. They might either install a booster amp or give you a different circuit. This *might* help. On the other hand, you just might be so far from the central office that this is all the quality you're going to get. You might also consider getting a "foreign exchange" phone line into that other central office if it's closer or has new equipment, but there's a fair chance that you just end up with an even longer link, still going to the old central office and thence to the other exchange...plus they'll want to charge you a bunch of money for it. Good luck... -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing: domain: grr@cbmvax.commodore.com Commodore, Engineering Department phone: 215-431-9349 (only by moonlite)
casey@gauss.llnl.gov (Casey Leedom) (01/11/91)
| From: grr@cbmvax.commodore.com (George Robbins) | | > The other possible source of these problems is the house wiring. The | > wiring in the house is very old, untwisted aluminum 3-pair. | | I'd find it really weird if it was aluminum and not tinned copper... Believe it. We've scraped away at it and it just stays silver all the way to the other side. I suppose it could be silver (very doubtful), but it really doesn't bend like silver. It has a tendency more to break than bend very well which makes it extremely difficult to deal with. | You can replace the wire if you want, but I wouldn't be too optimistic | about improvements. I'm not. But we really want to put service all over the house, run a net, etc., so I may as well just replace the trash. It certainly can't hurt my performance! | > 4. We're thinking of running twisted copper 6-pair throughout the | > house to accommodate future expansion with an Ethernet, AppleTalk | > net, and up to three phone lines. Does anyone see any problem | > with cross talk doing this? | | Shouldn't be too much a problem, although you might hear some "digital" | noise in your phone due to pick up of "RF" signals from the network | stuff. If this happens you can treat it the same way you would CB or Ham | Radio pickup. Uhmmm, what's the way one would treat CB or Ham Radio pickup? Sorry to be so dense. | I'd suggest complaining to the telco in terms they can understand. Tell | them that the volume is too low or there's ugly static any time it rains. Actually I tried that and then ended up admitting that I was looking to improve my modem service. As I mentioned in my first posting, my experience with PacBell technicians has been outstanding. The guy that came out to investigate my claim of weak voice service tried his hardest to do something, but he couldn't find any way to do a frequency sweep loss test. Apparently they used to have a number, PREFIX-0027, that would produce a 0DB frequency sweep, but they discontinued that number. Now all they've got is PREFIX-0020 which produces a 1004Hz tone at 0DB which isn't much help. A friend of mine who's been in the telco business forever claims that they have to have a way to test the frequency response of their loops and so the Central Office is probably just trying to keep the facility a secret so they don't have to fix lines that are showing low service. | They might either install a booster amp or give you a different circuit. | This *might* help. On the other hand, you just might be so far from the | central office that this is all the quality you're going to get. This is what I'm afraid of. The service technician who came out said that he could install some kind of passive device on the outside of my house but that it would hurt my modem service rather than improve it. (This was several weeks ago and my memory is fading. Sorry I can't remember the correct terms here.) He also said I'm on one of the better (newer) exchanges. I think I'm on a digital switch at the central office and the alternative is to be on some old analog equipment. Unfortunately I just can't really remember what was what any more though. Casey
casey@gauss.llnl.gov (Casey Leedom) (01/11/91)
| From: floyd@ims.alaska.edu (Floyd Davidson) | | >Thus, I'm going to get rid of the currently serially connected line | >arrangement. This should have been question 5: is this a sensible thing | >to do? | | For a modem line its best if there are no other extention lines | connected to it at all. Of course it may work quite well with | three or four on it, but... But this is what I was talking about doing ... Oh, I think see what you're saying. You're saying that if we have, say, two voice lines, which I'll call line 1 and line 2, and one modem line, which I'll call line 3, then, while I'll always hook those lines up to the same positions in the various terminals in various rooms, I should only hook up those lines to those terminals which I have a fair idea will be used. Thus, I would run line 1 (the house residential voice line) through most of the house, line 2 (the office voice line) to the office, and line 3 (the modem line) to the office. Since we're going to use a punchdown block and I'm going to leave plenty of spare wire at the punchdown block to strip back and rewire when necessary, this is a pretty reasonable scheme. I wish there was a junction box that I could use that I could hook everything up to and would provide switches to [en|dis]able each line to each terminal. I think I'll hit the local telephone equipment contractors outpet tomorrow to see what kinds of toys they have ... Casey office, the three bedrooms, the st Do you mean not run the modem telephone line into any room except where we're planning on using it?
floyd@ims.alaska.edu (Floyd Davidson) (01/11/91)
In article <89181@lll-winken.LLNL.GOV> casey@gauss.llnl.gov (Casey Leedom) writes: [...] >Do you mean not run the modem telephone line into any >room except where we're planning on using it? If it is possible, that is the best way. I have two extentions on a line I use for outgoing modem calls, so it isn't exactly the highest thing on the priority list! But if you have problems, one of the first things to try is removing everything except the modem. Floyd -- Floyd L. Davidson floyd@ims.alaska.edu Salcha, AK 99714 paycheck connection to Alascom, Inc. When I speak for them, one of us will be *out* of business in a hurry.
tad@ssc.UUCP (Tad Cook) (01/12/91)
In article <88953@lll-winken.LLNL.GOV>, casey@gauss.llnl.gov (Casey Leedom) writes: (Various stuff about modem problems deleted) > > I asked PacBell to come out and test the frequency response of the line > outside and inside the house to determine whether it was ``their > problem'' or my house wiring, but they don't seem to be able to do that. Doing a 1 KHz test is pretty standard. Doing a sweep is a little more exotic. Pac Bell in your area MAY be installing some Proctor 51200 Dialup Test Sets in the CO that will allow for frequency response testing, but I don't think they have done it yet. > The service technician who came out also mentioned that my house is > just about as far from the central office as I could get without being > assigned to a different CO. This could be a problem. I live about 5 blocks from my CO, a modern 5ESS, and the lines are great. If I lived several miles away, I could still have what the telco judges as adequate voice-grade service, but I might have problems with fast modems like yours. > > 1. I welcome any comments about frequency response testing and > getting PacBell to fix their wiring if it's the problem. > And just what are the nominal levels of service that they > do promise to provide? They are obligated to provide VOICE GRADE service. They will probably tell you that you need to pay Big Buck$ for a data quality line, which is really just a voice line with all the problems engineered out of it. > > 2. I welcome any comments about the potential problems that very > old, untwisted aluminum wiring might generate and in particular, > does anyone think it could be responsible for my frequency > response loss? I would replace the wiring from the demarc (Pac Bell's protector) to your modem with it's own 22 guage twisted pair. Although if the old wiring is in good shape it is probably not causing any problem, it would be best to do this so that you know that your wiring is good. After that, it might be interesting to do your OWN frequency response tests between 300-3000 Hz. Get a freq counter, a signal generator that will match a 600-900 ohm load, and use a quality DVM in the AC volts mode (preferably with true RMS reading) to measure the end-to-end loss at various fixed freqs across the spectrum (say every half-octave...300, 424, 600, 850, 1200, 1700, 2400 and 3400 Hz). Do this from one of your lines to the line that you are using for the modem (I think you said you had more than one phone line to your house?) Then run a comparison at another location where there are two convenient lines for a loop-around test like this, where you think the service is "good." You will need to dial from one line to the other, then terminate each side with a load that will hold the call, but not put noise on the line. An easy alternative would be to use old fashioned phones and unscrew the carbon mic in the handset. Another test...take your DVM, in the milliamps scale, and with the line idle, measure the current from tip to ring (across the line) that the meter draws from the line. If it is anything under about 25 milliamps, you are on a LONG loop from the C.O. Tad Cook Seattle, WA Packet: KT7H @ N7HFZ.WA.USA.NA Phone: 206/527-4089 MCI Mail: 3288544 Telex: 6503288544 MCI UW USENET:...uw-beaver!sumax!amc-gw!ssc!tad or, tad@ssc.UUCP
lstowell@pyrnova.pyramid.com (Lon Stowell) (01/15/91)
In article <764@ssc.UUCP> tad@ssc.UUCP (Tad Cook) writes: >In article <88953@lll-winken.LLNL.GOV>, casey@gauss.llnl.gov (Casey Leedom) writes: > >(Various stuff about modem problems deleted) >> >> I asked PacBell to come out and test the frequency response of the line >> outside and inside the house to determine whether it was ``their >> problem'' or my house wiring, but they don't seem to be able to do that. Pac Bell has a tariff for a "data conditioned local loop". I don't have the specific PUC filing nor the tariff number, but you should be able to get same from Pac Bell. They don't NORMALLY do frequency response (which really bothers VERRRRRY few modems these days..) measurements INSIDE your house, but this is available as "Engineered charges". If you can contact the Test Board supervisor, you might offer to go rent a TIMS if connections are important.... Be prepared to upgrade to a Data Link for your local loops, as to my knowledge, Pac Bell is NOT required to provide data grade local loops for the price of residential services... (NO flames please, I am commenting on the way it IS, not the way it ought to be....) You might also consider modem pooling at the CO.... Oh, yes, get rid of the aluminum. Copper Oxide is a semi-conductor and does strange things. Aluminum Oxide is a pretty good insulator...either WELD all of the connections or replace the wiring. (Never heard of a cheapskate contractor using aluminum for phone lines...just for power--which I think 60 Minutes covered adequately...)