(When) Can I use Cat5e wiring to carry a POTS signal to a DSL modem?

Question

According to this answer by @asndre,

It is very bad idea to use Cat5e-based wiring/cabling to establish a xDSL connection.

However, my telephone/internet provider technician recommended -- and gave me 100ft of -- Cat5e cable to carry the telephone signal to my DSL modem so that I could move the modem. Others have had the same experience.

Of course phone/internet provider technicians can be wrong. (Mine was wrong in saying that the WiFi modem he installed would reach the other end of the house with no problems.) But they could be right. Hopefully they have tried what they're recommending and found it to work.

Why could Cat5e for a DSL connection be a bad idea?

Cat5e wiring is not designed for so low frequencies and therefore does not feature an appropriate characteristic impedance (i.e. Z(f)) at the frequencies employed by xDSL which works in telephone line bandwidth (<20 kHz).

Using Cat5e for xDSL also results in dramatical signal attenuation, about 3~5 times worser comparing with an appropriate cable/wire.

To be short, POTN is about 600-Ohm techniques while Cat5e if for 100-Ohm ones.

That sounds to me like the answerer knew what he was talking about. (He also said he had done testing that supported his conclusions.) However the original question was not about the suitability of Cat5e cable for carrying a telephone signal to a DSL modem, so that issue didn't get discussed much.

On the other hand, maybe the results of the impedance mismatch are such that it wouldn't affect actual DSL performance in a typical household? I understand that the effect of impedance mismatch could be signal reflection (causing distortion) and reduced power transfer (= signal attenuation?). This Wikipedia article section suggests that impedance matching to telephone lines in modern devices is no longer very important due to active amplification and filtering.

Does anyone have experience or insight into whether using Cat5e wiring to carry a POTS signal to a DSL modem causes noticeable problems?

I fail to see where cat5 cable is worse than the normal telephone wiring in your house, or the 50 year wires used to connect the house to the DSLAM. ADSL is **made** to operate over cheap, plain old wire. Cat5 can't be any worse than that. — JRE, Dec 29 '18 at 13:26
And, I've used Cat5 and satellite TV cable to connect a DSL modem. No troubles. — JRE, Dec 29 '18 at 13:28
100 ft = about 30 m. Typically, xDSL spans km-long distances. It has enough power to overpass (i.e. waste on) an additional (let's take 1-km example) 30/1000=0.03=3% of the "renewed" full path. Try to look on your situation so your technician did have no cable other than Cat. 5e TP :-)). Yes, Cat. 5e is so universal while you are in the house/office (in-door), i.e. the link distances are (typically) no longer than about 100 m, but in the "out-door world's" distances it is not so. — asndre, Dec 29 '18 at 13:40
@JRE: Something could be mismatched to a particular use, without being "worse" in general. From what I read, plain old phone wire and Cat5e are specified to a certain impedance at difference frequency ranges. Certainly Cat5e cable is much better for uses like ethernet, but I wouldn't use it to hang a tire swing. — LarsH, Dec 29 '18 at 14:31
If you wish to test whether this matters for you, set up a computer beside your phone terminal block (where the DSL signal enters your house) and connect your modem with the full length of Cat5e(spread it on the floor, don't leave it in a coil), run a few connection speed and ping tests, then replace the cat5e with a short length of regular phone cable. Chances are you will have your full connection speed and near identical ping. If the Cat5e does show a measurable difference, shorten it to the length it would actually be installed and check again. — K H, Dec 29 '18 at 18:07
By the time you've tested you'll have a good idea of whether you want to take the windfall or spend money on a different cable. — K H, Dec 29 '18 at 18:11
Those who have commented, if you would submit an answer, I would be glad for the opportunity to upvote the useful ones and have more choice over what answer to accept. — LarsH, Dec 31 '18 at 04:00
https://electronics.stackexchange.com/questions/400001/cat5-for-telephony#comment982779_400037 - a close scenario but a slightly different problem: Untwisted vs. Cat 6. — asndre, Jan 02 '19 at 11:05
https://electronics.stackexchange.com/questions/153011/whats-the-best-phone-cable-for-dsl/153012#153012 – appropriate user experience + maybe spurious but some theory behind the ground. — asndre, Jan 02 '19 at 11:06

score 3 · Accepted Answer · edited Jun 11 '20 at 15:10

During my conversation with Jack Creasey, I “refreshed” in memory some normative material on ADSL and think that the next could be interesting:

Source: ITU-T G.992.1 (06/1999), ADSL.

Along with small guidelines and useful limitations, it could be considered “indirect evidence” that CAT5 cable is at least assumed as a possible in-home wiring by the then-days researchers.

Moreover, ITU-T G.996.1 (02/2001), Test procedures for DSL transceivers, considers CAT3 (predecessor?) cable appropriate among in-home wiring models (see subsection 6.2).

Maybe, this will give you additional confidence around your cat.5e cabling.

P.S. Also, Annex A/G.996.1 describes the characteristics (from 1 Hz to 5 MHz) of 26 AWG, 24 AWG, and 22 AWG cables (and test line models based on them) which are used in (lab) performance testing of G.992.1 ADSLs. Who knows, may be modern Cat. 5e fits the then-expected requirements...

score 2 · Answer 2 · answered Dec 29 '18 at 20:01

2

Short answer: Cat5 cable will work just fine for your DSL AND for your phone line.

The element missed by most answers and the listed one is a great example of how to be completely wrong, is that the characteristic impedance of a cable is frequency sensitive …..IT IS NOT A CONSTANT VALUE.

Plains old POTS telephone cable is NOT 100 Ohms at voice frequencies, it's actually more like 300-700 Ohms at the voice frequencies used over the phone line (about 300 - 4000Hz). The characteristic impedance drops as the frequency gets higher and asymptotically ends up at about 100-110 Ohms beyond 1Mhz.

Here is what the POTS cable looks like with frequency:

Here for more detail.

POTS only requires about 4kHz of bandwidth and the cable impedance is in the 300-700 Ohms range.

Next you need to understand that the bandwidth required for xDSL varies, but the spectrum usage looks like this:

More detail here.

Notice that the frequencies are all between 25kHz - 1MHz so the cable impedance will vary from around 140 Ohms to around 100 Ohms for all the frequencies involved.

Cat5 cable is just the same ….it's about 100 Ohms for the frequencies above 1 Mhz and you don't see specifications for frequencies below 1Mhz, but you can guess it tends toward 300-700 Ohms for voice frequencies.
You may find this informative.

While there will be differences between POTS telephone cable and Cat5 the differences will be small for the frequencies involved.

answered Dec 29 '18 at 20:01

Jack Creasey

21,428
2
15
29

While your answer may be correct, I don't see how you can say that the linked answer missed the fact that the characteristic impedance of a cable is frequency sensitive. It says that cat5e wiring is "not designed for so low frequencies" and "does not feature an appropriate characteristic impedance (i.e. Z(f)) **at the frequencies employed by xDSL**". – LarsH Dec 29 '18 at 23:18
@LarsH The linked answer says this: "To be short, POTN is about 600-Ohm techniques while Cat5e if for 100-Ohm ones." ....which compares the impedance of phone cable at audio frequencies to Cat5 beyond 1MHz ...which is completely incorrect. Both cables have about the same impedance whether at audio or RF frequencies. One could just as well say that phone cable is not and was never designed for RF frequencies, but is used for xDSL quite successfully. – Jack Creasey Dec 29 '18 at 23:24
1

@JackCreasey IMHO, [expression/function] Z(f) explicitly shows that "the characteristic impedance of a cable is frequency sensitive …..IT IS NOT A CONSTANT VALUE". Is not it for you? Why? In the linked answer there is no claim about Z(f)=Zo=const. Instead, the impedance is mentioned as a function of frequency and therefore expressed as Z(f). is not it? – asndre Dec 30 '18 at 09:22
1

@JackCreasey I could agree with your short answer ("Cat5 cable will work just fine for your DSL AND for your phone line") if you'll make some important addition to it like this: "at the distances insignificant to the POTN (i.e. from the provider to the household) span". – asndre Dec 30 '18 at 09:40
1

@JackCreasey I also want to draw your attention to the second factor in the linked answer: the signal attenuation. Taking in the consideration the same cat. 5e cabling, one can see that while its characteristic impedance in the range of 1 to 100 MHz is specified to be nearly "constant" (100 +/- 15 Ohms), its insertion loss in the same range behaves other, non-"constant" way: 2.2 dB/(90H + 100E)m @ 1 MHz, ..., 24 dB/(90H + 100E)m @ 100 MHz per TIA/EIA-568. And it is not specified below 1 MHz again... – asndre Dec 30 '18 at 10:09
@JackCreasey From the first source you cited: "Figure 4 obtained from theory ... of a cable 0.305 km long." The last source you cited gives only Zo (1 kHz) = 539 - 522j for Cat. 5 TP, for other frequencies it states: "More on this when I find a clean way to post the data." Until now, there is no data on other frequencies. Both sources state that real Z0(f) measurements will depend on the certain cable (line) length. Do we still can compare so directly? – asndre Dec 30 '18 at 12:43
@asndre why don't you just admit that as twisted pair cable the telephone and Cat5E are going to be much the same in the spectrum needed. You answer said they were absolutely not the same which is why I said you were completely wrong. Cat5 is used commonly for xdsl without problems. I suggest you alter your answer to say that. – Jack Creasey Dec 30 '18 at 16:32
@JackCreasey My answer was very conditional (imo), not absolute at all. What is made it so in your eyes? "Cat5 is used commonly for xdsl without problems" - at which distances? Could you share your experience on that? – asndre Dec 30 '18 at 19:58
@asndre Homeowners are rarely if ever required to run xDSL cables long distances. The phone utility is responsible for the street to home wiring and the homeowner simply connects to the internal phone points. Typically this will only be a few meters, and maximally a few 10's of meters. At these distances the difference between phone cable twisted pair and Cat5 twisted pair will be minimal over these distances. So can you define why you say using Cat5 is a "very bad idea" Also correct the fact that xDSL works at audio frequencies (<20kHz), which is patently wrong – Jack Creasey Dec 30 '18 at 20:05
@Jack You're saying that because the "short" summary at the end didn't mention frequency sensitivity, the answer "missed" frequency sensitivity, even though it mentioned frequency sensitivity twice? – LarsH Dec 31 '18 at 04:03
@LarsH No I'm saying that trying to compare the voice frequency impedance of telephone line to the approximately 100 Ohm impedance of Cat5 made the answer completely wrong. It compared Apples to Oranges. No amount of selecting exactly which sentence to mean what changes that. The answer said that you could NOT use Cat5 to replace tp telephone cable, when in fact you can. – Jack Creasey Dec 31 '18 at 04:13
@JackCreasey Please be careful, my answer said "it is very bad idea" not "it is impossible/prohibited" as you try to proclaim. Also i used the term "600-Ohm techniques" (not "600-Ohm impedance"!), because it is (imo) very common term (in telecom) with a very broad definition. Like the terms "100-Techniques", "50-Ohm techniques", ect, too... I'm not limiting anyone in what he/she can(not)/could(not), i'm only trying to show some possible (!) implications grounding on my customer/engineering experience. – asndre Dec 31 '18 at 09:54
@JackCreasey To be short, in the scope of ADSL per ITU-T Rec. G.992.1: ADSL psd mask has a limit of -92.5 dBm/Hz at **4 kHz**... (is it <20 kHz?) You can say that -92.5 dBm/Hz is so low but ITU-T Rec.99 thinks not, stating "_The step in the PSD mask at 4 kHz is to protect V.90 performance. Originally, the PSD mask continued the 21.5 dB/octave slope below 4 kHz hitting a floor of –97.5 dBm/Hz at 3400 Hz. It was recognized that this might impact V.90 performance, and so the floor was extended to 4 kHz._" (A longer answer is given in our discussion on my answer to the linked question.) – asndre Dec 31 '18 at 10:15
Great numbers @asndre, but absolutely nothing to do with xDSL frequencies per se. The specifications are for a protecting use of V34/V90 modems (which are inband, < 4kHz carriers) from interference form xDSL modulation artifacts. The 4kHz references the steps used to ensure minimal interference. xDSL does NOT use inband frequencies. Above the 4kHz limit for audio is a guardband up to about 25kHz where the first bin starts. To repeat xDSL does NOT operate at telephone audio frequencies (<4kHz) at all. – Jack Creasey Dec 31 '18 at 17:07
@asndre Perhaps something like this may help you understand the difference between inband (voice frequency) modems and the digital xDSL implementations which are out of bans: https://www.itu.int/ITU-T/worksem/asna/presentations/Session_6/asna_0604_s6_p4_palm.pdf While much effort was expended to ensure xDSL did not interfere with analog modems (v34/v90), today that is rarely a peoblem. – Jack Creasey Dec 31 '18 at 17:20
@JackCreasey Specification vs. presentation? (Hm... very interesting approach...) Now you claim "To repeat xDSL does NOT operate at telephone audio frequencies (<4kHz) at all", but early you claimed from about <20 kHz... and "xDSL modulation artifacts" from above 4 kHz... You are contradicting yourself. – asndre Jan 01 '19 at 15:28
@asndre I am not contradicting myself at all. xDSL does not operate at audio frequencies. I don't consider the first bin at 25kHz audio ...do you? I linked to the presentation since you don't seem to understand the frequencies (and impedances) involved. I was simply bringing the discussion to a point where you might understand. Enough has been said to clearly establish that you don't understand so I'll stop here. – Jack Creasey Jan 01 '19 at 16:03
@JackCreasey In common, I’m trying to distinguish between ADSL and other non-ADSL DSLs. In particular, where xDSL == ADSL per G.992.1 Annex A, I’m trying to distinguish between the nominal frequency of the lowest possible subcarrier (tone) and frequencies (including lower than the lowest possible subcarrier) that can be occupied by the spectrum (resulting from modulation) of a possible signal. But as I see to my regret, you are not. Thank for discussion. Goodbye. – asndre Jan 02 '19 at 11:07
I think your answer provides a (very interesting) answer to this question and points out the flaw in @asndre's reasoning about impedance, but your rough manner of replying to him seemed unnecessary. – Milind R Sep 15 '20 at 09:01

score 1 · Answer 3 · answered Mar 08 '20 at 17:15

I have spent 41 years working for local exchange carriers (ILECs). (The local phone company). I have also spent a good portion of my career designing SLIC circuits (subscriber line interface circuits).

The older homes have quad "station wire" that is not twisted pair and is not impedance controlled nor balanced very well. DSL does work on quad.

Seems to work better with CAT 5. Our techs only stock CAT5 on their trucks since the advent of DSL.
I wonder what the guy that said it will not work would expect us to use to wire up a DSL circuit? CAT 3? Good luck getting that. Moreover, when we convert a home to fiber to the home (FTTH) then we already have ethernet ready wiring.

(When) Can I use Cat5e wiring to carry a POTS signal to a DSL modem?

3 Answers3

Linked