Using the same feedline for HF/6 meters and 2 meters/70cm (with a diplexer)

Yesterday, a work party went to the remote HF station of the Utah Amateur radio club to install a new antenna for the lower HF bands.  At the site was an existing G5RV-type antenna which provided coverage on 80, 40, 20 and 10 meters (more or less) - but it didn't provide 160 meter coverage (mostly useful during winter months, after sunset) and this G5RV was not taking full advantage of the 65 foot tower on site, being anchored near the top, but sloping down toward ground that was also rising to the west.

Figure 1:
The exterior of the completed diplexer designed to allow
HF+6 meters to co-habitate with 2 meters and 70cm on the same feedline.
Click on the image for a larger version.

Taking this antenna design on as a project, former Utahn/club member Mike, WA7ARK, decided to take advantage of his recent research, simulating, and real-world testing of multi-band end-fed half-wave antennas ¹ and suggested a 160 meter end-fed half-wave wire:  If it worked as expected, it would provide useful coverage over the lower half of the 160 meter band, much (if not all) of 80/75 meters, 60 meters and 40 meters:  It may also be useful on 20 and 10 meters as well.

In short, the 1/2 half-wave antenna consists of approximately 260 feet of wire fed at roughly the 0.45 wavelength point by a 1:7 broadband transformer to provide a 49-fold impedance transformation.  When done properly, this combination can provide a reasonably good 50 ohm match on the first 3 or 4 half-wave multiples, (e.g. 160 meters=1/2 wave, 80 meters=2/2 wave, 60 meters=3/2 wave, 40 meters=4/2 wave, 20 meters=8/2 wave, 10 meters=16/2 wave.)  Much above 40 meters, the antenna was expected to lose effectiveness - at least in part due to the higher-order multiples, but also with the transformer "running out of steam" (getting lossier) at the higher frequencies.

But I digress...

No antenna is useful without some sort of feedline.  At the top of this tower was already-mounted a VHF/UHF fiberglass vertical fed with 1/2" Heliax ^(tm) hardline and rather than going through the trouble of running yet another feedline - which would be possible, but probably take more time than we'd have in just one day of our work party - I decided to construct a device that would allow this one feedline - which would be lower-loss than, say, RG-8 style coaxial cable - for both the HF and VHF/UHF antenna:  Because we'd be "force feeding" the antenna at impedances other than 50 ohms to attain greater operational bandwidth on the covered bands there was incentive to use the lowest-loss transmission line possible.


A "high/low" diplexer:

The solution to this problem is to employ a diplexer - a device that will take the signals from a common port and send high frequencies to the second port and low frequencies to the third port.  In this case, I started out with the basic design goals:

• Use an inductor-input low-pass filter with a cut-off set comfortably above the 6 meter band - say at 65 MHz.
• A capacitor-input high-pass filter with the cut-off set comfortably below the 2 meter band - perhaps 110-130 MHz.

Using the (free!) Elsie filter design program I plugged those values in and did some tweaking, ultimately deciding that an "N=5" Chebychev filter with 0.01dB ripple seemed to be appropriate.  I had to model these filters independently of each other because the free version of the program did not allow them to be bridged together at a common point - but taking the values given by the program and plugging them into "LTSpice" (by Linear Technologies - now Analog Devices) to simulate the combined circuit permitted a more analysis and "virtual" tweaking.

As expected, bridging the two filters at a common point "messes up" the response a bit - but for a circuit as simple as this, some experiment tweaking of values is all that is really needed.  Once I was satisfied with the result, I constructed one of these devices and analyzed it with my (relatively newly-acquired) DG8SAQ VNA to assess both the insertion loss and the matching.

The response for "6 meters and below" is as follows:

Figure 2:
The "low side" response of the diplexer with the VHF+ port terminated.  As can be seen, the insertion loss is below 0.5dB with a "reasonable" match at all frequencies 6 meters and below.  The isolation of this filter at 2 meters and above is well over 45dB.  The spurious response at the top end of this sweep (at around 725 MHz) is likely due to a resonance of the enclosure and has no bearing on its intended use.
Click on the image for a larger version.

At this point I should mention that the need for this filter arose rather suddenly:  About two weeks ago, we had taken inventory of what feedlines were available on the site and knew that we would be "short" a feedline - and in the likely event that we (probably) would not have the time to run a new one, I designed and constructed this diplexer - and a duplicate (one of these is required for each "end" of the cable!) - over the course of two evenings.  Had I more time I'm sure that I could have tweaked values a bit and reduced the insertion loss even more.

Moving the VNA to the VHF/UHF port and putting the load on the HF+6 meter port, I ran another sweep, which looked like this:

Figure 3:
The "high side" response of the diplexer with the HF+6 meter port terminated.  The insertion loss here is actually lower than that on the HF port - at least on the 2 and 70cm bands.    The isolation at 6 meters is a bit over 30 dB, increasing to over 60dB at 10 meters - more than adequate for our purposes.
Click on the image for a larger version.

As with the the "low" side, I'm sure that a bit of extra tweaking would have helped things a bit, but for its intended goal - providing an RF path to VHF/UHF vertical - its performance was plenty good - comparably to a commercially-available device.

Figure 4:
Inside the diplexer.  The diplexer was constructed in a box
that had previously been used for some satellite equipment.  The circuit
itself was built "dead bug" style on a piece of glass-epoxy circuit board.
The "common" HF-UHF in/out port is in the upper-left corner, the VHF-
UHF port in the upper right and the HF+6 meter port in the lower-left.
Note that the leads in the VHF/UHF path are kept as short as possible
with the components laying against the ground plane.
Click on the image for a larger version.

Figure 4 shows the interior of the constructed diplexer, built inside a Hammond 1590D die-cast aluminum case from a discarded piece of satellite equipment.  As it happens, the N-type connectors with attached lengths of UT-141 50 ohm hardline had been part of this same equipment and were put to use for the three input/output lines.  The use of the "N" type connectors were ultimately helpful as they are better-suited for outdoor use as they are designed to be weather-resistant on their own:  The connectors on the outdoor box were sealed with tape and wrap, anyway!

The capacitors used are NP0/C0G type ceramic disk, each rated for at least 1kV (and hi-pot tested to 3 kVAC - a bit over 4kV pk) and the inductors themselves are wound using tin-plated 12 AWG copper wire.  It is expected that this device should be able to handle at least several hundred watts on HF and 6 meters over a wide variety of mismatch conditions and 100 watts on 2 meters and 70cm.

Again, had I more time - and were it absolutely necessary to reduce the insertion loss even more - I would have done more tweaking of the capacitor and inductor values ².  While the insertion loss on the VHF/UHF port is gratifyingly low, it would have no doubt been even lower if surface-mount capacitors and 50 ohm strip-line had been employed.  Finally, this device could have been constructed in an enclosure of about 1/3rd this size - particularly if a circuit board had been made along with a bit of clever arrangement of the components - but I used what I had, in the time that I had.  Because we needed two of these - one at each end of the feedline - I replicated the first and was happy to get identical results.

Diagram:

The schematic diagram of the as-built filter along with some component information is depicted in Figure 5, below:

Figure 5:
The schematic diagram of the as-built filter along with information about the parts used. The "half turn" specified is simply due to the fact that when you wind a coil so that both wire ends point in the same direction, an extra half-turn naturally exists.  The major modification required when the input of the high-pass section was bridged with the low pass section was to change the value of C3 from 33pF to 18pF.
The "nominal" predicted values of the inductors are:  L1=L3=93nH;  L2=192nH;  L4=L5=47nH. In construction, extraneous factors will require a bit of tweaking of these values to achieve best performance.
Click on the image for a larger version.

All of the inductors are wound using 12 AWG tin-plated solid copper wire on a 3/8" (9.5mm) O.D. drill bit as a former.  12 AWG wire is probably overkill, but tin-plated wire was readily available in this size:  As small as 16 AWG wire could have been used - particularly on the "High" (2M/70cm) side with little impact on insertion loss, but since HF energy actually flows through L1-L3, I would not recommend using wire smaller than that:  The use of smaller wire would somewhat increase inductance for the same number of turns so coils using it would have to be modified accordingly.

All of the capacitors are of the low-loss NP0 (a.k.a. C0G) type and no other type of ceramic capacitor should be used.  If desired, Mica capacitors may also be used:  These may offer slightly lower losses - particularly on VHF and UHF - but they are typically much more expensive.  If only low powers are expected (less than 25 watts) then 50 volt capacitors will suffice, but if a full 100 watts is anticipated on HF - particularly if a radio's built-in antenna tuner is used - that C1, C2 and also C3 (which "sees" the HF energy) be rated for at least 500 volts with 1kV being preferred.  For VHF/UHF power levels up through 50 watts or so, 100 volt capacitors are adequate for C4 and C5.

It should go without saying that the junction between J1, C3 and the end of L1 be kept as short and compact as possible, as poor construction practices here - and with the remaining components in the VHF/UHF path - will result in higher losses.  It is recommended that when constructing this circuit on a ground plane such as a piece of un-etched circuit board material (as seen in Figure 4) that all of the VHF/UHF branch components be laid flat, as close to the ground substrate as possible:  If one wishes to etch a circuit board, a 50 ohm strip line  along with surface-mount capacitors of appropriate voltage rating would be preferred and lots of "vias" (or holes+wires tying the back side of the board to the front) should be used.

* * * * * * * * * * * * * * *

In use:

A pair of these devices were installed yesterday at the remote radio site:  One located atop the tower near the feedpoints of the VHF/UHF and the HF antennas and the other indoors, where the feedline enters the building, where the cable from the HF/6M port goes to the HF-6M transceiver and the other goes to the 2 meter/70cm dual band transceiver.

At least with moderate VSWR levels (e.g. up to 4:1 or so) the components within this diplexer should have minimal effect on matching - at least on the lower HF bands.  Because the "new" antenna was designed mainly for operation from 160 through 40 meters, the diplexer should be more or less "invisible" over this VSWR range - which is also about as much range as radios' built-in tuners can typically handle, anyway.  Although no scientific testing was done, there was no discernible effect on the performance of the VHF/UHF radio system.

For the diplexer located atop the tower, three small drain holes were drilled in the side facing downwards and in addition to the connectors (all N-type, which are ostensibly weatherproof) being sealed with appropriate methods, a bead of silicone was run along the seam of the die-cast box that was facing upwards.  These precautions should prevent moisture from accumulating in the box and based on past experience, it should experience a long service life.

* * *

Footnotes:

1 - Mike's recent experimentation has been with end-fed half-wave antenna.  While the traditional end-fed antenna (e.g. a "Zepp") has been used for about a century, this differs from the traditional implementation by using a broadband matching transformer to match to the high impedance "near" the end of the 1/2 wave section so that, unlike a traditional Zepp, it will work on multiple bands' half-wave harmonic multiples - and the use of this transformer has the added advantage of providing a DC ground for the wire to eliminate static accumulation.

The strategic placement of this transformer, along with a short "tail" of cable connected to the transformer and appropriate feedline decoupling, multi-band operation without inducing currents along the feedline itself, yielding performance very similar to that of a 1/2 wave dipole at its fundamental frequency while allowing operation on more harmonic frequencies than is typically possible with a standard center-fed dipole.   With antenna lengths appropriate for 160 and 75 meter phone, the 40 meter "resonance" turns out to be just above the 40 meter band, but a small inductor was placed in a 40 meter current node near the far end of the antenna and this moved the resonance near the middle of the band without having an effect 160 or 75 meters.  The transformer that was used was supplied by "myantennas.com" and was designed for the 160-40 meter frequency range.

The use of an end-fed half wave can be advantageous over a dipole since, unlike a dipole, there are only two "ends" that may need to be supported which may be more convenient in instances where there is a single elevated structure for attachment while avoiding the weight and sag of a feedpoint (which may include a balun) that might be located mid-span.  Its relatively high feedpoint impedance can, if the antenna is properly installed, reduce the amount of current flowing in other structures (e.g. the tower itself, the feedline) as compared to an antenna like a "sloper" and both reduce the probability of RFI being caused by transmitted energy as well as interference conducted from the ham shack's electrical system, onto the feedline and antenna - provided that the proper precautions are taken.

2 -  A VNA (Vector Network Analyzer) is pretty much the ideal tool for characterizing a device like this, but satisfactory results could have been obtained using simpler devices.  Insertion loss could be measured using a pair of known-accurate watt/VSWR meters with one on the input and another on the output of the "leg" of the filter, terminating the meter on the output side with a known-good 50 ohm load.  Spot-checks would be done on frequencies - such as on the band nearest the edge (e.g. 6 meters and 2 meters for the low and high sides, respectively) as well as the "next" bands over (10 meters and 70cm.)  Iterative measurements of best VSWR and lowest measured losses should get one "pretty darn close" to optimum.

Similarly, one could use the meter on the input to monitor for changes in the VSWR, iteratively tweaking the inductors as necessary to minimize both VSWR and insertion loss.

* * *

This page stolen from ka7oei.blogspot.com

[End]