Twisted pair RCA cables again

I wrote few years ago blog article Unshielded RCA cable is bad design. I just received yesterday a comment on it:
“It’s not always that simple. There are equipment combinations which REQUIRE twisted pair RCA. Here’s explanation:

So I watched this Truth about RCA signal cables video mentioned on this comment:

The video was pretty well made and I can fully agree most of the things said on the video. Here are some information picked from the video.

Here we discuss the circumstances of when it is appropriate to use a twisted pair RCA cable vs a Coaxial RCA cable:
“Coaxial RCA cable is good and twisted pair is better”
“That’s not the case”
“There are specific cases where you would want to use each one of them”

You want to use coaxial if you have amplifier with single-ended inputs. In those inputs the outside of RCA is grounded. This type of inputs are used on average amplifiers (70-80% inputs in audio equipment). So for normal single ended RCA connections the coaxial cable is the right cable type.

The video says that if you have amplifier with differential input you want to use twisted pair wiring. Those are found on some of the high end amplifiers Only a small percentage of equipment have this type of RCA inputs.

If you mix those different cables and different input types, they don’t work that well: the connection become easily noisy.

According to the video if you have amplifier with differential input you would want to use twisted pair wiring. Those are found on some of the high end amplifiers Only a small percentage of equipment have this type of RCA inputs. I cam agree this. I have very rarely seen those differential input RCA connectors anywhere except some car audio equipment.

There is one considerable benefit on differential RCA input: The advantage is that there is no direct connection between the two grounds of the source and the amp, so no possibility for ground loops that, which could set up a voltage across the ground lead, which could lead to hum etc. There are no other advantages in sonic performance than the ground issue.

The video maker comments on Truth about RCA signal cables video page:

If the head unit has SE outputs and the amplifier has SE inputs = coax cable

If the head unit has SE outputs and the amplifier has Diff inputs = twisted pair

If the head unit has Diff outputs and the amplifier has Diff inputs = twister pair

If the head unit has Diff outputs and the amplifier has SE inputs = bad no matter what as you will shorting out half of the head unit’s outputs. In this case you would need an amplifier with Diff inputs or a converter to convert the head unit’s outputs to SE

According to those comments you need to what type of inputs and outputs your devices have to select the most suitable cable type. The differential inputs and and outs are like how XLR /balanced inputs work.

I said in my original Unshielded RCA cable is bad design article: When you wire unshielded twisted pair to unbalanced signal source and receiver RCA connectors, most of the good properties of the UTP cable noise rejection are lost. This holds true.

When you have system where you have unbalanced signal source and differential receiver, you get some of the benefits of differential system, but not all. The differential RCA receiver can get rid of the noise that is coupled as common mode noise to the twisted pair. In systems that use Balanced Line Technology the twisted pair cable construction combined with differential receivers and signal sources make sure that both inductively and capacitively coupled external noise to the wire pair gets converted to common mode noise that can be removed in differential receiver.

When you have system where your signal source is single-ended, things are somewhat different. Let’s consider first case where the single-ended signal source is grounded. In this case one of the signal wires is at ground potential and other carries the signal. The coupled magnetic fields generate the same amount of noise to both wires, so the magnetically coupled noise is still removed well differential receiver.

The story is different for capacitively coupled noise: There will be some capacitively coupled noise induced to the signal carrying wire (how much depends on electrical field strength and source impedance), but the same field does not cause any noticeable voltage on the grounded wire (it’s impedance is almost zero!). All this means that the capacitively coupled noise will not be coupled as common mode noise, and the differential receiver cannot get rid if it. Unshielded unbalanced cables are very sensitive to capacitive coupling especially when circuit impedance is high (impedance mostly determined by signal source output impedance in HIFI systems). The differential receiver does here nothing to help in this.

So I don’t think that unshielded twisted pair cable in this case would guarantee to be considerably better that coaxial cable. Depending on your noise source type, it could be somewhat better (only magnetic noise) or it could be considerably worse (against capacitively coupled noise on high impedance systems).

The situation for capacitively coupled noise is somewhat different if the signal source equipment is not grounded. In this case the impedance different the capacitively coupled signal sees is much smaller, so the signal gets coupled more equally to both twisted pair wires (not completely equally), and the differential receiver can get rid of most of the capacitively coupled noise to the cable. The downside with ungrounded mains powered equipment is that there is always some leakage from mains to equipment case, and this leakage can easily generate common mode noise to the cable that is much higher in amplitude than the audio signal on it (few volts to tens of volts depending on impedances and leakages). Usually at least some of that voltage gets to audio signal as noise because the differential inputs have their limitations on their common mode noise canceling capabilities (limited CMRR capabilies typically 40-70dB and sometimes the noise voltage can exceed the common mode voltage range the input can operate with). So if you are working with differential inputs you generally want the signal source to be grounded in way or another (or to be completely floating battery powered gadget or microphone).

The best cable for a setup that uses differential RCA inputs and single-ended output would be shielded twisted pair. The twisted pair construction would perform well against magnetic noise, and the shield (grounded at least on signal end) would get rid of the capacitive noise. So this would be the optimal cable.

A differential input only really works best with a balanced output and twisted pair/screened interconnect cable. This is how professional audio equipment that use Balanced Line Technology and XLR/phone connectors are typically wired. With this kind of balanced output and differential input system an unshielded twisted pair can also work acceptably.

The second best option for interconnection is unbalanced (single-ended) output wired to shielded twisted pair cable that is connected to differential input. Trying to use unshielded twisted pair here will give considerably worse performance.

So I can still stay on my original claim that Unshielded RCA cable is bad design. This will hold true very well. Exceptions to this rule (if there are any) are really rare. I can also say that twisted pair construction on RCA cables is better on some applications when the cable construction is shielded twisted pair.

In case the twisted pair cable used for interconnection is shielded twisted pair cable, I can agree on those comments on Truth about RCA signal cables video page:

If the head unit has SE outputs and the amplifier has SE inputs = coax cable

If the head unit has SE outputs and the amplifier has Diff inputs = twisted pair

If the head unit has Diff outputs and the amplifier has Diff inputs = twister pair

If the head unit has Diff outputs and the amplifier has SE inputs = bad no matter what as you will shorting out half of the head unit’s outputs. In this case you would need an amplifier with Diff inputs or a converter to convert the head unit’s outputs to SE

Feel free to comment if you disagree with my analysis and conclusions.


  1. James Hackintosh says:

    I remember your article back awhile ago. Ive come across this again using google somehow. I work for the phone company , as a certified engineer for the CO and outside cable plant. Besides Dense Wave Division Multiplexing on our lit single-mode fiber runs… (many colors of light on 1 fiber), We use TP allot. Actually Our entire cable plant is mostly Twisted pair and it has to be. I also own a home and ive used the stuff i bring home from work for my projects, after seeing the cheezy aftermarket audio cables for car stereo installs, I got curious. I could see it was TP in there… clear see through insulation. I ran some from a composite video cam directly to my dvr with much success so i thought , okay.. this works without a balun… lets try my audio gear. I ran from my Dolby Prologic II unit… a 40 foot single pair of TP to the subwoofer on the other side of the livingroom. No hum. I ran a 70 ft 2 pair from the livingroom to the bedroom … to see what R and L channels would give… and injected line level audio through it.. no hum. However whenever i hook my tv upto anything audio related…. i get hum on everything . After trial and error… it was a surge protector that caused the hum. Apparently the strip had no ground because of an ikea product that was supplying it with power, internally was not attached to ground (it was brand new)…possibly defective… This resulted in the power strip having an open ground, however it tied all equipment to a common ground wire. Then the LED that tells us that it is protected….. it was dumping 120v AC from the Hot leg… to ground… (the led was attached to ground). This made every device “hot”… and shockingly.

    I threw the surge protector onto the roof. After that, everything worked and i measured 0v across ground. no hum. I decided to replace the ikea fixture . No hum . I left Ground disconnected since i have a surge arrestor at breakerbox. no hum. This is very interesting, i wonder how much defective equipment contributes to hum.

    As of 2013 i replaced all components in my livingroom except Dolby prologic II unit (rack mount) with car stereo amps (solar panels on the roof) in favor of my DC bus. I use telecom grade cat 5 patch pairs (not in a cable)… and lots of them, and run them to wall plates… My 1000ft roll of RG-6 goes unused outside thanks to the TP. Not one device contributes hum, and my system is clean. I haven’t even got to my spectrum analyses Fluke branded tool yet. As it turns out… The build defects into the TP cabling. Its called Graded TP at the factory but nobody seems to know what it means. Basically interference on the pair cancels itself out , specifically 60 hz. I am sure a noisy computer or cellphone might create some noise but i have not heard it yet. The fluke meter injects a signal at one end and measures it at the other. When we run services at our CO to the customer premises we are talking about 24,000 ft at voice grade cat 3 and sometimes 2. This requires a balanced line and a earth bonded metallic foil and steel for our common 1200 pair buried cables. However, in our home, its 40 ft max with most segments under 15 ft. There is noway 15 feet is going to show any discernible effect to audio, video, or even Ethernet… at 15 feet unless a person lived next to an AM radio station. I ran this in my walls. I can fit more TP on a patch panel then i can with coax… and it is how i manage things. Generally , without numbers… my basic home experiment with it was a success. Technically speaking, The Fluke shows everything… every channel that is possible…. every harmonic… reflection… the db loss, TDR… it shows that less twists will contribute loss because harmonics from interference and the signal reflections itself… will distort the original signal slightly, technically. Also 60hz hum is detectible with untwisted pair and increases with footage usually but not by much. With a cat 5 twist…. there is plenty of bandwidth to accommodate a analog signal for 15feet and 40 feet. However i will say this, all of these wires could act like an antenna and draw harmful atmospheric charges close to equipment… if a person lived in canada…(northern lights) sun-solar storm… So that is something to consider. If a person had a tesla coil laboratory setup , TP would need to be in a metal conduit and bonded to earth in the likelyhood of saturation of voltage and harmonics in the air…. What i am trying to say, yeah i get it.. impedance mismatch, and grounded side of a fake balanced pair… ok. However my results at home work. When i tell people ive ran 120 volts AC over cat 5… they wonder how. Our DS1 and DS2 lines are over 200 Volts of Pulsed Digital squarewaves… onto 22 AWG wire… wrap your head around that one.. and in the same cable as 1000 screaming Pots line customers. TP is what we work with, and we find many uses for it.. we can all agree that a coax would make for a good hangmans noose and TP might not match that unless it was a service drop wire..then the Airomid yarn (Kevlar) strength member and 4 copper conductors would work.

    Arguments set aside… -48 volts DC is used allot in telecom. It works really well for POE. Of course its DC, thats why. However We injected “60 Volts AC” yes 60 volts (sixty volts alternating current) over the thing… and got a elevated level 60 hz hum… but surprisingly by half and with some TP we got test yields of 1/3 AC coupling… and under a stationset wire (non-tp) it was there… you could hear it . This test was done with a buttset and talk battery and also with unbalanced linelevel source and receiver. I dont recommed utilizing 120 volts onto a single pair. A sunset telecom set was used in place of the fluke for this test.

    Once i had a problem with a immature neighbor cutting our wires on our house,,, telephone… satellite… etc.. So i ran all my new wires inside… and installed a few “fake” wires outside… that i connected to 240V AC… I actually heard and saw the flash outside… he yelped. turns out that he, was a she. She was cutting our stuff. I didnt press charges, but the police charged her with things…i dont know what ever came of that.

    So you see, i think you maybe writing about this to get page hits . I think it maybe a waste of peoples time . Nothing you say works in the real world tests. I wish i could prove what you are saying, however i cannot. Its seems like a chassis attached rca “ring” and a signal “tip” would yield some bizarre thing…such as noise…. but in my tests the noise is rejected more as the cable is longer and then it reaches a point where the signal itself is rejected at certain frequencies, probably in relation to the twists per inch.

    Interesting stuff.

  2. Tomi says:

    Thank you for your feedback.

    I know that entire telephone company cable plant is mostly Twisted pair and it has to be. I have designed circuits that interface to those. Twisted pair works very well on telecom environment when the equipment that interface to those cables have balanced interface: POTS, ISDN, xDSL, T1, Ethernet, professional audio equipment balanced audio interface, etc..

    If you connect unbalanced signals to those twisted pairs, the performance of the cabling drops considerably: it will pick up noise much more easily (mains humming and other noise) and there will be considerable crosstalk from wire pair to another. Those effects are considerable when using unshielded twisted pair cable (shielded versions work much better with unbalanced signals than unshielded).

    Depends on the case if those problems with this are at acceptable level or at unacceptable level. I know there are cases where things have worked quite well with twisted pair, and many cases where the noise and crosstalk has been a serious issue. I have home experiences where RCA signals directly connected to UTP cable has has been noisy with lots of humming, but when I used balanced connection (professional mixer and/or audio baluns) the UTP cable has worked without noise problems.
    I have personal experiences on using UTP cable on both home HIFI setups and as part of professional audio/video setups…

    I have done several real life and laboratory test with UTP cable and unbalanced interfaces. I know both theory and practice.

    The cable characteristics have considerable effect on noise pickup with twisted pair cable. But also what you connect to the cable ends: balanced connections work always well; unbalanlanced interface connected to twisted pair cable performance depends very much on the conditions (noise levels around the cable, are equipment grounded or ungrounded etc..)

  3. Zack says:

    So how do we know if our amps have single-input or differential-input? I called Alpine for my amps and was advised by the tech that differential input was the way to go. Sadly, however, this key information is not prominently listed in any products I’ve observed, whether head units or amplifiers. Seems like it should be, because it’s obviously important for SQ.

    • Tomi Engdahl says:

      If the manyfacturer does not list if the amplifier is single-input or differential-input, to verifying that fact there are few ways:
      1. Access to circuit diagram of the device (for example access to service manual)
      2. Open the device and knowleadge electrnics designer takes a look at the circuit design
      3. Do a small series of laboratory tests to measure impedances and how the inputs react to differential and common mode signals

  4. Tomi Engdahl says:

    Twisted pair: Good at lower frequencies. Good
    protection to magnetic fields.
    Coax: More uniform characteristic impedance.
    Suited from DC to several hundred MHz.

    J: At a first though grounding the cable at both
    ends would reduce the influence from the
    magnetic field. However by doing so we also
    establish a ground loop that contributes with

  5. Tomi Engdahl says:

    Use a Twist and Other Popular Wires to Reduce EMI/RFI

    A twist. “The Twist.” A twisted pair. Alexander Graham Bell patented twisted-pair wires in 1881. Remarkably, we still use them because they work so well. In addition, today by harnessing incredible computer power in field-programmable gate arrays (FPGAs), circuit simulators, and filter design programs we can make a twisted-pair wire even more useful in data communication.


Leave a Comment

Your email address will not be published. Required fields are marked *