The KiwiSDR 2 online store is open for orders! Please visit

Combining/Diplexing Antennas to Single Kiwi

I find that the higher HF reception is lacking, and would like to try a combo of a separate high dipole and preamp for 15 MHz and up. I'm currently running a Wellbrook loop.

Anybody working on that here? It seems that if the performance of the loop is such that there is a natural rolloff at that range, that some filter/combiner design where the loop is low-passed at that frequency and a 2nd antenna is highpassed at about the same frequency? This may also be useful for folks using a Loop-On-Ground (which I may try as well, in the effort to reduce a very local noise source).


  • edited August 2019
    I did try it using a high pass filter and simple t-piece. I generally have to reduce the signal from my active loop antenna anyway so figured it was worth a shot as I had the bits to hand.
    I know this is not really what you are asking but my experience was that where the two antennas coverage overlapped the signal did not seem diminished (very short term test!!) and up the higher end there was an improvement. One slightly unexpected result was that the worst ear crushing rough noise I was getting at the time was slightly greater but audibly softened by the slightly out of phase second signal.
    The "High pass" I used was simply a SMA-SMA bias Tee (with no power connected) with a frequency range specified "from 5MHz", that seemed to roll off nicely under 10MHz.

    As my QTH is more suited to magnetic field rather than electrical antennas I did struggle to reduce the extra interference introduced by the vertical (and it's coax pickup) and that alone stopped me spending too long on it. At another location I'd give it more of a go, maybe even trying a shorted quarter wave stub to band pass just 10m (for example).
  • edited August 2019
    I have been working on exactly this for some time. I started with a variety of broadband approaches several years ago. A prototype was published by W6SFH in QST a couple of years ago. I've visited both short dipoles and loops as well as various FET/BJT approaches such as several iterations on the Trask design.
    My target has been to get to or very close to the "quiet rural" ITU noise floor with an antenna (system) that can cover the range of the kiwi. It turns out that with small antennas the limiting point is the same as that of the kiwi - upper HF. Although 'radiation resistance' and the associated signal&noise levels are dropping with increasing wavelength, the noise floor at these long wavelengths is rising faster so that allows a very small antenna to work well from LF-midHF. The problem is that to get to the ITU limit above ~14 MHz (or so) one needs to get rather close to KTB. This is not presently possible with available devices. For a dipole, the device that can provide the input Z (low capacity, high R) to avoid mismatch loss can not simultaneously handle the broadband onslaught of highlevel signals and very low noise floor. I have been able to achieve this with an OpAmp design but only VLF up to midHF.

    Lately I've been working on a hybrid passive/active approach just as you suggest. But to make this system work with a passive over an [octave] requires some complexity. The Q of a single dipole (around 7) is too high to avoid mismatch loss (see paper by Fano) that raises the effective noise floor. But there are techniques that can do this. For example, a biconical or modified-biconical can be built from wire and give only ~2 dB mismatch loss from 14-30 MHz (and above). It also appears to be possible to couple an active dipole with a passive one to get partial-hybrid, though not over an octave. With a suitable diplexer at, say, 12 MHz, one can combine the best of the active antenna performance with that of the passive and get to a suitably low noise floor.

    Presently I'm trying to understand the effects of ground/earth proximity and polarization better. I have kiwis outfitted with both active antennas and multiple V/H passive dipoles (with LNA) available should you like to investigate. The active dipoles are on N6GN/K3 and N6GN/K2 (a remote site) while orthogonal 21 MHz dipoles are at N6GN/K. All three have antenna switches so you can play around, though be cautioned that I'm always messing with things so there's no guarantee what you'll find. I try to keep the Antenna labels current but don't always succeed.
    Write me if interested in any particular part of this.

    Glenn n6gn
  • My own goals are more modest and in line with what you say in the 2nd paragraph. My location isn't burdened with many high level signals in that range (maybe CB and business band on the upper end, at most?). I'd be happy to get something for the upper amateur HF bands, more consistent WWV signals and the occasional broadcaster. My previous Delta Loop did fairly well on these but also was plagued by local interference from VLF to 8 MHz. Its been some time since I've worked these equations so if you've some references that'd be helpful. I wonder how the calculations for a 'terminated' dipole would work out like a T2FD, when married to a suitable LNA, like an RPA-1?
  • Speaking of the T2FD antenna, I just put one up the other week. I used to have one several years ago, but it was damaged after some winter storms. This one is 58 ft long, the spacers are 20" pieces of 1" PVC pipe. Fed with a 9:1 balun into 75 ohm RG6 coax. I did some experimenting with the tilt angle, the gory details of construction as well as some tests/measurements are here:,56285.0.html

    Summary, I ended up with about a 30 or 35 degree from vertical tilt angle, which seems to produce the strongest signals as measured by several MW stations (I chose MW vs HF because I wanted relatively fade-free signals to work with, results could be different on HF I suppose).

    I am really impressed with the performance, and I have a bit of an antenna farm as it is (670 ft horizontal sky loop, 500 ft and 300 ft beverages, several dipoles). It's a solid performer from MW through 10 meters. I do not have it on my KiwSDR because I have the 43/48 meter folded dipole connected to it, as that antenna produces somewhat stronger signals on those two bands as you might expect, and the main purpose for my KiwiSDR is for people to listen to pirates on those two bands.

    On MW, it picks up a semi local pirate station that is about 10 miles away on 1620, I suspect it runs a few watts. Better reception than either my large sky loop or beverage. On the CB band, it has significantly stronger signals than any of my other antennas, even the 10m dipole.

    I do live in a fairly rural/RF quiet area.

    If I were to put up a second KiwiSDR, this is the antenna I'd connect to it.
  • As I usually think about it, the problem is about noise/QRM floor. For a more-or-less omnidirectional antenna, an antenna's aperture is roughly the same, a short dipole or monopole over perfect ground 'catches' the same power irrespective of its electrical size. A one inch dipole and a full half-wave dipole on 3 MHz each have nearly the same pattern and *if matched* deliver the same power. The rub, of course, is that small antennas can't be well matched because their impedance is too awful, at least short of superconducting materials.
    An antenna that covers the Kiwi's range is going to be small at some wavelength, thus the question of delivering the propagated noise floor to a Kiwi is, I think, a good one. If one can hear the propagated noise, unhindered by system noise, IMD, near-field or other ingress such as common mode noise currents, any antenna will do about the same. But as an antenna gets small the only way to get to the SNR produced by a distant station or to the propagated noise floor which is limiting that SNR, is to couple 'well enough'. It turns out at low frequencies that the only way to do this is to build a 'voltage probe' - a preamp/receiver with such a high input Z that there is low mismatch loss. Trying to conjugately match (power match) to a small antenna isn't possible as evidenced by the roll-off of even large half-wave dipoles by LF or VLF when terminated in 50 ohms.

    But interestingly, the target noise floor (using ITU as a guide) is climbing extremely quickly as one goes lower, so much so that even a short antenna along with a suitable and achievable input Z can reach it without device noise dominating. For us, the problem of short antennas is the other end of the spectrum - mid to upper HF. ITU shows galactic noise dominating above 20 MHz, maybe above mid-Kiwi-range depending where you are and modulo seasonal and diurnal variations.

    So to make a broad band antenna system we can use a short active antenna from 10 kHz to, say 10 MHz but can't match well enough above that to reach the galactic limit. Even a 2x1m dipole has low enough 'radiation resistance' that the target levels are too low compared to the noise of even very good devices. We've just added super-low noise to the high impedance requirement. But it gets worse in most localities. Once we've created a pretty-broad 'voltage probe' for e-fields most places see hundreds of millivolts, if not volts, peak-peak signals. These big guys must not create inter-mod that is large compared to the shrinking voltages of a short/small antenna. And it's not just IMD, all the other ingress mechanisms get more relevant compared to these tiny signals. CM currents in particular can easily dominate. Fortunately near-field coupling falls of with distance much faster than inverse-square so can perhaps be mitigated by proper antenna location, even in residential settings.

    It's for these reasons that I've gone to using a very wideband OpAmp that has both high impedance input as well as low noise and low distortion (feedback provides this until one reaches a slew-rate limit, at least). But even with the best I can find, my 2x1m active dipole only gets to mid-HF. Above that I need to go back to a passive antenna BUT I have to match to it with low enough mismatch loss that I don't make the system noise, IMD and the ingress mechanisms dominant. To do that, I'm looking at maintaining balance via a dipole and 'excellent' balun (40-60 dB of balance may be required, more than is commercially available in broad band baluns) an LNA (which can be at 50 ohms) along with structures which have low enough mismatch loss that the galactic limit is still achievable.

    This gets me to where I am, an dipole active antenna (AA) for the low half and a biconical-like structure for the upper part. A four-leaf biconical, roughly 6m tip-tip, built on a insulating vertical mast and made from wire can be matched with ~2 dB of mismatch loss from 14 MHz through 30 MHz. Using a diplexer to combine this with the existing AA can get me coverage over the entire range. Any other option which can't be matched similarly well to the radiation resistance (not the losses in the antenna or earth), while maintaining acceptable directionality doesn't seem as attractive a solution. I'm not stuck on a biconical, I just want good match everywhere over 10-30 MHz and something reasonably small and HOA friendly. Electrically small antennas, either probe-like or loop can't do this.
    Made from AWG #28 something like this can do it

    At least that's what I'm thinking today.

    Glenn n6gn
  • Hi Glen,

    You should be able to make the whole structure active and a lot slimmer.

    The problem with a lot of E-Probe antennas is that they tend to use a very small plate antenna, which only has a capacitance of a few pF. If you use a fat cylinder about 1m long to give a capacitance of around 30pF. The HF performance becomes considerably better, as the ratio of antenna element capacitance to amplifier input capacitance is much improved.

    The downside is that you need to use an amplifier with good IMD performance as the overall signal levels are much higher. I have used the Trask fig5 amplifier with a brass mesh tube to increase the antenna capacitance quite successfully, and until a few weeks ago (most antennas have now been disconnected prior to a change of location) it was one of the switched antennas that was available on my public KiWi

    The bicone angle defines the feed-point impedance when used over it's designed frequency range, i.e. where the elements are roughly equal to 1/4 wave long at the lowest design frequency. If you make the taper really narrow and long you end up with a feed-point impedance in region of the hundreds of ohms.

    Both of these factors would seem to lend themselves to the construction of an active antenna of the type you describe.

    Here's a couple of patents that make interesting reading.

    Incidentally I noted Chris's comments about my TC2M broadband vertical antenna on the HF Underground forum. The requirement for a decent ground plane was mainly associated with it's overall efficiency when used as a transmit antenna (which was the focus of the article the quotes were taken from). The one I had running on my KiWi only had a minimal set of radials, but it still performed well as a receive antenna where the delivered Signal to Noise rather than absolute gain is the main consideration.

    Adding an effective ground plane to any vertical antenna will improve its efficiency, gain and low angle radiation, but it shouldn't be considered a show stopper, as a vertical may be easier to install in certain locations, especially if you don't have an existing or handy support structure for something bigger.


    Martin - G8JNJ
  • Martin, how is the TC2M with respect to local noise pickup? My experience with verticals in the past was that they seemed to have higher noise levels, one reason I have shied away from the since then.
  • >
    >how is the TC2M with respect to local noise pickup?

    If you decouple the feed line they are not too bad. It's a pity the antennas are currently dismantled, otherwise you could have compared them yourself. But generally speaking the TC2M was the best overall performer of all the antennas I've tried so far, and quite a few people have go on to build their own versions after using my KiWi.


    Martin - G8JNJ
  • edited August 2019
    Martin writes
    "You should be able to make the whole structure active and a lot slimmer.
    The problem with a lot of E-Probe antennas is that they tend to use a very small plate antenna, which only has a capacitance of a few pF. If you use a fat cylinder about 1m long to give a capacitance of around 30pF. The HF performance becomes considerably better, as the ratio of antenna element capacitance to amplifier input capacitance is much improved. "

    I don't think this solves the problem. Fundamentally the radiation resistance of a short antenna, the real part of the impedance presented goes as length squared. Adding inter-element capacitive reactance, sometimes called 'dead capacitance' doesn't change this so the active device and ingress noise still has to be very low noise compared to that low signal voltage. It turns out that the input C of the amplifier is almost the least of the worries as long as one stays below a few pF which generally isn't difficult with FETs or FET input OpAmps. A short 'probe' only intercepts a short region of space. The best one can do is 'catch' that without mismatch attenuating it. Similarly the input R of the stage is generally not a cause of noise above a pole that is down in LF. As I mentioned, the ITU levels are so high at VLF that they swamp what would otherwise be a problem.

    The biconical patents you reference have much in common with the Flying Antenna I built and previously wrote about. In that case it was a discone - a biconical in reflection - with an extension. As shown it could give multi-decade matched performance, 400 MHz to 10 GHz, with only a small 'glitch' in the pattern in one or two regions.

    I've not yet found a small solution to VLF-HF coverage down to the ITU target in a single active design, thus my interest in a hybrid and broad band matched antennas.

    Glenn n6gn

    [Edited to fix the URL to "Flying Antenna" video]
  • edited August 2019
    >Adding inter-element capacitive reactance, sometimes called 'dead capacitance' doesn't change this

    Agreed that is an issue, and a tube doesn't give quite as much improvement as you would expect.

    A wire frame inverted 'pyramid' dimensioned to give the same value of capacitance as the fat tube, gave improved performance over the tube on the HF bands, however it was a bit unwieldy by comparison.

    Other shapes may provide better results for a similar outside perimeter.


    Martin - G8JNJ
  • I was involved in the design of a similar antenna 25 years ago..... I'll see what details I can find.
  • Hi Glen,

    I've just realised that we had covered some of this ground before, in an email exchange back in December last year.

    My apologies for revisiting some of these themes.

    I think I'd better take some time to re-read the notes that I made at the time :-)


    Martin - G8JNJ
  • Martin,
    FWIW, I played a little with your suggestion of matching to a 'slimmer' antenna at a higher impedance. Broadband matching is a never-ending story. I spent some time trying relatively simple approaches to match the S Parameters that NEC2 says a simplified ~7m tip-tip 4-leaf biconical and an 8m tip-tip dipole exhibit. Each was made from small gauge wire so as to be fairly HOA friendly. Doing my best for a couple of hours, I couldn't find a way to get near the lower mismatch loss of the biconical with a dipole. No doubt moving the dipole to 'somewhat fatter" will move in the direction of the performance of the biconical but it also approaches, if it doesn't exceed, its dimensions.
    Here's an example simulation output comparing the two:

    There are endless possibilities to try and perhaps someone else will take up trying others but at this point, given the simplicity of the approach and the apparent success, I think I'm inclined to build a hybrid using the biconical-type approach and the active antenna I'm already using. I think that with readily available components, probably a GALi+ sort of approach for the LNA and a diplexer, that the two can push down pretty close to the ITU lower noise limit.

    I still have to come up with a better understanding of the differences I see between low horizontal and low vertical dipoles. Why the two can produce such similar SNRs but 10 dB different N is still a puzzlement. Perhaps it just happens that low take-off-angle improvement of the vertical counteres the low angle propagated (inverse-square) noise increase. I need to take more data.

    Glenn n6gn
  • Hi Glenn,

    Hmm, that's disappointing and somewhat unexpected, as my understanding is that some skeleton Bicones used for EMC measurement purposes have a 300 Ohm feedpoint impedance, and have an impedance transforming Balun built into them, as it's necessary for accurate calibration.

    Unfortunately my old EMC antenna catalogues that contained some detailed drawings are packed away at the moment, and I can't easily access them.

    However I did find this in the 'Handbook of Antennas for EMC' by Thereza Macnamara P145.

    Any chance you could email me the antenna NEC files so that I can have a play with them ?


    Martin - G8JNJ
  • Hi Glenn,

    > probably a GALi+ sort of approach for the LNA and a diplexer, that the two can push down pretty close to the ITU lower noise limit.

    There are at least a couple of Mini-Circuits devices that have similar and usable IMD performance, but there is a difference between them

    Gali-74 has a fairly constant 50 Ohm input impedance below 50MHz.

    PGA-103 has a gradually rising input impedance below 50MHz and is about 1K ohm at 1MHz.

    I have successfully used PGA-103's as the basis of HF active antennas, and this characteristic may be advantageous in a combined bicone & active dipole design.


    Martin - G8JNJ
  • Thanks for the biconical reference, I really haven't looked at any yet. The issue isn't (just) the feedpoint impedance, essentially I think it is the Q of the structure. To get a reasonably good match over a broad band the issue is not only the real part but the ratio of reactive/real. On a Smith chart this is rotating and it is this sort of circumstance that Fano addresses in his famous paper.
    The simplified wire-biconical I modeled and used to generate impedance data, the one I showed higher up in this thread, was made from closely spaced parallel conductors in order to drop the Q over the octave of interest. Those S-parameters were exported to QUCS for modelling and optimization and how I did my admittedly quick comparison with a thin dipole. A simple matching network

    produced the comparative mismatches shown previously.

    I'm not expecting any severe problems with the LNA since it appears that I can pretty easily get good match referenced to 50 ohms. No doubt using quiet devices that run enough current and a structure with good symmetry/balance to help with IMD is important. Because I don't need to go down to MF or LF, as I did with the OpAmp approach, using ferrite is fair game and simplifies things.

    I'll mail you (and anyone else that wants it) the .NEC file I used for modelling.I'd post here but the forum seems finicky about file types.

    Glenn n6gn
  • I'm prepping for new build year and want to tackle this high HF problem in the spring. Had anybody modeled these approaches over average ground or built one for evaluation? If not, I am interested in trying it - even the vertical biconical - as I'm getting much more interest from CBers and would like to improve performance for that.
  • edited December 2019
    my system uses a bi-conical of sorts, the pix will come up when you access it. I have a xfrmr, HPF, and LNA at the feed. I have AM Notch filters downstream.

    The "dipole" is balanced so I use balanced stuff to the preamp, a Dallas Lankford based Norton that you can build or buy from N4CY. The flow is:

    Modified OE-254 antenna -> 16:1 73-ferrite isolation transformer -> 3 dB attenuator -> 35 MHz LPF -> LNA -> feedline -> 2-way magic-tee splitter -> Nooelect AM distill band filter to each radio.

    Everything is HB escept the mechanics of the antenna, the NYCY LA and the AM Distill

    I live in an RF swamp and even have severe OV from TV/FM TX < 1 mile away. I need to change that 35 MHz LPF to something better when WX permits
  • I have also found that I need to use larger core than you might expect for RX due to the high levels of energy here in the swamp. I use 1/2" 73 binocular cores.
  • N0EMP built a prototype of the modified biconical that I suggested and previously modeled with NEC2. We measured it in situ with the bottom less than 1m above earth and I then built a simple matching network intended to match it over 12-30 MHz. Here are the 4NEC2 model vs. measured results:

    I think the red was modeled and blue measured but as you can see it doesn't make much difference because it works as expected. Mismatch loss over the range is low and it should be a good antenna for both broadband transmit as well as receive over almost a 3:1 frequency range.
    This was measured over relatively flat, snow covered and slightly muddy earth while the NEC model was for any of the 'Real Ground', "Minninec Ground" sorts of models, it didn't matter much which was used. Pattern varies somewhat with frequency but here's what the model says for 20 MHz

    which makes it appear to be a fairly good DX/low-angle antenna as was expected and intended for this frequency range.
    I am still concerned about the imbalance caused by mounting it too close to the earth since that will generate CM currents however because it is a full size passive antenna, as contrasted to a short dipole mounted high (which would have a worse CM/differential current ratio) I think it may be OK. Also because it is passive and low impedance, one can not only match it to 1:1 easily for transmit but also filter as needed to protect against alias images and any monster signals that the bandpass nature of the design doesn't do natively. Much below 10 MHz the structure looks rather like a high pass filter due to mismatch loss, by 5 MHz it's already down by >30 dB (beyond limits of the graph)

    Tip-tip it is around 6m long and perhaps smaller than many other antenna methods. I think Martin has a design that has slightly more bandwidth but is larger. This one was built around a telescoping fiberglass mast with another goal of being relatively HOA friendly It's not self supporting but one set of guys may be enough in most environments.

    When the snow/ice subsides so we have access, we may put it at the remote kiwi at n6gn/k2.
    We have 7 50 kW class NIST transmitters completely LOS at <20km to deal with so this may require some additional filtering.
    Glenn n6gn
  • edited December 2019
    My modified OE254 is neither a true biconical nor real small, but is compact and works well 630-10M. The picture below shows it during testing, it has since been raised up 8 ft. higher.

  • Can you provide S-parameters for LF-through HF for that antenna as mounted? I have a hard time believing that it 'works well' at the low end since it is too small to have good efficiency against real world materials/grounds. It is very likely just fine for receive but of very low efficiency with any sort of broadband matching structure below 40m or so. What is the total length?
  • edited December 2019
    no parameters, sorry. 3X8ft elements top and bottom. It is roughly the same as my Pxel loop on 475 KHz
  • Too bad. Even the results from a nanoVNA measurement (which are generally pretty good) would allow a pretty good estimate of broad band efficiency and give an idea over what range in the spectrum it might be sufficient to get down to the ITU noise floor in a wide band matching configuration.
  • Does anyone has a Pixel Loop characterized? That would be easy to compare, even in the snow and cold....
  • edited December 2019
    The problem with comparing with a small antenna is that the real part of the radiation resistance is swamped by the antenna material, matching loss and on receive, preamp noise and ingress, thus efficiency isn't as easily deduced. The (small) Rr is calculable but not very useful in determining loop performance since that is dominated by other factors. Like a short dipole or mono-pole, it doesn't attempt a broad band match.

    But a larger antenna such as you describe attempts to be a matched structure, the measured S11 can be useful for determining the utility in a broad band situation. ITU noise power (our target) sourced by the real part of the measured impedance is a known so given that impedance the mismatch loss in getting to [50] ohms can be known too. This gives a way to know how close to the propagated/galactic noise floor such an antenna can get us.
    Any antenna under a few wavelengths in size produces the same SNR - at 160m a 1cm dipole and a full half-wave dipole each have about the same aperture- but getting to that and transferring it to a 50 ohm receiver over a broad range of frequencies is the problem. For a physically small antenna the matching problem is prohibitive.
  • I'll revisit antennas in the spring! While it is unseasonably warm here right now, it's also rainy. Snow will follow....
  • @n6gn - what are the construction details on that antenna? I've got fiberglass ready for this. I was thinking of using a combiner and a SHP-20+ on this antenna, with the other port for the Wellbrook.
  • @w1euj It's a four leaf subset of a biconical as shown near the beginning of this thread. I've not written up a construction article for it. Perhaps after we've put it at the remote site, maybe in conjunction with an existing active dipole to cover LF - 10 MHz I'll publish something.
  • aww heck, just the dims on the overall length, hip width and hip to tip length and I'll take it from there.
Sign In or Register to comment.