Nobody suggested that most of such noise was in the GHz range. Why make unequivocally useless comments that lack all relevance?
Such noise components *do* exist, whether James Knott has yet had enough experience to know about them, or not.
Sigh, another non sequitur.
That doesn't even make a good strawman, so why beat on it?
You are, again, confusing useful spectrum with actual spectrum.
No one suggested that in a properly designed, installed, and functioning system such interference will occur. We have *all* agreed from the start that 10baseT utilizing CAT3 or CAT5 cable with RJ-45 connectors was *designed* to work with POTS lines on the same cable.
The question and discussion point was what potentials exist for problems from miswired connectors and defects such as kinked cable. Not to mention the claims by some that 20 Hz ring voltage from a POTS line could be applied directly to a pair running 10baseT without even causing errors, much less damaging the network interface. That particular subtopic resulted from questions about how 20 Hz could ever cause any interference, given the disparate frequency.
All of that has been discussed in detail, with specifications of
10baseT and telecom systems compared to demonstrate the likely effects under various circumstances... and all despite your unwillingness to understand most of it.
Incredibly horrible phone lines exist almost everywhere. What causes them might indeed mystify *you*... ;-)
On the other hand some of us actually do have enough experience and the theoretical background to understand, and at least make attempts at explaining, what causes impairments on telecom circuits, whether they are voice or data.
My position on that was exceedingly clear, and was posted prior to Rich's fine comments. Perhaps your confusion is caused by not having understood it? Here's the exact quote, which I really do not think is easy to misunderstand unintentionally:
"That is in fact what CAT5 was designed for!"
No amount of arguing from you will convince anyone that you have read or understood this thread if you continue to make such misdirected statements.
??? ethernet transceivers are coupling transformers. They fundamentally respond to current, not voltage (although they have a characteristic impedence and it takes voltage to produce current).
DC ringer. AC ringers don't need contacts.
I very much doubt it, especialy if there are no other phone/LF services nearby. Silver satin and various forms of untwisted cable are perfectly fine for voice. I do not see much difference between untwisted cable and split pairs unless the split conductors are themselves paired.
Or does it take current to produce voltage? Or are you chasing your tail?
The fact is the *receiver* is a voltage sensing device.
And just how many telsets use a DC "ringer"? And why is ring current AC? ("I've never heard of such".)
Well, I can lead you to the water, but...
Like I said, the result is *precisely* the same.
That is simply *not* true. The amount of unbalance is less at lower frequencies, and to the degree that VF circuits use fewer high frequency components they are affected less. Some data formats of course us fewer low frequency components, so they are necessarily affected more. But at any given frequency the imbalance and resulting impairments are *precisely* the same.
While a handful of silver satin patch cords that are 6 feet long will work fine at voice frequencies, attempting to use longer or more cords is an invitation to trouble.
The same is true of data circuits. I've seen people wire DSX-1 and T1 circuits with silver satin, in a telephone office no less! It worked. And because it worked we didn't ask for an outage to correct it either.
Incidentally, in another group someone mentioned replacing ~70M of flat satin with twisted pair to correct problems with an ADSL circuit. In the discussion that followed, one fellow laid out 150 feet of RG6 coax, and patched his ADSL through it! It worked. He said the phone had a barely audible buzz.
Interesting examples, because the 70m of flat satin could be about half expected to work, and I doubt that anyone would expect 50m of coax to work. But the one didn't and the other did, /in/ /those/ /particular/ /situations/.
Well barbed wire works fine too, as long as there are no external fields to induce interference.
Unbalance on a cable may or may not result in a circuit malfunction, but to the degree that there is unbalance there is potential for interference. The degree of imbalance from a split pair is relatively greater for any given length of cable than is the imbalance from non-twisted cable pairs.
Which is to say that if you use a 6' flat satin cord with a split pair in it, there is almost *no* chance that the imbalance will be enough to present a problem. It can be measured though, if you have some very good test equipment.
I've never seen a DC ringer used on a telephone, though I have seen them with intercoms, that use telephone sets. In that situation though, the buzzer was on a separate pair.
{"Ringing can not interfere with 10BaseT; it's the wrong freq.."}
Here's another way to think about it. Move from the frequency domain to the time domain. At point t1, the 20Hz signal is zero. But at point t2-t6 it's big vice that Ethernet on the adjacent pair....and that's what may KO the bits going by...
Oh, its even worse than that. The ringers I recall from the days of mechanical bells all had inductors that had a little flapper (just like on a relay) that normally had an air-gap. When the flapper was pulled over that gap closed. That changed the inductance greatly. The resulting dL/dt puts a nice spike into the otherwise smooth 20hz sine wave.
The full equation for the voltage across the inductor is really
e = L di/dt + i dL/dt
We are so used to working with fixed-inductance inductors that it is easy to forget about the (i * dL/dt) term.
To see this dL/dt effect in action one can drive a relay with a low-frequency sine wave (say 1hz - 10hz) and put a scope probe across the relay. At the point the relay pulls in one should see a little spike. It might help to increase the signal generator's impedance a bit by adding a series resistor.
That's a good example. And these kinds of devices may not be common, but in some places they aren't rare.
The 400 ohms value is unlikely to be the battery resistance either. Originally that was a typical value when line current, sensed by a "line relay", was supplied through two windings of a polar (balanced) relay. The typical line relay had 200 ohms DC resistance per winding.
Today the loop current is often regulated, and hence the effective series resistance can be almost anything. Different manufacturers regulate at different loop currents, usually from
25 to 45 milliamps, so the range of values is all over the map. Particularly because, again with equipment like subscriber line carrier and loop extenders, the loop current might be as high as
120 milliamps, though the original recommendation was to not exceed half that.
Telephone loops vary greatly. Moreover, several specs say something to the effect that 95% of the plant should meet a given criteria. The point being that an impaired, but functional, telephone line is allowed when that is the only possible way to provide service to Granny out on the Farm.
Now who is being pedantic? Voice is nowhere near data in frequency. So the truism of idential impairments has no practical effect.
??? most of telco outside wire is Cat0, essentially identical to silver satin but solid with weather/burial protection. A lot of older wire is even worse, with even paper insulation. Yet hundreds of millions of circuits all work fine over thousands of feet per run.
Of course twisted pair is better. Expecially an EMI noisy environment. But the difference is marginal. With digital, marginal effects can make large differences.
Precisely. Which is why local field strength is an issue. A split pair causes no trouble if there's nothing to interfere with it (troublesome freq). One split voice pair in a shared sheath isn't much problem, although it will pick up more 60 Hz hum. Two are definitely a problem, especially since they're likely to be split on each other. Either way, the data won't care so long as it's not split.
Granted, but I see that difference small (~20%) comparing with with twisted pair. And under some circumstances, non-twisted pairs would be _worse_ than split pairs (think side-by-side flat pairs vs USOC)
More than that. I've pushed 10baseT fine through 50ft silver satin with USOC pairing. It totally fails as side-by-side. I suspect 10baseT could be pushed through a good length of quad with diagonal pairing so long as it wasn't too kinked.
I hope not to have been less than pedantic about *anything* in this thread! The entire discussion has been pedantic from the start. What's your point?
I try to be pedagogic too, but you and James Knott are very difficult to instruct!
That is simply *not* true. What is or not "data" does not equate to frequency in any way. "Voice" distinctly defines a relative ranges of bandwidths, but "data" does not.
You just happen to work only with data services that arbitrarily use high rates. What about a 7 baud circuit? Or even a 45.5 baud circuit?
(To be pedantic, I'm assuming above that by "data" you meant "digital data", because in fact "voice" is one form of "analog data". ;-)
That is not true. Almost all outside cable plant is twisted pair. (Though not quite all, only because some relatively short runs of drop wire are non-twisted.) I don't know that today there are any remaining runs of 1 mile or more using open wire line...
Other than various kinds of two pair drop wire there are *no* multi-pair cables that are non-twisted. (Though there are at least two non-twisted conductors in every cable, but I'll leave the purpose and use of those as an exercise for the reader.)
All of it on twisted pair. That is significantly distinct from silver satin, which is not twisted pair.
It is not hardly "marginal", it is *vastly* better.
I should probably also remark specifically on the idea that marginal effects can make a large difference for digital services. The opposite is true. Digital is immune to many "marginal" impairments, which is specifically the reason todays telecommunications networks are virtually all digital. The SNR of the data remains high (hence the error rate low)) despite changes of marginal impairments (such as the facility SNR), with the additional significance that the "noise" is not additive (multi-link circuits have an end to end SNR equal to the highest individual link SNR rather than the ratio of the signal to the sum the noise from all links, as it would be with analog facilities).
That is vitally important to the design and implementation of telecom facilities. Because the error rate remains low (even when the SNR change is *far* from "marginal") it is possible to use m-ary (multilevel) encoding schemes that trade SNR for bandwidth to achieve greater capacity. That is, for example, the essence of V.90 digital modem protocols compared for example to V.34 analog modem protocols!
Of course there are very few examples of locations where that is the case, given that today almost any telephone service is where there are AC power distribution services too.
One split pair in a shared sheath can only be a non-problem as I noted above... in an environment where there is no external field. That means, for a multi-pair cable, that it effectively is
*always* a problem if any other pairs are used.
In fact however, external fields are not the only problem with split pairs. That is merely the only cause of induced noise. Longitudinal imbalance will result in increased transmission losses too, which are frequency dependent and as the frequency in use increases that can be a fatal impairment too. Which of course is true for either digital or analog uses.
You do know what a split pair is? It *always* involves two pairs. Whether both of those pairs are used is of little consequence in a cable with other active pairs.
The data *will* care. How much depends; but given any significant distance or external induction the effects will be quite significant.
"Under some circumstances" means nothing, because under an equal number of other circumstances the opposite would be just a true.
That is a foolish statement. At 6' there is little risk. At
12' there is significant risk. At 50' it is pure folly, but sometimes luck makes it work (while you watch, and then it fails when you aren't there).
Of course you can "push it through". But what you can't do is install that and walk away with any confidence at all that it will continue to work. The question is not if there will be a failure, but is merely a matter of how long until something causes it to fail. That is why there are standards and why installation to standards has advantage. It *continues* to work, even when other factors change at a later date.
A 45.5 baud 130 mA circuit maybe, Floyd? With an KSR 28 on the end....
Huh?
I agree with Floyd here. Only butt-ugly 2 conductor drop is untwisted. And I don't know that Ma even uses it anymore..
[pulp]
If you think pulp cable is not twisted, you've got some manhole [OOPS "maintenance hatchway"] time coming. All that lead-sheathed cable is quite twisted. [A local crew just cut 600' of ~1200 pair pulp out and replaced it with icky-pic. I stopped and chatted several times. The MH was so cramped they were doing only 100 pairs a day....]
....
Take a battery-power scope and go find a field far enough away from the grid that you can not see 60Hz on a 10' long antenna...
I think he's talking about drop wire. I know GTE used to use non-twisted pair drop wire in a lot of places. I always thought it was a damned fool idea, but...
...
Used to drive GTE right up the walls...
Back in the early 80's I worked on Eielson AFB, south of Fairbanks, and most of the time they had a humongus 1800 Hz tone induced into every pair on the base. We did eventually determine the exact source, but were not able to get it fixed. The base power plant had one particular generator that produced it. If that generator was on line, it was everywhere.
Pedantic is not usually considered desireable. This discussion started with a very simple question to which I and others said "It's OK so long as you don't split a pair".
Please see the Subject: line for a precise definition of the data & voice we are discussing. I maintain my point that these vastly differing frequencies (aided by balanced signalling) greatly reduces the _effect_ of interference.
Newer cable most certainly has some twist. Older cable was more rolled as a bundle(s) than twisted pairs. The roll helps against external fields (60 cycle hum) but doesn't help crosstalk.
Certainly vastly at 10baseT (and moreso at 100baseTX), but the difference is much less at ADSL freq.
Yes, quite true. The downside of digital is less discussed. When the error rate significantly exceeds the design BER, then the whole thing fails entirely. Think of a BER of 0.1% on TCP/IP. A split pair on 10baseT often causes less.
We are talking a single POTS circuit and 10baseT two pairs inside a single sheath. See the OP.
External fields are not the only cause of induced noise. Crosstalk is very severe in a split pair cable, and governs many 10/100 systems. It's what causes longer lengths of Cat3 to fail in
100baseTX.
Longitudinal imbalance? A split pair is nowhere near as bad as putting raw signal (no baluns) across coax as you cited. The impedence of the shield is hardly specified, and even the signal pathlength and velocity of propagation (slew) will be different. Yet you reported it worked over 150'.
The impedence of a split pair will be off quite a bit, but at least both splits will be the same if they're against similar (grounded/floating) conductors.
Of course I know what split pairs are. I like to say: "Electrons may be color blind, but they know who their dance (twist) partners are." And obviously splitting pairs involves more than one pair.
Consider the classic newbie goof -- RJ45 with pairs in sequence. pins1&2 are paired and 10baseT in one direction runs fine. Pins3&6 are split so the other direction is full of errors (may not fail at 10, but most likely will at 100). But if POTS has been put where it usually is (4&5), this pair is also split, but against the ethernet split! I'd have to double check the Left Hand Rule, but I believe this doesn't cancel but amplifies crosstalk.
Folly? It's been working fine for a year, even with considerable flex. It can now be easily fixed, but there's just no real driver until it fails.
Of course it's a failure risk. But people have a right to take informed risks. Standards are good and should be followed, but sometimes the cost of complaince is greater than the risk-adjusted cost of failure.
Cabling-Design.com Forums 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.