This process is not onerous: it takes a little more than 2 hours. For an excellent antenna on top band it is worth the small effort twice each year. Until I come up with a better idea the ritual will continue. It will take longer starting this year since I intend to double or quadruple the set of 8 30 meter long radials. Other changes to improve performance are also planned.
Although summer is not peak season on top band there are contests and DX to be chased. A second antenna removes that hole in my calendar and provides a fall back in case of main antenna failure. Ice storms, wind and other periodic calamities can destroy a quarter-wave of wire in the air.
From the start I planned for a 160 meter option on my 3-element vertical yagi for 80 meters. The idea was to turn the driven element into a short (loaded) 160 meter vertical. The radial field for the five 80 meter elements is extensive and therefore low loss; the 4 parasitic elements are disconnected from their radials for 160 meter operation. Loading lowers efficiency but if not excessive it is a price well worth paying. After all, a few decibels of loss is better than -∞ db!
As constructed the 80 meter yagi switching system and shack control head support 160 meters. A relay in the switching system selects an alternative path to the driven element. See the previously linked article for the design and construction of the 80 meter yagi control box. I won't repeat any of that information this article.My short term solution is to install a base coil to the driven element. In future I would like to move the coil farther up the tower, a project that will reduce loss.
I made several mistakes right from the start. Since these are educational I will tell you about them. Too often errors are excised from articles (and far too many science papers). The presented results are "sterile" when the all important process and experiment errors are absent.
If you build this or a similar antenna you may make the similar mistakes. By documenting those mistakes my hope is keep others from repeating them.
Mistake #1
The first was to estimate the coil inductance from the EZNEC model rather than by antenna measurement. The model said the antenna impedance at 1.83 MHz was approximately 20 - j320 Ω. This requires an inductance of 31.5 μH to cancel the capacitive reactance. There is still the low R value to deal with but until I knew the measured coil ESR (equivalent series resistance) and radial system equivalent resistance the true feed point resistance.
I built a coil with an inductance of 31 μH and a Q of 400 using K6STI's Coil program. The modern version of this software appears to work well. Alternatively there is an online coil calculator by ON4AA that includes Q, reactance and other data that works well but has a few quirks.
The coil form is a PVC coupler for 3.5" pipe which, including insulation thickness gives a coil diameter of 3.5". Form loss is negligible at 1.8 MHz. A "fat" coil like this has a high Q and therefore a low ESR which reduces the typically high I²R loss of a base loaded short vertical.
After winding it using scrap AWG 14 THHN wire I measured the inductance with my ancient LCR meter. Stray inductance of the leads is 6 to 7 μH which is subtracted from the measurement. Accuracy is further affected by the meter making the measurement at AF rather than RF. Adjustment of the inductance must be done in the field.
The coil is connected between the 160 meter stud on the control box and the driven element (tower). For the initial test I didn't do this since it requires opening the box and manually powering the switching circuit. Instead I disconnected the 80 meter stud and connected the coil to it.Since the unpowered (default) state is omni-directional 80 meter mode and the DPDT relay to select 80 or 160 meter stud is on the output of the 80 meter switchable L-network this temporary connection is the same as normal use. More on this design curiosity later.
The antenna did not behave as expected. Resonance was near 1300 kHz, very wide of the mark. Bypassing the control box I measured the impedance of the driven element at 1.83 MHz without the coil as 17.5 - j167 Ω. R is in close agreement with the model but X is twice the magnitude.
The coil is obviously far too large. It was a mistake to rely on the model and skip measuring the impedance before designing and building the coil.
Mistake #2
When I wired the control box I attached the 160 meter stud (via relay) to the output of the L-network. This was expedient because the L-network input was not very accessible. Since the network used a low pass topology and modelling confirmed that the low pass L-network had a low impedance at 160 meters I believed this would work. Wrong!
There are two reasons why this was a mistake. The first is that although the L-network is low pass it is not transparent. Here are the measured impedances of the driven element on 1.83 MHz:
- Transmission line port: 3.2 - j66 Ω
- Isolated driven element: 17.5 - j167 Ω
This is not a disaster. If the 160 meter base coil brings the match reasonably close to 50 Ω at 1.83 MHz the SWR will be low regardless of the L-network phase shift. Unfortunately it's not that simple. First, the base coil lifts the antenna R very little, so the SWR can be no better than 2.5. Were that deemed acceptable it still wouldn't suffice.
Although the 80 meter L-network is pass through at 1.8 MHz the addition of the 160 meter base coil changes that. The L-network and added series coil are not separable: we have instead made a completely different L-network, one that is not transparent on 160 meters. The shunt capacitor is the 600 pf used for the 80 meter omni-directional mode and the series inductance is the sum of the ~1.5 μH of the L-network coil and that of the 160 meter base coil.
There is no escape. The 160 meter mode of the 80 meter yagi requires its own L-network. Designing one with TLW is pretty simple.
The new L-network is in series with the 80 meter L-network. With the 160 meter network in place the 80 meter L-network is once again pass through since its output port is near 50 Ω on 160 meters. They do not otherwise interact.
Network loss is low. The shown C and L loss is for component Q of 1000 and 200, respectively, at 1500 watts. The Q of my coil is calculated at 400 so the loss is half that shown. I also run less than 1000 watts (our legal limit is lower than the US) so the loss is proportionally lower. Loss in 80 meter L-network should be very low and has not been calculated. This can be done in EZNEC with careful adjustment of component ESR per the calculated Q.
As shown further below I initially installed a 2000 pf shunt capacitor and the coil was adjusted in the field. The low pass L-network design works.
Mistake #3
Let's see if you can follow along without a detailed schematic. There are two L-networks in series, the first for 80 meters and the second for 160 meters. The 80 meter L-network is always in line. Its antenna side port is switched by relay between the driven element and the 160 meter L-network. Each of these has a stud.
The relay is only activated for the 160 meter mode. When 160 meters is selected the L-networks ports are connected. However, at all times the 160 meter L-network series coil is connected to the driven element and the shunt capacitor is connected to the radial hub. Therefore when an 80 meter mode is selected there is a series LC circuit formed by those components across the antenna port of the 80 meter L-network.
This could be a problem. The series LC circuit resonant frequency is F = 1 / (2π sqrt(LC)). That's the frequency at which impedance is minimum and current maximum because XL = -XC. In this case the frequency is in the vicinity of 950 kHz. The problem is that the series LC circuit has a low enough impedance on 80 meters to take appreciable current and therefore upset tuning of the L-network and add loss on 80 meters.
The calculated impedance at 3.5 MHz is ZL + ZC = 330 Ω. This is simply a voltage divider. Current in the series LC circuit is less than 10% of the total due to an 80 meter impedance of 30 Ω (omni modes) or less than 20 Ω (yagi modes). This isn't large but neither is it negligible. Performance on 80 meters may be affected. Correcting that problem requires a relay to disconnect the 160 meter L-network from the radials when operating on 80 meters.
Calculation can do wonders here however I simply measured the SWR. There is negligible effect on all 6 of the 80 meter modes. On air it the 80 meter array continues to perform as it should. I decided to leave it alone, at least for now, since there does not appear to be a problem in practice.
Tuning it up
With all the mistakes fixed or accounted for I proceeded to tune the antenna on 160 meters. The coil was adjusted until the minimum SWR was approximately where I wanted it, at 1830 kHz. A shunt 2000 pf mica capacitor was added between the radials (ground) and the 160 meter stud.
It should be obvious that this is a temporary installation. The capacitor will be put inside the control box and adjusted to improve the matched SWR. I have small ceramic capacitors for this task. The coil is large to maximize Q so it will remain on the exterior. It will be given mechanical support and a rain cover to preserve Q in wet weather.
Despite the high Q coil and capacitor the SWR curve is pretty good. By decreasing coil inductance a small amount the match will move a little higher and achieve a low SWR from 1800 to 1900 kHz. Although my interest is CW during major contests activity does extend that high.
The series LC circuit formed by the added coil and capacitor has not changed the 80 meter SWR for all of its 6 modes. I didn't expect to get that lucky. There is no need to use a relay for the capacitor.
Tuning was facilitated by the +12 VDC on the spare control line. This was my first opportunity putting it to use. With some squeezing of the fingers past the various cables I used an alligator clip to switch among the 7 modes: 6 for 80 meters and 1 for 160 meters.
Performance: model
A short vertical has poor performance relative to a full size vertical. This antenna cannot do as well as my primary antenna. That isn't its purpose. It's intended to be my year round antenna that is always available when my primary antenna is removed during the haying season.
Relative performance is limited by the following:- Short antennas have "fat" patterns. For a short vertical the elevation beam width is larger so there is less gain at low angles. DX suffers although there is a boost for shorter paths.
- Base coil loss converts power to heat. This can be mitigated with a high Q coil and its lower ESR, which I did (see calculation above).
- Short antennas have a low radiation resistance. Since the radiation resistance is in series with the ground loss more power is dissipated in the ground as in a larger antenna. This can be mitigated with a more extensive radial system.
Performance: on air
In practice the difference is difficult to discern. QSB cycles are out of phase due to the different locations. The actual losses in the short vertical may be less than predicted. The equivalent resistance of the 80 meter radial system cannot be easily calculated since there is a complex combination of relatively short radials (20 m) on the driven element interconnected in numerous places to the 4 parasite 15 meter radials systems. However it is likely quite favourable based on its measured resistance (sum of radiation resistance and ground loss) and the extent of the combined radial system.
The first test is noise. When noise is equal in all directions two antennas of equal efficiency will produce the same noise power in the receiver. The new antenna is quieter by a little less than 1 S-unit. However from the pattern shown above the primary antenna is not omni-directional: it has relative gain away from the 150' tower at a heading of approximately 200°. Noise is always stronger in that direction since weather is more intense towards the south. I see the same noise level difference on the 80 meter yagi, with northeast and northwest directions quieter. The difference disappears during daylight when the band is largely closed, so the noise is not local QRN. The directional effect due to the new 140' tower is not included (yet) and it is 40 meters southwest of the primary antenna.
When evening arrived I listened at intervals. Activity was light. US stations were consistently stronger on the primary antenna by 1 to 2 S-units, depending on direction and measurement uncertainty due to QSB. European stations later at night were typically no more than 1 S-unit stronger on the primary antenna. So far this is consistent with the models, understanding that on the FTdx5000 the S-meter scaling is non-linear and is no more than 4 db per S-unit. It also gives me a nudge to do something about that gain loss towards Europe due to tower interaction with the primary antenna.
I have made a couple of contacts with the antenna and it does get out okay. I have not asked anyone to participate in an A-B test but may do so later. Since it takes a kilowatt without blowing up the coil loss is not extreme. I have not run out to the hay field to check its temperature!
I will continue to compare the antennas until early May. That's when the primary antenna will be rolled up for the season and I only have the new antenna for top band. For the present I am pleased with it even though I have to give up a few decibels of gain.
80 meter checkup
Because this was the first time I had the control circuitry for the 80 meter yagi open since it was completed last fall I did a brief maintenance checkup. The inside of the box was dry and there was no discolouration of metal surfaces and wire insulation. In fact it looked pristine.
The one problem that had developed was severe attenuation on receive, for all operating modes. When it first appeared as an intermittent problem I thought it might be a relay in the 8×2 antenna switch.Swapping the coax with another port made no difference. Transmit performance was unaffected. Corroded and abused relay contacts behave in this fashion but it wasn't the relays.The cause was dissimilar metal galvanic corrosion. In my haste last fall I attached the wire to the driven element (tower X-brace) with nothing more than a stainless hose clamp. I had intended to solder on a tinned lug so that galvanic corrosion would be slow. Copper to zinc are far apart on the anodic index.
Sanding off the corrosion and reattaching the wire restored receive performance. I'll make the necessary modification in the coming days or weeks. The same goes for the temporary joint between the 160 meter coil and tower.
No comments:
Post a Comment
All comments are moderated, and should appear within one day of submission.