ad: hrd-2

Twisted Pair Feedlines - Some Practical Answers!

Discussion in 'Antennas, Feedlines, Towers & Rotors' started by G3NJV, Jan 18, 2015.

Thread Status:
Not open for further replies.
ad: L-HROutlet
ad: l-rl
ad: Left-3
ad: abrind-2
ad: HRDLLC-2
ad: MessiPaoloni-1
ad: L-Geochron
ad: L-MFJ
ad: Left-2
  1. G3NJV

    G3NJV Ham Member QRZ Page

    Over the years many QRZ threads have visited the topic of twisted pair feedlines. Few of these threads have provide really informed answers. I have some practical experience and results I hope will prove useful.

    It all started with summer games testing Vertical versus Delta Loop versus Moxon on the back lawn. The Moxon won hands down. My 17m Moxon was designed with Moxgen and made with wet string and bamboos cut from my garden here in extreme south west England, where a brisk westerly wind coats your teeth in salt from the nearby ocean. All the following information relates to 17m but you can use whatever frequency you want in the calculations.

    My Moxon was made to "fly" suspended by a cord under some trees in my garden. The feedline hangs down from the driven element and weight is a big issue, both for feeder and avoidance of putting a balun at the feedpoint. I used 1 square mm PVC insulated wire with OD = 2.24mm to make twisted pair balanced feedline. Two inches from the antenna socket of my rig I put the feedline 4 turns through 10 stacked ferrite toroids to make a choke balun. SWR was 3:1. The IC-736 tuner coped and I had good contacts.

    Later, investigating the SWR, I realised it was caused by the feedline not being 50 ohm. Assuming the Moxon was its designed 50 ohm the SWR suggested the feeder was maybe either x3 (150 ohms) or divide by 3 (17 ohms). It didn't really matter, so long as the feeder was electrically a multiple of half waves long the antenna impedance could be transformed accurately to the rig.

    To get the feedline to a multiple of half waves at 18.13 MHz, the first step was to accurately measure its physical length and do a simple calculation to find the frequency where the feedline SHOULD be one halfwave. Then disconnect the Moxon and replace it with a 50 ohm load. Tune the rig down (a long way!) to find the frequency where SWR dips close to 1:1. Velocity factor is calculated from f-measured divided by f-calculated. In my case it turned out around 0.63. Using this velocity factor, calculate how long electrically your feeder actually is at 18.13 MHz. Add or subtract length as convenient to achieve an exact multiple half wave length, where SWR dips close to 1:1 at 18.13 MHz and rises either side. I achieved this on my 50 ft feeder by adding another 62 inches. Reconnecting the antenna gave SWR 1.2:1 and some good DX without needing the tuner. As an added bonus the SWR was near 1:1 @ 28.5 MHz also!

    I later found an online impedance calculator which gives a figure of Zo = 105 ohms for my wire dimensions.

    I was happy until I tried out my newly restored linear with a pair of 3-500z on the Moxon on 28.5 MHz. The tuning and loading drifted all the time I tried adjusting it. Something was changing. I stopped and checked things. The feeder was burning hot 8 feet from the amp - so hot it had fused the two twisted wires together. The choke balun was cold, it was the twisted wire feeder that was heating up.

    Further use on 17m again gave noticeable heating of the feedline.

    I broke the system down and did some tests. Even if your feeder is completely unknown impedance and velocity factor, it is still easy to measure its loss. As described above, simply terminate the feeder in 50 ohm and adjust frequency until low SWR close to 1:1 is found. Then set the transmitter to give a known input to the feeder and measure what comes out the far end. In my case putting 30 watts into the feeder @18.13 MHz resulted in 17 watts at the far end, a 2.5dB drop over 54 feet. This loss of 5dB per 100 feet is VERY BAD! Removing the choke balun had no effect. Untwisting perhaps 95% of the twists I had originally put in had no effect on loss but a big effect on velocity factor.

    Using a bench power supply to put 2 Amps DC through each leg of the feeder in turn I used a DVM to measure input voltage actually on the wire ends. This gave accurate DC resistance of 0.267 ohms each leg - not enough to account for the power loss.

    Twisted pair offers huge advantages as a balanced feedline which is easy to make and use. The constraints are that it is almost impossible to get 50 or 75 ohm impedance but 100 ohm is quite easy. Loss depends on copper size and material used for the wire insulation, following the same established rules as coaxial cable. Air, foam and PTFE give best results. On wire only PFTE (Teflon) is common. Constructed of commonly available 1 square mm stranded PVC wire, it works great for reception giving approx. 100 ohm impedance with a velocity factor of 0.6 ~ 0.7 depending on how hard you twist it. You can expect about 1 ohm DC loop resistance over 100 feet of feeder accounting for perhaps 5 ~ 15% RF loss. However the measured loss of 5dB @18 MHz per 100 feet I experienced is much much worse than the copper loss would suggest, worse than even small coaxial cable and no good for transmitting. This loss can only be accounted for as loss in the PVC insulation. The exact same copper core with Teflon (PTFE) insulation would probably give excellent results (but here in the UK - is stunningly expensive). Use it if you can get it! Wires harvested from Networking CAT cable for higher data rates may also have superior low loss insulation ideal for reception or QRP transmitting but as this is small signal wire, be careful about transmit power. Remember two 100 ohm twisted pairs connected in parallel will make a 50 ohm line. Also check out how to make 4 wire twisted quad line.

    Of particular interest in my results is a method of measuring loss in "any" impedance twisted pair transmission line and attributing this to copper or dielectric loss. As the loss in my feeder was unchanged after removing 95% of twists in the line, it suggests that even if all twists are removed and the wires moved slightly apart the loss may well remain constant. If this is the case, only complete removal of the PVC from the dielectric path would lower the loss - leading to classic open wire feeder with copper conductors and air insulation. Of further concern is similar loss caused by PVC insulation on antenna wire which (unlike transmission line) is impossible to measure. It has been widely suggested that PVC insulation may be introduced deliberately as dielectric loading to shorten an antenna. Finally, it is worthwhile to remember that increased loss in either an antenna or transmission line almost always results in what appears to be an improved SWR.

    In summary, my PVC insulated twisted pair feedline is both simple, lightweight and practical. However the PVC causes unacceptable high loss for transmitting. Previous QRZ threads have posed largely unanswered questions and few practical results about using twisted pair feeder. Noted is the poor performance using rubber zipline but also the good results with Teflon insulated wire. Bearing in mind that the conventional 50 ohm impedance line is almost impossible to make in a twisted pair, successful use of twisted pair feedline will simply depend on good RF dielectric material and good conductor size - no different to any other type of transmission line. It would be very interesting to hear from anyone who can carry out similar loss tests using PTFE or CAT type wires.

    73's Paul
    Last edited: Jan 18, 2015
  2. N7EKU

    N7EKU Ham Member QRZ Page

    Hi Paul,

    Thanks much! Your post is very interesting to read :)

    An additional thing to note about using feedline this way, is that rain or ice will have a big effect on its velocity factor. So it would only be good for dry weather.

    One thing you could try is certain types of CAT5 or 6 network cable. Some of that is made with FEP insulation which is pretty similar to PTFE. It should have better dielectric properties and should also shed moisture better. You can tell if it has this kind of insulation as it is pretty resistant to burning when touched with a hot soldering iron. CAT cable is solid conductor though, so it may not stand flexing in the wind too well.


  3. G3TXQ

    G3TXQ Ham Member QRZ Page


    A few observations:

    1) You quote the DC loop resistance, but say nothing about the RF resistance. At 18MHz it will be about 20 times the DC resistance, although that's still not enough to account for all your losses.

    2) Your line was operating under mis-matched conditions, so there will be some further small loss from that mechanism.

    3) Did you track how the velocity factor changed as you untwisted the wire? That's usually a helpful indicator of how much of the field is contained within PVC rather than in air.

    4) I'm very sceptical about your suggestion that removing all the twists and moving the wires apart may cause the losses to remain constant. That's not been my experience, nor the experience of the many folk who have built open-wire line with PVC insulated wire and not experienced 5dB/100ft loss at 18MHz! For the same reason I'm sceptical about your hint that there may be significant losses from using PVC insulated antenna wire.

    If you've not already found the page, I reported some TL loss measurements under different conditions here:

    Steve G3TXQ
  4. KH6AQ

    KH6AQ Ham Member QRZ Page

    Paul, that is interesting data. The matched copper loss at 18 MHz, not accounting for proximity effect, works out to be 1.4 dB/100' with the 2:1 VSWR adding 0.3 dB for a total of 1.7 dB/100 ft. The link below suggests a fudge factor of 2.3 for proximity effect which would make your measured loss 1.7 dB X 2.3 = 3.9 dB. So, perhaps 4 dB of the measured 5 dB is copper loss.

    A way to test this is to measure loss at two frequencies. Copper loss is proportional to the square root of frequency while dielectric loss is linearly proportional to frequency. If the loss is all copper the loss at 29.7 MHz will be 6.4 dB and if the loss is all dielectric the loss will be 8.3 dB.
    Last edited: Jan 18, 2015
    AJ6KZ likes this.
  5. G3TXQ

    G3TXQ Ham Member QRZ Page

    Loss measurements made this afternoon on a PTFE twisted pair:


    Wire was 2x38ft of Farnell part 118-4148. Centre conductor 19x0.2mm (18AWG); Jacket OD 2mm

    Zo varied between 140 Ohms and 170 Ohms from 1MHz to 50 MHz.
    Vf was 0.83

    Losses up to 30MHz are totally accounted for by copper loss.

    Steve G3TXQ
  6. G3TXQ

    G3TXQ Ham Member QRZ Page

    For comparison, here are the copper losses calculated for a TL with 18AWG conductors and Zo=165 Ohms:

    10MHz: 0.48dB/100ft
    20MHz: 0.69dB/100ft
    30MHz: 0.85dB/100ft
    40MHz: 1.00dB/100ft
    50MHz: 1.13dB/100ft

    Steve G3TXQ
  7. KM1H

    KM1H Ham Member QRZ Page

    Also dont forget that both 72 and 300 Ohm PVC twin lead was popular in the 50-60's and not subject to high losses. I used a 40M 72 Ohm twinlead fed dipole on 80 thru 10M at 50-100W initially and later about 700W CW and AM carrier (PP 250TH's with 810 modulators) on 80/40/20M; nothing melted. Also a home brew 4 el 6M and 5/5 2M fed with 300 Ohm that both worked rather well.

    Twisting the feeder helps balance the line but only a twist or two per foot is needed on the higher HF frequencies.

    With Beverages on 160-30M two twists per 10' is sufficient when using various types of balanced line for 2 wire direction switchable versions; I use 5 of these antennas of 500-750' length that perform rather well running thru the woods here under all weather conditions. The line impedance is about 140 Ohms using military telephone wire of multi conductor siamese pairs; not WD-1A.

    With OWL spacing over about 3" the balance effect is gone which goes right back to the perennial arguments of 600 Ohm line not being truly balanced and radiating some amount of RF.
    But that is another subject for another time.

  8. KH6AQ

    KH6AQ Ham Member QRZ Page

    G3TXQ, excellent and as you know your data shows no proximity loss and the loss vs. frequency shows copper loss to be the dominant loss mechanism. This points to the PVC twisted pair as having high dielectric loss. PVC twisted pair loss data at 18 and 54 MHz would tell if this is correct.
  9. W5DXP

    W5DXP Ham Member QRZ Page

    For what it's worth, here are the loss graphs from my 1957 ARRL Handbook. 14-023 was the Z0=75 ohm parallel feedline recommended for transmitting, 14-080 was for receiving only. For HF operation, 214-023 was lower loss than RG-58. The bottom 'A' line is for Z0=600 ohm open wire feedline. 14-079 is Z0=150 ohm. 14-022, 14-056, and 14-076 are Z0=300 ohm lines. Insulation for the parallel lines is polyethylene.

  10. W0LEV

    W0LEV Ham Member QRZ Page

    About a decade ago (maybe more), our small group of hams at HP (the "real" HP, knot of recent) decided to entertain themselves on a Friday afternoon. We had contemplated the idea of a cheap and easy feedline embodied as a twisted pair. We ordered some good grade #18 twisted teflon-insulated wire for the experiment. When the wire arrived (we went for 1000' - we all chipped in on the cost) - we made 25 feet of tightly twisted "transmission line". If I remember correctly, the characteristic impedance came out around 90 or so ohms. The length looked good on the VNA, so one of us made a 50-foot length and installed it on our home dipole array. Same experience. Looks and behaves well at low powers, but won't tolerate power - even witnin a matched system. I still have what's left of the roll of wire.

    Theory sez TP (twisted pair) transmission line should work well in to a balanced load driven differentially. That which we built worked "OK" up to about 50 MHz at low power. Above that, the losses became to great to consider serious use, even for receiving.

    Now to the digital world: Ethernet started out using coaxial cable. Radiation was NOT a problem. You could operate an AM or FM radio in a building with installed Ethernet. Coax was expensive. The digital world went to TP. Radiation became a MAJOR problem for anyone attempting to use a radio inside a building outfitted with even "shielded" TP (never mind the digital incantation of the "shield"). Of course, the digital embodiment of differential drive is not true DM since each phase references "ground" reference of the digital circuitry. I've found that withn 5 to 8 wavelengths of "digital" TP, the "signal" pretty much becomes common mode and radiates like crazy. We didn't have that problem when Ethernet was in Coax.

    Why does TP look so good in theory, but falls flat in practice? Maybe we need to go back to basics: Mr. Maxwell ? I no longer do that kind of integral calculus in my retired years. Anyone up for a real challenge???

    Dave - W0LEV
Thread Status:
Not open for further replies.

Share This Page