Measuring CAT 5E frequency response

[jord4231]

Hi, Hoping someone can shed some light on what is going on here,
I have done a simple test with a long length of cat5e cable 50m or so, its a professional grade audio cable.
Using a sweep generator set at 20hz to 20 khz and a oscilloscope I wanted to measure the frequency response of the cable.
First I connected the sweep generators ground to the scope, then ran the signal wire through one strand only of the cat cable, What I got was a signal with greatly reduced amplitude at the lower and top end see first photo.
Then I uncoiled the cable (it was on a roll and took the same measurement again) see photo 2 Looks like I've made my self a big inductor...

I then sent the signal down a twisted pair and the response was a lot flatter (see photo 3)

What exactly is going on here? especially in photo two?. Is the other wire that is twisted around the wire I sent the signal down acting as a choke, inductor / filter? why the reduction in amplitude.
What way would you go about this experiment?
Thanks!

[gf]

Cat5e is not a dedicated audio cable, but actually a twisted-pair Ethernet cable, with a bandwidth of about 100 MHz and a characteristic impedance of about 100 Ohm. The signal is supposed to be transmitted symmetrically balanced over the twisted pairs. The shield is not supposed to act as signal return path. Despite the different designation, I could still imagine that the cable type may work fine for symmetrical balanced audio connections (e.g. XLR).

EDIT: Changed the term "symmetrical" to "balanced".

[iMo]

.. and the cable has to be impedance matched at both sides, of course..

[Nuno_pt]

I know of several guys using Cat 5E and Cat 6 for audio, this cable avoid ground loops, RFI, and some other stuff.

[gf]

I know of several guys using Cat 5E and Cat 6 for audio, this cable avoid ground loops, RFI, and some other stuff.

The key for avoiding ground loops is rather that the audio equipment has symmetrical balanced inputs and outputs. If this is not granted (like for most consumer equipment), a suitable cable alone won't help either.

EDIT: Changed the term "symmetrical" to "balanced".

[madires]

Please see https://en.wikipedia.org/wiki/Balanced_audio and start from there. BTW, your sweep generator has an unbalanced output.

[gf]

.. and the cable has to be impedance matched at both sides, of course..

At high frequencies, certainly yes. At audio frequencies usually voltage bridging (which is typically used in the audio world) sufficies.

[magic]

The symmetry is rather suspicious. Are you sure that your generator doesn't sweep from 20 to 20k and then from 20k to 20?

[Nuno_pt]

Look here < http://audiosystemsgroup.com/AES-RFI-SF08.pdf >

[TimFox]

.. and the cable has to be impedance matched at both sides, of course..

At high frequencies, certainly yes. At audio frequencies usually voltage bridging (which is typically used in the audio world) sufficies.

For many applications, matching the impedance of the transmission line at only one end will suffice.  If the load matches, it will absorb the incident wave and prevent reflections.  If the source matches, there will be a +1 reflection at the load, restoring the original voltage, and the full reflection is absorbed in the source.

[madires]

I don't see any benefit of using Ethernet patch cable over proper audio cable (also shielded TP) for balanced signals. The Ethernet patch cables are less flexible.

[gf]

.. and the cable has to be impedance matched at both sides, of course..

At high frequencies, certainly yes. At audio frequencies usually voltage bridging (which is typically used in the audio world) sufficies.

For many applications, matching the impedance of the transmission line at only one end will suffice.  If the load matches, it will absorb the incident wave and prevent reflections.  If the source matches, there will be a +1 reflection at the load, restoring the original voltage, and the full reflection is absorbed in the source.

Sure, but at low frequencies, short cables are usually not considered to be transmission lines.

See http://audiosystemsgroup.com/TransLines-LowFreq.pdf

[...] Thus, at  audio  frequencies, a cable less than 2,000 ft long is no more complicated than its series resistance and parallel capacitance.  As the cable becomes longer, or as frequency increases, the cable will begin to behave as a transmission line. [...]

[DimitriP]

I know of several guys using Cat 5E and Cat 6 for audio, this cable avoid ground loops, RFI, and some other stuff.

The only "magic" to eliminating ground loops "by using cat5" has nothing to do with cat5 itself , it has to do with not using a ...ground!

You probably want to read this:  https://www.belden.com/blog/broadcast/the-strange-world-of-cat-5e-and-cat-6
Bottom line is you must use a balanced source to a balanced input.

[jord4231]

Firstly I'd like to thank everyone for their response so far.
I'd like to clarify a few things.
When I say this is professional grade audio cable this is what I'm talking about. This is not a solid core cat5e but rather a multi stranded flexible cat5e
I should have been a bit more specific on the cable type sorry.
We are talking of about 50meters of this
Rated: CAT 5e
Shielding: S/UTP
Diameter: 6.2mm
Conductor: AWG26/19
Impedance: 100Ω
Capacitance: 50 pF/m
This is the cable http://soundtools.com/products/cable/supercat.php

The Original idea for this test is because they sell these things called cat snakes http://soundtools.com/products/catfamily/catbox.php it carries a balanced audio signal down a pair with the shield as the ground.

I'm not sure if I phrased my question correctly originally or If I have misunderstood some of the replys, so I'll have another go.

With the signal (centre pin) of my signal gen connected through one single wire in this cable, and the ground of the function gen connected directly to the scope why has it given me such a rolled off response at the start and end of the sweep. I'm guessing as others have said its just due to the impedance mismatch

Down the length of this cable 1 strand measures about 7ohms

[fourfathom]

I don't see any benefit of using Ethernet patch cable over proper audio cable (also shielded TP) for balanced signals. The Ethernet patch cables are less flexible.

Sure, there's a benefit: shielded ethernet cable is cheap, good audio cable is expensive.  There are of course disadvantages, including the lack of flexibility you mention.

[jord4231]

Ok I think I have answered my own question
All these tests are no good and have to be redone. For starters the signal from the sweep generator is unbalanced to overcome this we need a thing called a BALUN it needs to be a ratio of 2:1. Secondly we are what we are seeing in these photos especially the second one is the reflections inside the cable because it must be terminated at both ends with a 100 - 120ohm Resistive load

[DimitriP]

The Original idea for this test is because they sell these things called cat snakes http://soundtools.com/products/catfamily/catbox.php it carries a balanced audio signal down a pair with the shield as the ground.

YOu may have missed the point....what eliminates the ground loop in the"catsnake" arrangement  is that they don't use a ground. Each pair carries one single solitary balanced audio signal.
 
And it doesn't have to be "fancy" twisted pair either.
What eliminated the ground loop is not using a ground signals.
What makes it noise free is that it's balanced.
The best part about cat5 cable is that it's "cheap".   http://www.etslan.com/blue-jeans-cable.cgi

In bulk it's between $350 to $1000 for a 1000ft spool so 0.35 to 1 per foot.
https://www.mouser.com/Search/Refine?Keyword=belden+1700A

You resistance measurement of 7 Ohms for 50 meters sounds high.
https://catalog.belden.com/index.cfm?event=pd&p=PF_1700A

[jord4231]

This Particular cable the ground runs down the shield see photo.
What I wanted to do was see the frequency response of the cable I was not concerned with ground loops or noise

[gf]

Secondly we are what we are seeing in these photos especially the second one is the reflections inside the cable because it must be terminated at both ends with a 100 - 120ohm Resistive load

At audio frequencies and with a still reasonably short cable length you won't notice reflections. But with this circuit

With the signal (centre pin) of my signal gen connected through one single wire in this cable, and the ground of the function gen connected directly to the scope

your cable is forming a inductor with quite some turns as long as the cable is coiled, and still a loop with a large area when the cable is uncoiled, since the signal return path does not pass through the cable.

[gf]

The signal current and signal return current are supposed to flow through the two wires of a twisted pair. Since there are 4 twisted pairs, the cable can carry 4 signals (hopefully with sufficiently low cross-talk). No signal current (or signal return current) is supposed to flow though the shield.

[StillTrying]

Just to be different...

I don't think it's anything to do with transmission lines or inductance, I think it's just pixel aliasing on the scope's display.
If it is, setting the scope's acquisition to Peak-Detect should fix it.