Friday, November 27, 2015

Conspiracy of Silence

This week the low bands were quite good. An extended period of quiet geomagnetic conditions lowered the noise and ionospheric attenuation. With CQ WW on the horizon many hams dispersed to various locations around the globe to have some fun as desirable contest multipliers. It is also prime DXpedition season. All of this is to say that lots of interesting, and sometimes rare DX was there for the taking on 40, 80 and 160.

When neighbourhood QRN allowed I prowled the CW segments of 40 and 80 (no 160 meter antenna, unfortunately) at sunrise and sunset, and late evening when the terminator favoured conditions to distant countries. Most of the DX was of the more ordinary variety. They were still fun to work, if only to hone my pile-up skills.

The DX spotting clusters were a beehive of activity. I was there as well, taking my cue from spots and making a few of my own. For the most part it was more fun to just tune and listen, finding the DX on my own. It's good that I enjoy this mode of operating since I must do so during contests, where I always operate in non-assisted categories. It is also mandatory when DXing with QRP: there's no point in finding a DXpedition when a pile-up has already formed, everyone having been attracted by the global spotting networks. QRP is not competitive in pile-ups.

Tuning the gaps

Band maps in popular logging software make it easy to tune from spot to spot, picking off the DX stations of interest. Motivations vary, from the rare ATNO (all time new one) to new band-countries, or the simple joy of working rare and semi-rare DX. However there is more out there than what the band map will tell you. For that you need to tune the gaps, those blank areas on the band map between spotted call signs.

Most often what you'll find is silence or stations that don't attract the attention of DXers. It isn't often someone will spot a VE3 call such as my own; few need VE3 for a DX award! However, those unspotted station may offer an opportunity to engage in a longer QSO, if that is your pleasure.

Hams whose primary interest is DXing either listen a lot or monitor the cluster spots while going about their lives or doing some off-air activity in the shack. It is fascinating to hear a seemingly dead band fill with hundreds or thousands of stations within a few minutes of a DXpedition spot relaying around the globe. Those are hams tuning the DX spotting networks, not their radios

The truth is most DXers do not bother to tune the gaps. Instead they stay silent, only activating their rigs when what they want appears. The longer they've been in the DXing game the less they find of interest to work. That's how they enjoy the hobby. For myself that seems a little sad, though there's nothing wrong with it. Out hobby has room for all types.

Now let's go back to tuning those gaps since I'm a ham who enjoys finding stations to work.

The conspiracy

While tuning 40 CW early one evening this week I came across a few US stations calling someone, and doing so at a relatively slow speed. Looking at the band map there only white space around that frequency. Whoever it was had not yet been spotted. A weak though perfectly copyable, sending slowly, came back to someone. That peaked my interest.

Flipping between my two inverted vees I found reception best on the east-west antenna. Since the only darkness to my west at that hour was within North America I reasoned that the station must be to the east. That might mean Africa. I was briefly left guessing until he signed his call two QSOs later.

When he signed on completion of the next QSO I learned it was XT2AW. He's been active lately and I had worked him on one or two of the higher HF bands. Not too rare, though a really nice catch on 40. This would be a new band-country for me (when I returned to the hobby in 2013 I restarted my DX count). I awaited my chance and listened. The band map continued to show a blank space. Then I started calling.

I worried that someone would spot him and bring in a pile-up before I could get through. An inverted vee is not a pile-up buster, nor is 100 watts. Yet no one was spotting him. For the moment I was protected from insurmountable competition. It was as if the others, like me, had stumbled across him while tuning the band and wanted to keep him to ourselves for a little while. Consider it a form of courtesy, to give you a prime opportunity as compensation for making the effort to find the DX without outside assistance.

Within a few minutes I had him logged. All this time I never heard more than 3 stations at a time calling him. As with those who came before me I declined to spot him, thus extending the courtesy to later arrivals.

The conspiracy had held. I listened for a moment longer then went DX hunting on 80 meters. When I returned 10 minutes later there was a medium sized pile-up on the XT. Glancing at the band map I saw that someone had finally spotted him.

Reality check

Was it truly a conspiracy? There's no way to know without interviewing everyone involved. It could be nothing more than a coincidence. If so it's a common coincidence since it's something I've witnessed many times and always under similar circumstances. Each time I felt no urge to spot the DX. Perhaps my sense of courtesy really is common to other DXers. That would explain the appearance of these silent conspiracies.

It doesn't really matter. Whether true or not it behaves like one so it's useful to view it that way. It's also quite a common occurrence. If it looks like a duck, quacks like a duck and walks like a duck, we are free to assume it's a duck.

Should I or shouldn't I?

The conspiracy won't last. It never does. The only question is how long it will last, or how long it ought to last. Someone will spot the DX. Perhaps you. How do you decide?

There's no one right answer. One thing I look for is whether the DX station "sounds" like a casual operator or someone who wants to run. In the latter case I will often send a spot quite soon after working him. Other times I'll wait until there are no callers left. Since I often don't hang around after logging a QSO I do nothing at all. So it was with XT2AW. I thus left the spotting decision to someone else.

It comes down to your own sense of what's right and appropriate. On that I can make no recommendations. I will only say that the next time you encounter a situation as I described above to stop and think, and ask yourself why you are spotting the DX. Of course it is appropriate to return the favour of spots if you benefit from the spots of others. I would never discourage that. Only sometimes it may be best to delay or forego the opportunity. It's your choice.

With that I'll leave you for a few days. It's time to torture myself for 48 hours by operating the CQ WW CW contest with QRP.

Wednesday, November 25, 2015

Choices on Adapting Prop Pitch Motors to Rotator Service

My plans for a larger station continue. I have the tower, the transmission line, a few antennas and assorted components for building long-boom yagis. I now also have a couple of rotators, or at least one rotator and one rotator-to-be. This is what I want to discuss today.

Big antennas require powerful rotators. The key metrics are turning torque and braking torque. Since purpose-designed large rotators are expensive many hams have adapted surplus prop (propellor) pitch motors from aircraft of an earlier era. They are ideal in that they have high turning and braking torque, and a shaft rotation of close to 1 rpm.

However a motor is just that: a motor. While important, a motor is only one component of a rotator. The two motors I now own came from the same ham that sold me the tower. One comes with a complete platform and drive system for the tower, and a homebrew controller, all of which is ready to go . The other prop pitch motor was in storage as a spare. Both appear to be working well.


On the left is the motor that had been up the tower. The hardware is almost all original equipment. When refurbished many choose to rebuild with new hardware, as seen on the right motor. The shell contains the motor and the body contains the reduction drive. The crown gear for the shaft is underneath (hidden in this view). I have the matching shafts for both motors. Wires on the motors have been rerouted to exit at the joint between shell and body. On the far right of the photo is a tower plate and thrust bearing to support the mast and chain drive. I am in the process of overhauling the bearings.

My plans for some large yagis require rotators that are up to the task, and prop pitch rotators fit the requirement well. I have some choices ahead of me and some work to prepare these motors for use in my next station.  In this article I'll walk through those points.

History

Rather than repeat here the history of prop pitch motors I will link to the excellent material posted by K7NV on his web site. My motors are the small size, perfect for the antennas I have in mind. The bigger motors are often used to power rotating towers.

Prop pitch motors have not been manufactured since around 1960, when (primarily military) aircraft technology advanced. It's humbling to think that the prop pitch motors sitting on my work bench are quite likely older than I am. That they continue to perform well in countless antenna farms is impressive.

Power and cabling

The clip leads get warm to the touch in this test
The motor within the small prop pitch motor runs at ~9,500 rpm. That's fast! The motors in the larger units are only a little slower. The reduction drive brings this down to ~1 rpm (I measured 0.75 rpm with 24 VDC), just about perfect for a rotator of this size. The motor is not only fast it is powerful. On a bench test one drew 8 A and the other drew 7 A at 24 VDC, or about 180 watts. This has implications for the power supply and cables.

The power supply is easy enough since a motor does not require low-ripple DC. A typical motor power supply consists of just 3 components: transformer, bridge rectifier and filter capacitor. The capacitor is sometimes omitted. Components should be rated for continuous duty.

In the simplest configuration 3 wires go to the motor: common, clockwise and counterclockwise. Power is applied between common (not necessarily tied to station ground) and one of the other wires. Alternatively a relay at the rotator can switch directions, with a lower gauge wire to power the relay. That can save some money at the risk of a relay failure at the top of the tower right when it'll hurt you most: a contest or DXpedition.

From Ohm's law we expect the motor to present a 3 Ω DC resistance (R = E/I), and that's just what I measured. This is low enough that the gauge of the motor wires must be carefully selected. According to K7NV 10 AWG wire can be used for up to 300 foot (90 meter) runs. That isn't much since the large antennas that are likely being turned on high up and far away. For example, 150' out to the tower and another 150' to the top.

Since 1,000' (300 m) of 10 AWG copper wire has ~1 Ω resistance his suggestion implies that a 20% voltage drop (from 24 to 20 volts) is acceptable. In a complete return circuit of 600' of wire we have 0.6 Ω in series with the 3 Ω motor. Smaller gauge wire can be used if the power supply voltage is raised. For example, the same run done with 12 AWG wire would require 28 VDC to achieve the same result. Clearly it is best to match the power supply voltage to the length and gauge of the motor wires.

Mast coupling

The drive shaft exits from the bottom of the motor. If it is to be used as a conventional rotator the motor must be mounted upside down (shaft pointing up) and fitted with a clamp for the mast. A typical installation would have the motor flush against a steel plate affixed to the tower, with a hole for the shaft. Optionally a thrust bearing can be used to direct the vertical force of the mast and antennas directly to the tower.

There are reports that this approach should only be taken if the oil in the reduction drive is replaced with grease. Otherwise the oil will eventually foul the motor. This procedure requires a complete disassembly of the reduction drive.

The other approach is to mount the motor upright (shaft pointing down) on a platform attached to the side of the tower, and use a chain to turn the mast. Rotating towers are similar but with the motor anchored to the ground. More mechanical work is involved for a chain drive system which, unless you have the skills and tools, will require the services of a machine shop. It is important that both the mast and motor shaft use thrust bearings since the drive torque will impart a large lateral force on both mast and motor shaft. Two bearing are needed for the motor to ensure no lateral force is transmitted to the motor.

There is just such a drive system in the tower package I purchased, all of it custom built many years ago. Since it is beautifully built and works well I intend to use it. However it is extremely heavy and had to be disassembled before I could carry it by myself. For the second motor I will have to decide how to proceed when (if) I have another tall tower and large array. I can defer this decision for a year or two.

Direction indicator

Apart from situating the shack window where you can see the tower and antennas, there are two common methods for indicating direction with a prop pitch rotator: pulse and pot.

In the former case, a pulse (momentary contact closure) circuit is made for every rotation of the motor or a selected point in the reduction drive. With a prop pitch motor this can be challenging since it's rotating from 7,000 to 9,500 rpm (115 to 160 times per second). K7NV sells a magnet-driven reed switch add-on to the motor axle.

The controller counts the pulses to calculate degrees of rotation. The controller requires a calibration circuit to assign a reference point and degrees per pulse. If pulses are missed the error will grow over time and require recalibration. This reminds me of my first rotator, the CDE AR-22R, which used a contact pulse to drive a solenoid to ratchet the direction indicator in the controller. It was flimsy and unreliable, requiring constant recalibration. Pulse technology has gotten much better over the decades!

In the case of a pot (potentiometer) the mast or shaft directly drives the wiper of a pot, the resistance of which indicates direction. The popular Ham rotator series uses this method. In this case a linear pot is attached to the motor & drive assembly and the pot wiper is turned by the bell housing as it rotates. You must carefully align the ring gear, pot, stops and bell housing during assembly. Trim pots in the controller set zero and full scale.

For the prop pitch motor a mechanism must be built to drive a linear pot. A calibration circuit in the controller is desirable to allow over-rotation (more than 360°) and not require climbing the tower to make any adjustments. A universal calibration circuit sets the resistance for the selected rotation stops in clockwise and counterclockwise directions.

The prop pitch motor and controller I acquired uses this style of direction indicator. An op-amp configured as a voltage comparator is calibrated with a trim pot for zero scale, and another in a buffer amp set full scale. Modern controllers typically use microprocessors and software calibration. Not all commercial controllers support both means of direction indication so this must be decided before purchasing a unit. For example the EA4TX controller does not support pulse direction indicators. The Green Heron built RT-21pp by K7NV supports pulse.

Regardless of the method used it is advisable to have rotation limits set by hardware or software to prevent destruction of the coax to the antennas.

Weatherproofing

In its original application the motor is protected from the weather by the propeller conical hub cover. In its designed horizontal orientation the motor is designed to keep the lubricant inside. To keep the weather out there are places in the body of the motor that must be sealed.

Both of the my motors have been weatherproofed with sealant in the all the right places. That was long ago so some maintenance is required. The two critical areas are the motor cover and the motor drive.

The motor wires in the original design often exit via a hole in the side of the motor cover. Most often the wires are rerouted to exit closer to the body of the motor where the hole is less exposed to the weather. The original holes and the new one must be sealed against moisture intrusion.

Unless the original adapter plate is used the shaft seal is exposed to the weather. This is usually not a concern when it is mounted upright (chain drive). But when mounted upside down to couple directly to the mast a cover over the motor ought to be used, even if the adapter plate is present. The oil seal around the shaft is old and is insufficient to repel wind-driven rain or melting snow and ice.

Controller choice

The ham I bought the motor from built his own controller. It is quite old so it is entirely analogue. The motor power supply is quite simple, as described above. Switches choose direction of rotation and another is for the transformer primary. Thus the motor power supply is off when the motor is idle. The direction indicator uses a pot with a planetary drive on the mast, as described earlier. The indicator itself is a meter from a junked Ham-M rotator.

My first inclination is to keep the retro controller but to move it to a surplus Ham IV controller. The brake and rotation switches can be rigged to operate the motor. However there is no limit mechanism so it is necessary to avoid going beyond the standard 360° rotation range. Well, you may occasionally want to do so to catch an elusive multiplier or new country, but only if you prepared by installing enough coax slack in the connection to the yagis.

The commercial controllers I am looking at are the K7NV/Green Heron RT-21pp and the EA4TX. Both are well regarded. I described their different approaches to direction indication up above (with links to their product pages).

The RT-21pp is the more sophisticated and expensive of the pair, including electronic start and stop power ramps, manual and computer control, power supply and motor pulse unit. The EA4TX requires an external motor power supply, which it switches with relays (no power ramp), computer control (manual control appears minimally adequate) and potentiometer-based direction indicator. It is also significantly less expensive.

Should I choose a commercial unit, especially to have computer control, I currently lean towards the EA4TX product. I have the pot already so I won't have to purchase and install pulse units for the motors. From my observations of the motor start and stop behaviour (see below) I should be able to get by without power ramping.

Stop, Start and Braking

It takes about 2 seconds for these two motors to spin up to full speed and about the same to spin down. Under load the time will increase a small amount. The reduction drive of ~10,000:1 cannot react instantaneously. This is good since it is not desirable that large antennas accelerate too quickly.

The K7NV/Green Heron controller chops the power with a solid state switch to electronically ramp power at the start and end of rotation. This is particularly helpful for the subset of motors that have an integrated brake, which could, if left as is, risk antenna damage. Since my motors are not of this type I do not absolutely require this controller feature.

The braking torque in a prop pitch motor mostly comes from the reduction drive; it is very difficult to turn the motor by applying torque to the output shaft. For example, momentum of a turning yagi or asymmetric wind load. That is usually sufficient with a prop pitch motor to make a brake redundant. On the commonly used Ham series rotators it is possible, with the brake released, to turn or stop a turning rotator with your bare hands. I've done this for laughs when I was younger, though I don't recommend it since you could injure your hands or wrists if you don't grab the bell housing in just the right manner.

Some of the large commercial rotators with worm drives (Alfa Spid, Prosistel, etc.) don't need a brake for much the same reason. However those rotators should use a controller with an electronic ramp (manual or automatic) to gradually accelerate and decelerate the rotator.

Preparing for use

The first motor will not go into service until at least late 2016. The other will require a tower mount and chain drive system to be designed and built. However that is not possible until I know what model of tower it will be used with. That takes me to perhaps 2017. Until then the second motor is a spare. They are easily interchanged on the existing motor platform and drive system.

Other maintenance includes the thrust bearings, weatherproofing and, possibly, internal inspection and lubrication. I also want to refurbish the controller. This is a project for next year since the motors will not be required until my first large tower is installed.

For me this will be an excellent learning experience. I've never had the chance to directly play with prop pitch motors, only to use them a couple of times at other stations. Learning new things is good for the soul. But right now I need to go and fish a thrust bearing out of the oil and solvent it's been soak in for the past 2 weeks. I am attempting to salvage it rather than buy a new one.

Monday, November 23, 2015

Pile-ups and Dual Receive

Along with many DXers around the world I chased the VK9WA Willis Island DXpedition. The DXpedition is now history, and I logged several QSOs. Two of those came relatively easy, with the rest more difficult. I failed to get through on 20 and 80, though 80 was almost impossible with 100 watts and a less-than competitive vertical. Most often I missed them because I was unable to get on the air.

For the bands where I did get through, 17 and 40 were the toughest. On 17 meters this was due to only having an inverted vee. On 40 this was due to...only having an inverted vee. I used the same transceiver feature to enhance my chances on those bands: dual receive. If you haven't considered this possibility you may benefit from the following description of the tactic that exploits this feature.

Dual receive set up for working VK9WA on 40 CW. Notice which RX and TX lamps are lit.

The FT-1000 MP, like a number of other transceivers, has two independent receivers. On some it is built in and on others it is an option. The RF front-end and audio sections are most often common to both receivers. There is only one transmitter, which can be switched between receiver VFOs.

Dual (or multiple) VFO transceivers can in some cases be used for the same pile-up tactic, but typically not to the same degree of effectiveness. I will focus on "true" dual receive rigs. You may also be able to do it on an SDR rig.

Problem statement

At first blush one might think dual VFOs are enough, or even just RIT & XIT. You tune the VFO to DX transmit frequency and use XIT or the other VFO to transmit split. The second VFO or RIT/XIT offset knob is tuned to find the station the DX is working. This is the pivot position from which you set the transmit frequency in preparation for your call. Where you tune of course depends on the DX listening pattern and your guess at where he'll next listen. This is the common approach.

There are problems doing it this way, even if it suffices in most cases. First, the RIT knob is typically small and has a fine tuning rate. That slows you down when you need to move quickly. This is in addition to pressing the RIT button a couple of times: once to switch from listening to the DX to the pile-up and again to go back. With a good DX operator you may have just 2 seconds to find the station and another second to set the frequency for your transmission.

Another problem is that you can only listen to one frequency at a time. It is easy to miss the completion of the QSO. The DX operator most often sends a simple "TU" on CW, which you can easily miss, you end up missing your turn to call. Or you call when you should not, or press the wrong button and call where you should not. It's easy to do, unfortunately.

The RIT/XIT knob on the FT-1000 MP (outside the frame of the photo above) is the same size as the small knob at the lower right. RIT and XIT states appear on the lower right of the main display when enabled -- which they are not in the photo since I'm using dual receive. It is difficult to spin the RIT/XIT knob quickly, plus the tuning rate is slow and the maximum offset is 10 kHz. These were all impediments in the VK9WA CW pile-ups.

Dual receive drastically changes the situation to your advantage. There are ways to best use the feature.

Open wide

The way I use dual receive in a pile-up is as follows:
  1. Main receiver fixed on the DX transmit frequency. The sub-receiver is used to find callers and to transmit. Start by tuning to the DX on the main receiver and pressing A->B (or equivalent) to bring the sub-receiver to the same frequency.
  2. Activate dual receive. In most rigs this routes the audio of each receiver to one side of the stereo audio output. Obviously this works best with headphones.
  3. Adjust audio balance so that the main receiver audio is at full volume and the sub-receiver at reduced volume. You need to clearly hear the DX. The pile-up at low volume almost always suffices. If the audio balance feature isn't available you can try positioning the ear piece corresponding to the sub-receiver partially off your ear lobe.
  4. Find the caller with the sub-receiver and decide whether to offset up or down, and how much, to make your call. As in any pile-up you'll soon discover the pattern and if conditions are favourable you'll make the QSO.
With true dual receive you have the choice of roofing and IF filters and DSP on each receiver to improve reception. I use the full force of what's available on the main receiver to cut the clutter of QRM, be it deliberate, ignorance or error in setting split. However, on the sub-receiver I leave the IF wide open. On CW this means using the SSB filter (2.7 kHz). Sound like an odd choice? It is often the right choice. Let's see why that is.
  • You'll find the calling station faster. Often you won't even have to shift the sub-receiver to hear the caller since the DX usually shifts listening frequency in small steps, steps that are well within the wide filter's bandwidth.
  • The intent is not to copy the caller well, only to identify which signal it is. The successful ham's transmission is a giveaway. You'll hear something like "R" first, or just "5NN TU", which is very distinctive in comparison to the horde sending their calls. And make no mistake, just because the DX calls a specific station other stations don't shut up. The QRM is non-stop on the rare ones. It made be rude and unfair, but you must play the hand you're dealt. Complaining won't get the DX in the log.
  • Only if you don't hear the caller should you tune the sub-receiver. If you can, remember your starting point since you may want to flip back there if you fail to find the caller (happens a lot) and the DX pattern is to make small offsets between QSOs. Be careful with the knob since it tunes very fast in comparison to RIT. The speed is only a benefit if you have a steady hand.
  • In most pile-ups the callers tend to cluster around the last successful caller. In response the DX will often spin randomly or will make large frequency shifts every few QSOs. You'll respond to those more quickly than others with the wide receiver filter. Not only can you find the caller faster this way you never miss when the DX resumes transmitting. The number of potentially winning calls is much higher with dual receive.
Another thing you can do is discover if the DX ever hunts out the relatively sparse areas in the listening window or the fringes to find stations that are easier to copy. Some of the VK9WA operators did so. This happened a couple of times while I tried to get through on 17 meters, with the DX occasionally listening up +10 to +12 kHz rather than the more usual smaller split. So I moved up there and within a minute had them in the log. That was easier than working hard on every call to stake out a calling frequency. I might not have gotten through otherwise.

Is it worth the cost?

Dual receivers are pretty much necessary to use the described tactics. It may not be adequately achievable with either software VFOs or SDR. For example, the use of roofing filters requires that two software VFOs must tune within the pass band of that roofing filter. Otherwise the DSP bandwidth filter can't do its job. Additional hardware that includes at least a portion of the receiver electronics is needed.

If your rig already has dual receive you are set. Otherwise be prepared to pay in the vicinity of US$500 to add it, if the option is available. For those for whom chasing the rare DX is a passion the cost is almost certainly acceptable.

Only you can decide whether the value of this feature is worth the cost. I didn't choose the FT-1000 MP for this feature, but I am glad that I have it.

Tuesday, November 17, 2015

Performance of Coil-loaded 40 Meter Elements

Of all the articles on this blog by far the most popular are those for 40 meter antennas. In particular wire yagis and other simple antennas with gain. This is not surprising. Achieving gain on antennas for 20 meters and above is routine, while on 80 and 160 meters it is almost always out of reach. That leaves 40 meters: a band with great possibilities, for which gain antennas are just within reach for many hams. But it also challenging, thus the interest in practical designs. With the solar cycle declining 40 will become the "go to" band over the next several years.

If you've followed this blog for a while you'll know that 40 meter gain antennas are very much top of my mind. I cannot fit in a gain antenna at this location, not even a wire yagi, and certainly not the 2-element yagi I have in storage. So I plan ahead.

One of the questions I am pondering is the relative performance of 40 meter rotatable yagis with shortened elements. A full-sized 3-element or 4-element yagi is a monster: typically well over 100 kg weight and up to 30 ft² (3 m²) of wind load. Although the tower I recently purchased can handle this size antenna there is the matter of erection and maintenance. I want to be certain the effort is justified. So if comparable performance is achievable with a smaller yagi that becomes an attractive option.

The subject of compact and full-size 40 meter yagis is something I hope to address in future. In this article I will focus on coil-loaded tubing elements. Rather than address a complete yagi with short elements, which complicates the design in several respects, there is some insight to be gained by investigating a single loaded element.

The basic design

With EZNEC I modelled a dipole with 25 mm (1") constant diameter aluminum tubing. This is not a realistic design, which would require tapered tubes. The Leeson correction in EZNEC cannot be properly applied to elements with loads, be they coils, capacity hats and other systems. For the current study there is no need to find the exact length: the NEC2 engine will tell us what we need to learn about performance. Getting the dimension exact can be dealt with when and if an antenna is built.

I placed the loaded dipole in free space to avoid the effect of ground. Ground interaction with the near-field of a yagi is in any case less that than of a single-element antenna, allowing the model behaviour to be most useful. Coils are placed at the midpoint of each half element. This is a typical placement for loading coils since they become less effective and larger further outward and can decrease element efficiency when placed further inward.

Element length is then progressively shortened from full-size to just under half-size. Inductance is set so that resonance (R+0j) in all cases is 7.100 MHz. Gain is calculated with selected values of coil Q from 100 to 800.

The higher the Q the lower the loss in the coil: Q = X / R, with X (inductive reactance) a function of coil inductance and frequency. Since R = X / Q it is a simple matter to calculate the ESR (equivalent series resistance) of the coil for known values of X and Q. As Q declines, R increases. As R increases so does the power dissipated by the coils, lowering performance and limiting high power operation.


Notice the current profile when coils are inserted. In a short element (50% full size shown above) the current decreases only a little from element centre to the coil, then sharply declines to zero at the element tip.

Coil inductance, Q and radiation resistance

The broadside gain of a dipole in free space is ~2.15 dbi. Even with perfect (zero loss) coils the gain decreases as the element is shortened. That power goes into broadening the pattern. Gain will decline further due to coil loss and conductor loss. The latter is negligible for elements made from aluminum tubing, so it is the coils we must optimize.

The radiation resistance of a λ/2 dipole declines as it is shortened. Since the (loss) resistances of the coil and conductor are in series with the radiation resistance, as the dipole gets shorter the loss increases. The matching network to transform the net feed point impedance to 50 Ω also has increasing loss, although this is not addressed in my model. Total loss sets the practical limit to how short an element can be made and still have acceptable performance.

Capacity hats and linear loading also introduce loss, an amount that depends on configuration and, again, on how short the element is made. There is no free lunch. Managing loss has a cost. Our job is to optimize the design to maximize performance and minimize complexity and cost.

As loading increases the antenna bandwidth also declines. This is mainly due to the lower radiation resistance, whereby the impedance Z = R+Xj changes more rapidly as X comes to dominate R. That is, the rate of feed point impedance change is more rapid as the frequency of operation moves away from resonance (Z = R+0j).

When the element is incorporated into a yagi, already a high-Q antenna, the problem can become worse. Since that is my ultimate goal I have an incentive to find a design that is not too long and not too short, but just right.

Putting it all together

With all the design components in place we are ready to run the numbers through EZNEC. In this way we can develop a picture of what to expect from an inductively-shortened 40 meter tubing element. As mentioned earlier, element diameter is fixed at 25 mm and the position of the coil is always at the midpoint of each dipole leg (half element).

For those unfamiliar with how coil Q is affected by its construction I strongly recommend you read what W8JI has to say on the topic of loading inductors. There is no need for me to reiterate what he so describes so well. You can also look at VE6WZ's designs to see what a high-Q loading coil for a low-band yagi looks like.


The chart includes the case of a lossless coil (Q=∞). Although not physically attainable, it shows the maximum gain that a shortened dipole can achieve. W want to get as close as possible to that value with a realistic design. As stated earlier, short dipoles have less than 2.15 dbi broadside gain just because they are short, irrespective of loss. I stopped shortening at 40% of full size since at these short lengths the coil loss becomes unmanagable.

We can summarize what this chart tells us:
  • Coil Q is less important when the element is close to full size. That is, the loss may be negligible. So build the coil more for endurance than high Q,
  • High coil Q cannot prevent large loss when the element is very short. Q=800 is approximately the best we can attain for a practical coil in this application, and Q=600 may be best achievable.
For example, a 50% size element the loss is about -0.5 db with Q=400 coils, and gain with respect to a full-size element -0.3 db due to its shortness. Net gain is therefore ~1.35 dbi.

My own rule of thumb is to keep the net gain above 1.7 dbi, which allows for -0.45 db due to a combination of coil loss and element shortness. When elements above this value are incorporated into a yagi the bandwidth and performance can usually be successfully managed. The Cushcraft XM240 elements are just above this cut-off, with elements ~65% of full-size and a coil Q of ~200.

If your performance objective is more modest a low target may be justified. It's a personal choice, provided the loss does not grow so large that operating QRO becomes a risk. More on this below.

Implications for high power and trap antennas

To be specific about loss let's take an example. You are running 1,000 watts to a 70% size dipole whose coils have a Q of 100. The gain is -0.43 db with respect to one with perfect coils (Q=∞). Coil loss is therefore ~100 watts, or 50 watts per coil.

Depending on the mode and the weather this can easily damage the coils and render the antenna inoperable. Consider that a coil with Q=100 is typically close wound with enamel wire on a solid dielectric core. It may not be easy to shed that much heat in many ambient conditions, especially when combined with a weatherproof coating encasing the coil.

The similar situation applies to trap antennas, including tri-band yagis. On bands where a trap passes the RF it is not doing so without effect. On those bands the traps behaves as an inductor, and all of the above discussion applies. Trapped elements are shorter than full size because they are inductively loaded dipoles.

A tri-band yagi element on 20 meters has both the 10 and 15 meter traps behaving as inductors. The Q of those inductors is perhaps no better than 150. The loss can be significant, limiting gain and dissipating substantial heat. It is for this reason the 15 meter trap on Hy-Gain tri-band yagis rated for maximum legal power are wound from copper rather than aluminum. The lower resistance is needed on 20 meters, not 15.

Construction issues (real coils)

We cannot always design a coil for highest Q. There are practical limits, the most serious being fragility. In climates like mine the effect of ice (freezing rain) is an ever-present hazard. Also consider corrosion and fatigue from bending in the wind.

It must also be built around on outside a non-conducting structural member such as fibreglass which is needed to place the coil in series with the element yet maintain the yagi's structural strength.

These can be dealt with, though as the coils become larger so do the challenges. The coils on an XM240 are not high performance, but get the job done at a modest loss. Think about the trade-offs if you ever go this route. Again, pay close attention to what W8JI has to say on this topic and what VE6WZ has achieved with the coils he's built for a hostile climate.

Last, there is no good way to model a tapered yagi element with loading elements using NEC2, even with the Leeson correction. Be prepared to tune the elements, individually, on the tower. Or invest in NEC4.

My next steps

At some point, perhaps over the winter, I will model a few yagis with coil-loaded elements and see what I can come up with. This has been done before, though not by me and perhaps not with the same set of performance criteria. Even if I go no further than models there will be something new to learn. As usual I will share that learning on the blog.

I will also share my thoughts on other element shortening techniques. There are some I like and others that I do not. In the end I may yet opt for a full-size, 3-element 40 meter yagi, or go with one of the large wire yagis I've previously described.

Thursday, November 12, 2015

Generations of CW Keyers

Keyers have been on my mind lately. Now that I am intending on operating less QRP I have to change my CW setup. For QRP it was quite simple: use the KX3 internal keyer to send manually and use macros within N1MM Logger+ to send "KY" commands for contest QSO exchanges. That works well, but only for that one rig. To use my FT-1000 MP in contests requires an external keyer that can integrated with contest software. It would be a solution that is transceiver independent. This becomes important for when I eventually buy one or more modern high-performance transceivers for my future station.

To that end I recently purchased a K1EL WinKeyer USB kit. This product has become the keyer of choice for many contesters, and for good reason. The kit is easy to put together, only taking 2 hours of my time to build and test. It took a little longer to puzzle out the documentation to get it properly configured and integrated with my shack laptop PC. I can see myself buying more, as needed, when I build my next station. One is enough for my current modest setup.

The distant past

Electronic keyers have been around for decades. The earliest ones appeared in the tube era. They were simple devices that used a single-lever paddle. Push it in one direction to generate dits, and in the other direction for dahs. Proper timing of code elements was the responsibility of the operator.

By the 1960s the dual-lever squeeze key arrived. This offered improved ergonomics, though nothing more at first. Eventually the iambic keyer emerged to fully exploit the dual-lever key, where closing both contacts sends dits and dahs in sequence. With the introduction of logic chips, flip flops were used to allow element buffering and improved inter-element timing schemes. Sending fast, clean CW with reduced operator fatigue was a revelation to those of us entering the hobby in the early 1970s.

It was also a boon for CW contesting. Soon enough the appearance of reasonably-priced memory chips drove the introduction of memory keyers. The ergonomics of recording messages was at first dreadful. This was overlooked by contesters such as myself since the pain of programming the keyer was more than compensated by the improvement in contest scores.

Accu-Keyer and Accu-Memory

As a relatively new ham and budding contester in 1975 the series of QST articles for the Accu-Keyer and Accu-Memory peaked the interest of me and a few friends with similar interests. We purchased the boards from the authors and built these units. We were very pleased with the results. Contesting became a lot more fun. The authors turned the articles into commercial products that did well for a time.

We were teenagers at the time and our skills at building electronics were not yet very good. He (VE4VV) used a Nye Viking squeeze key and I bought the venerable Brown Brothers dual-lever paddles; this was before Bencher paddles were introduced. I still have those Brown Brothers paddles and I still use them on occasion. The Accu-Memory itself was retired to the junk box in 1985.

The photo shows the inside of the Accu-Memory keyer I built, complete with AC power supply. It could key both solid state and tube rigs by choosing the output jack. The push buttons I used were awful. You make poor choices like that when you have little spare cash. The screw on the top cover prevents the front panel from bending when the operator (me) bashes those buttons in a futile attempt to make them work.

Only about one year later I learned the art of designing digital circuitry at university and discovered the horror that was the Accu-Memory design. The way they reduced the part count was by carefully engineering signalling delays between gates. The RC timing circuits on gate outputs are mandatory! With them they exploited race conditions that would otherwise require additional digital stages. While the criticism is valid I must admit they were inexpensive and worked. If there were glitches it sometimes helped to try different chips of the same type.

CMOS Super Keyer

When I bought a house and built a modestly competitive station in 1985 I upgraded everything in my shack. Looking for a better keyer design I found what I wanted (again) in QST. This was the CMOS Super Keyer by KC0Q and N0II in the October 1981 issue. The keyer evolved several times since and appears to have had considerable commercial success. Like the Accu-Memory this design also used discrete logic chips, except that they were better. CMOS allowed for low power consumption and therefore battery operation. It had a few new and interesting feature, though it was still just a memory keyer.

The keyer could run on 3 AA cells for years, without the need for a power switch. Constant power was necessary to ensure the dynamic memories would not be erased. Idle power consumption was advertised as typically 10 μA. My ammeter had a resolution of 10 μA and the keyer I built read zero. The first set of batteries I put in lasted for 7 years!

Unlike the Accu-Memory this keyer did not depend on race conditions. The design was much cleaner. The board was of the plug-in type, which made service simpler and avoided rat's nest wiring to other components. The power consumption was so low that the memories would stay active when the board was pulled for many minutes, despite the lack of power to the dynamic CMOS memory chip.

The buttons I used on this keyer look similar to those on the Accu-Memory but are in fact much better. I had the money by then to do it right. The buttons and switches are labelled, a luxury I skipped on the Accu-Memory. After all these years it's unclear what the buttons on that older device do. Assuming it still works it would require some experimentation to discover their functions.

This keyer continues to work well, and I actually used it in this past weekend's ARRL Sweepstakes contest. However like other keyers of its generation it is difficult to program. The character timing during memory playback was not perfect. It also did not help that for speed control I used a rotary switch rather than a potentiometer. It's an old technology solution to allow definitive steps to speed changes. The 12 position switch does not allow for sufficiently fine speed adjustment.


The keyer includes positive and negative keying circuits. This was superfluous even in the mid-1980s since by then I, like many hams, only used solid state transceivers. I did build in support for keying of multiple rigs (bottom left rotary switch). Modern technology integrates that function with the computer for SO2R contest operation.

WinKeyer

Last week I decided it was time to move on to the latest in CW keyer technology. Several generations of technology had passed by while I was out of the hobby from late 1992 through 2012. The CMOS Super Keyer, although excellent for its time, does not withstand comparison to what is now available.

The K1EL WinKey is now a part of my station, as seen in the photo above -- the older keyer is in the frame only as part of the photo-op. It has many features of value to those of us that operate contests. Best of all is the ability to integrate with contest logging software and general logging software. Placing the buttons on top is also a plus since there is no danger of pushing the keyer backwards when manually playing the memories.

With it there are no timing issues from using PC generated CW, since most desktop operating systems deal poorly with real-time applications. I had considered this option when I recently modified my FT-1000 MP. Memories and speed control are integrated with logging so that your hands might never need to leave the keyboard during CW contests. Operator fatigue is lessened, thus improving enjoyment and results.

I plan the first serious use of this keyer during the upcoming CQ WW contest. Until then I am experimenting with its features as I use it for daily operating. It's proceeding well so far, though I have had some difficulty getting a few questions resolved with the documentation. One problem I did encounter is that I've run out of USB ports on the PC. It only has three and I already use all of them. I unplugged the KX3 to make room. I may eventually need a USB hub to support the increasing number of external devices connected to the computer.

I don't know how much further CW keyer technology needs to advance since there may be little more of real value to be had. Most of the advancement is now on the receiving side, including skimmers and readers. Perhaps this will become the final generation of keyer technology in our hobby.

Wednesday, November 4, 2015

80 Meter Vertical Follow-up

Following the CQ WW contest I made a few small changes to the 80 meter vertical. These should be sufficiently interesting to others to justify this follow up.

Radial spacing

The 8 radials were not equally spaced when the antenna was first built. I was in a rush and so used a protractor to estimate 45° intervals (8 x 45° = 360°), and at least ensuring that opposite radials were aligned to be 180° apart. To set my mind at ease I decided to do a better job of it. The radials are easy to move since they lie on top of the sod.

Given the radial length and radial number it is easy to calculate and lay them out symmetrically to ensure equal currents. The process is described in many amateur radio texts, and is in any case easy to derive if you have a rudimentary knowledge of trigonometry. What you do is use an equation to calculate the distance between the ends of adjacent radials. I derived it since it was easier at the time than hunting around for the right book and searching for the equation. For my 8 radials of 8 meters length the distance is 6.12 meters.

With than distance calculated you start with any radial and a tape measure. Take the adjacent radial in hand and find the point where the radial end and measured distance coincide. Repeat for the remaining radials. To be on the safe side check that opposite radials are collinear; this requires an even number of radials.

After this task was completed I checked the SWR. Resonance had shifted upward by 50 kHz. This was not unexpected since published experiments confirm that symmetry affects radial current equalization, an effect that increases in magnitude when the number of radials is small. Ground loss is minimized when radial currents are equal.

Re-tuning the L network

Although the radial symmetry effect did not require retuning the antenna it did become a concern when the second inverted vee for 40 meters was put up. Due to the interaction the 80 meter SWR increased to between 1.7 and 2.0 from 3.5 MHz to 3.8 MHz. There was no time to remedy this before CQ WW. For that contest the problem was not significant since I knew I'd be making very few 80 meters QSOs with QRP SSB.

The L network box was unsealed and inspected. For a change it was dry inside. The weatherproofing is working.

I knew that to lower the frequency of minimum SWR I'd have to reduce the value of the series capacitor. I disconnected the three 82 pf capacitors that parallel the 500 pf doorknob capacitor, keeping two of them in parallel so that I had 3 combinations of capacitance to try.

To adjust the coil without unsoldering the tap I compressed the coil to increase the inductance. From playing with TLW I expected this to be the required direction of change. This turned out to be the case. I folded up short lengths of cardboard to adjust the compression, although any dielectric material can be used. A business card separates the coil and coax connector.

By trying various combinations of capacitance and coil compression I was able to find one that resulted in an SWR below 1.5 from 3.5 MHz to 3.8 MHz.

Mission accomplished I resealed the box and put the finishing touches on weather sealing the boot on the coax connector.

Common mode woes

There is common mode incursion into the shack from the vertical, just as there was with the loaded half sloper. The only place I've noticed it is the behaviour of the Ham-M rotator control head. When I had the half sloper the lamps in the meter would dimly light when running 100 watts. The vertical affects the meter deflection.

As discussed in the original article I knew very well that I'd have common mode problems due to the overhead runs of coax and control cables from the tower to the house. The coil I made with the extra coax to the vertical feed point doesn't accomplish much, just as I predicted. I decided to conduct an experiment.

I dipped into my junk box and pulled out two snap-on ferrite cores. Over the years I have kept some of these around to deal with EMI on stereos, telephones, televisions and other household appliances, either in my house or those of my neighbours. I was able to fit 4 turns of rotator cable through the cores, placing this choke coil near the cable entrance to the house.

The result was...interesting. I didn't expect a miracle, since this choke is not really adequate for 80 meters, only to see if it could achieve some common mode suppression. It didn't. Rather than reduce the problem it just changed it. Meter deflection continued as before, except at different frequencies -- the rotator EMI is frequency dependent. Sometimes the meter moved left and sometimes right, sometimes a little and sometimes a lot.

My conclusion is that the common mode is also coming in on the outside of the coaxial cables (RG213), so that the choke on the rotator cable only serves to modify standing waves on all the cables. I have no good solution to this other than a massive effort to add chokes on all cables that are suitable for 80 meters, or to bury the cables. Neither is an option in my present situation. I can leave with it. However, running a kilowatt would change my opinion.

I can of course use capacitors on all the rotator cable wires to bypass RF to ground. That won't solve the common mode problem but could make it less visible; out of sight, out of mind.

Ready for winter

Other than some cleaning up I am done with antenna work for the season. The weather is unseasonably warm this week, which is great for antenna and tower work, except that I don't need it. Everything is where it needs to be, performance is as good as it will get and all connections are waterproofed. Now I need only work on the shack electronics and software in preparation for upcoming contests and DXpeditions.