Building the PICaYAGI self-tuning HF Beam - 20m to 10m band. Part1

© G3XJP 2011 PICaYAGI - Part 1 Page 1 v10 of April 6, 2011

ONCE UPON A TIME Yagi and Uda invented their beam. Actually, it was in 1926.

We were soon building monoband Yagis for 20/15/10m and shortly thereafter the popular trapped 3-element tribander. Then along came the WARC bands and since then we have been struggling for a viable strategy for a single beam to cover all five bands. Several different commercial approaches have evolved over the years but none of them meets my requirements - for reasons discussed shortly.

PROJECT SUMMARY.

This is a constructional project to build your own 5-band (20m-10m) Yagi that automatically grows and shrinks its element lengths as a function of your Tx frequency.

This is intended as a repeatable design but by the very nature of mechanical construction, I would expect significant variations in implementation. This is not therefore a prescriptive approach.

You may need to choose different materials (eg tube sizes) and you may think of better engineering implementations. To this end, some of the blind alleys and false trails are also mentioned since there is rather more learning to be derived from mistakes than from success.

All my previous PICaPROJECTS [1], were rigorously tested for reproducibility by multiple builds prior to publication. This one is different. In fact as of Spring 2011, my PICaYAGI is the only one in existence. So although the formal engineering drawings have been rigorously checked for accuracy, this can never be a substitute for verification by building. Further, this design uses some novel concepts which inevitably increases the risk. So although mine works brilliantly, there can be no absolute guarantee that this is reproducible. There is much good faith and conviction - but no warranty. This project also has attitude! This is homebrew amateur radio so the project objective is about more than just acquiring a beam. It is also about increasing personal skills and “learning through doing”. And about improvising and value for your money. And absolutely about increasing your station’s capability. But ultimately about the incremental pleasure and satisfaction that derive from using anything you have made yourself. It has been an incredible source of learning for me and was chosen for that very reason and also because I wanted a decent beam to complete my home-brew station. Like many people, I have dabbled with EZNEC and built a few sundry beams in my time. I thought it would just be a matter of joining up the dots to discover if this one would work because I didn’t want to invest in all that aluminium if it would not. Suffice it to say that after 12 months of trying to understand if it would even work at all, I still didn’t know for sure and I had to buy the materials, build it and use it before I found out. So even if you are not contemplating this design, the lessons are completely general.

YOUR INFRASTRUCTURE REQUIREMENTS.

As in all the other PICaPROJECTS, the need for any serious workshop technology has been carefully avoided:

· No welding, lathes, mills etc.

· There are some GRP (glass reinforced plastic aka fibreglass) parts to be fabricated - but nothing so serious that they can’t be addressed with a typical “car repair kit”.

· You will need an old PC - and the means to program a PIC. The software itself is provided as both source and object code at no charge - strictly for your personal amateur use.

· You need the ability to shrink heatshrink tube.

· You need a cheap pop rivet gun.

· Some RF performance modelling software is useful - and the best in my experience is free.

· In previous projects [1] you will find several different ways of making your own PCBs, any of which can be used here.

To summarise, this antenna can be built entirely from your own amateur resources and you don’t need to pay anyone else to do any of it for you. So I won’t be authorising any commercial PCBs to circumvent the project objective, though there may possibly be scope for bulk buys of raw materials.The latest software and the latest updates are available at [1] with general build advice and support on the PICaYAGI Yahoo Group [2].

ACKNOWLEDGEMENTS.
This concept has been evolving since November 2008 on the PICaYAGI Group. This has a large number of members a few of whom have been active and completely indispensable. My special thanks to Dave G3SUL, Harold W4ZCB and Chris Stake. And not least Fran, who has been frequently called upon for a second pair of eyes and hands and her ability to proof read stuff she doesn’t claim to understand to see if it makes sense.

MY PERSONAL REQUIREMENTS.

Like everyone else I want multi-band coverage but, being greedy, I also want mono band performance. 
Actually, I want better. I want monofrequency performance. That is, a beam tuned for the spot frequency I’m on at the time, not compromised by the need for coverage of an entire band. But for me, the dominant consideration has to be visual impact and at least in the UK, I suspect it is for many others too. Frankly, I don’t want some monster polluting my skyline and I can’t imagine why any of my neighbours would either. So I guess my fictional engineering figure of merit would be along the lines of “dB performance per metre of obstructed sky”.

HOW DO OTHERS DO IT?

Always a good first question! Taking the 3-el Yagi as a performance datum, there are several completely different strategies for 5-band beams on offer:- One approach is the interlaced 5-band 2-el Quad. Then there are log periodic arrays and log- Yagis. Various designs interlace multiple Yagis on the same boom, some also with log cells. A radically different approach is taken by SteppIR™ whose Yagi elements change electrical length as a function of frequency. None of these approaches satisfy my basic requirements. Why not?

· A Quad with spreaders long enough to cover 20m is a highly visible 3-dimensional structure.

· Any log periodic array gives relatively poor performance per element and so uses its quota of “obstructed sky” inefficiently. On that same scale, 5-band interlaced designs are even worse since most of the elements do nothing most of the time. So when you hear someone on 20m proudly announce that he is using an 11-el Yagi, I just hope he realises that only 3 of them are actually in use.

·The SteppIR Yagi is fixed in mechanical size (as opposed to electrical size) so whatever band you are on - and even when not in use - the visual impact is always worst case, namely that of a 20m beam.

For me, all these approaches deliver ~ 3-el Yagi performance but score poorly on visual impact. With commercial offerings, there are also significant value for money considerations and some performance issues. A glance at any price list shows the former is self-evidently an issue.

The latter will be addressed later. In this context, it is worth restating my personal ethos and hence this Project’s objective. By definition there can never be any commercial / packaged solution that meets my needs (and Licence requirement) for experimentation and selflearning.

But any temptation to purchase is easily resisted in this case since they don’t appear to meet my requirements anyway. And for less than 20% of the price, I can build something that does.

PICaYAGI An auto-sizing HF beam - Part 1 G3XJP/M during early proof of concept trials in Spring 2010.

© G3XJP 2011 PICaYAGI - Part 1 Page 2 v10 of April 6, 2011

DREAM ON.
For several years I have been idly contemplating ways of building an HF Yagi which grows and shrinks in size as a function of frequency. Plasma? Conductive liquids? A scaled up version of those evil party blow-out toys? In Sept 2008 with the prospect of some sun spots I decided to get full time serious. And with the end of 10 years of previous project support commitments, I was able to.

I want something about the size of a 10m Yagi when not in use (ie the vast majority of every 24 hour day) and there are three themes that were pursued in parallel and which needed to converge to turn the dream into reality. Changing element resonant frequency. I began by looking at the most obvious approach, namely telescoping aluminium tubing. The fundamental issue is that any rubbing aluminium surfaces will promptly seize. Any other metal is way too heavy and any metal-to-metal contact has a very short life outdoors. After much debate, Dave G3SUL came up with the brilliant idea of sing heatshrink tubing applied to an inner sliding aluminium tube as a dielectric layer. This provides a reactive coupling between the sliding and fixed tubes. Much engineering detail follows later but this has remained the approach ever since.

Frequency span is however a problem. The essential issue is that from the bottom end of 20m to the top end of 10m is more than a 2:1 ratio. And the nature of any telescoping tube arrangement (given an overlap) is that the length range is less than a 2:1 ratio. The reality is worse than that because the smaller diameter inner tube increases the resonant frequency as does any element droop. And dielectric coupling reduces the effective length near full extension also. So for many months we looked at telescopes within telescopes and some extraordinary pulley arrangements to drive it all in and out. None of it felt engineeringly realistic. At about this time I fixed on a Yagi with a fixed length boom and with fixed element spacing - simply to bound the task.

Figure 1 shows the chosen tubing configuration. The essential approach was to make the fixed inboard length of each element from two ~4m tubes, overlapped and clamped together and fixed to the boom. This gives a net length of just over 2m per side - which can be pre-adjusted by altering the overlap to make a director, a reflector or a driven element for the top end of 10m. Then with sliding inner tubes also ~4m in length and with a minimum overlap of ~1m, these elements can be extended to over 5m per side to hopefully cover down to the bottom end of 20m. Motive power. I started out looking at pneumatics. This has the appeal that air is invisible, not very heavy and comes for free. And I was already using compressed air for the mast - though that is not a prerequisite for others and certainly not part of this project. Ultimately I think I could have made it work but the problem lies in measuring the resultant element length. There is little virtue in formulae like “2.3 bars applied for 3.7 secs = QSY down 50kHz.” Stepper motors are an instinctive choice for repeatable positioning. But given the torque requirement, they were immediately ruled out as too bulky, too heavy and too expensive. On the basis of size, weight, torque, price and availability I finally settled on a £6 electric screwdriver from a UK DIY chain. I use one per element.

This has a 3V6 motor and an epicyclic gearbox delivering 1.5 Nm of torque. Measuring element length. My first approach was using sonar range-finding with a piezo sounder on each element tip and an electret microphone near the centre of the boom. Unfortunately the speed of sound in air varies significantly with temperature. The complication of adding a temperature sensor and all the inevitable calibration curves in the software felt entirely disproportionate. So I pragmatically abandoned this otherwise excellent bird scarer. Laser range finding is prohibitively expensive. The obvious approach was to attach some “string” to the element, spool it on a shaft and then count turns with a shaft encoder. Having made that leap, the obvious next step was to use that same string to also pull the element in and out and abandon the pneumatics. For the string itself, my first thoughts focused on beaded cord as commonly used on domestic window blinds. This is basically nylon braid with moulded nylon balls at about 10mm intervals. The idea was to drive it with a sprocket, much like a water wheel. The great virtue of this approach is that you can pass the cord once round the sprocket and the same length of cord is used for both IN and OUT purposes. In other words, there is no spooling requirement. In trials, I could not get it to work effectively because the diameter of the sprocket needed to grip several balls was too large - such that the screwdriver didn’t deliver enough torque. Also, fitting the ball diameter in the gap between the fixed and moving parts of the element was distinctly marginal. On the recommendation of Paul, G0ILO I went instead for braided fishing line. I tried several on the basis of their advertising claims but “zero stretch” proved to be somewhat overstated. Some of them appeared to just go on stretching forever! On advice from Steve G4ZBV I avoided anything based on Kevlar on the grounds that it is self-abrading - ie fine until you spool it tightly. Finally I got in touch with some serious kiteflyers on the West Coast and ended up with some Dyneema braided braid which is less than 0.5mm diameter and has a breaking strain of 80lb / 36kg.

That is incredible performance. It is also essentially (if not actually) zero stretch at the loads I am applying which greatly simplifies the inevitable software task of converting counted shaft encoder slots into a repeatable element length. This wonder braid is both readily available around the world and inexpensive.

SUMMARY SO FAR.

Having settled on the underpinning mechanical technology, there followed the engineering to make it work reliably and all the issues of RF performance. There were still several outstanding conceptual problems:

· Would this design approach produce enough change in element length?

· Indeed, how much length change do you actually need to achieve the frequency span?

· What sort of boom length would be needed to give good performance on all 5 bands?

· Indeed, what exactly is “good performance”?

· How on earth do you feed a driven element - which cannot be split across all 5 bands?

· Could it all be made to stand up in a wind?

Time for some antenna modelling!

Not to scale

Overlapped fixed tubes

Sliding inner tube

FIGURE 1: PICaYAGI tuneable elements - illustrated near full extension.

First installation on the pump-up mast in late Spring 2010.

The mast also helps with visual impact – and the castle on the hill in the background shows why I need all the help I can get. You can also see I have a bit of a ground slope challenge.

© G3XJP 2011 PICaYAGI - Part 1 Page 3 v10 of April 6, 2011

WHY MODEL AT ALL?

For the simple reason that it would be essentially impossible to find the optimum dimensions of the real antenna by cut and try. In this design, you start out with at least five interdependent variables - namely element lengths, boom length and element spacings. To say nothing of the feed arrangements. The task is to determine the best dimensions such that it will perform well over all five bands and at the same time, have enough mechanical strength to have a reasonable chance of survival.

The task is iterative and it would cost a fortune in time and materials to attempt on a real antenna.
Frankly, you would be unlikely to ever get there. Most likely you would probably produce something that “worked” but that was far from optimum. Worst of all, you would never know it.This topic is therefore particularly relevant if you are considering changing any of the element spacings or tube diameters and of interest to anyone designing or assessing any antenna. 
The subject of modelling splits by tradition into the two tasks of ensuring mechanical integrity and optimising RF performance. In fact, both affect both.

MODELLING MECHANICAL BEHAVIOUR.

How do the professionals designers do it? A commercial manufacturer cannot know eg the likely windspeeds or ice-loading at the customer site because obviously, they are all different. So they have to design to some established standards - which typically have generous margins for error.

These are typically national standards so there are then all the issues of certification if you want to market the product abroad. All this costs much money and it inevitably finds its way to the bottom line and the price.

For the one-off amateur it can be (and should be) very different. Round here for example, earthquakes, tornados, hurricanes and significant ice storms are all very rare events. Not unknown, but we are talking typically once in 50 years stuff here in up-state Herefordshire. Would you really want to design to handle them all? Because if so, you go very fast down the road of diminishing returns and end up using gargantuan tube wall thicknesses and need a beefy (but very low) tower to hold it all up - and a rotator that would start a jumbo jet on a cold morning.

A lot of these rugged designs originate in the USA where they certainly do need to design for significant environmental extremes. But there again, by UK standards they have some extremely tolerant planning legislation for amateur antenna installations. Pragmatically, the most likely outcome of designing to handle those same risk levels in most parts of the UK is that you would end up with no antenna in the sky at all. And indeed for some people in some circumstances, that may well be exactly the best outcome. Ultimately, only you can assess the risks and the consequences that you face.

The realistic consideration I believe is the price of failure. I’m lucky (actually it’s long term lifestyle choices) to have my mast in a field that is well out of range of any human activity. The only person that ever ventures into the drop radius of the beam is me - and I can choose when I do so.

But obviously, I still want to design for reasonable survivability of my precious antenna.

Needless to say, we are always trying to hold down the total weight, turning radius and crosssectional area in order to minimise cost, visual impact, wind loading, mast requirements and rotator torque. However, all this down-sizing tends to work against both RF performance and the ability to withstand ice loading. It is a vicious circle. I don’t have a wind-tunnel here and I’m also not inclined to add weights to the elements to discover when they would break under ice loading.

My pragmatic approach in the end was to design for RF performance using materials that were available and “felt” as though they might handle the stresses using the ARRL Antenna Book HF Yagi mechanical designs for guidance.

About the same time, Dave G3SUL pointed me in the direction of the basic equations for cantilevered beams. This is frightening stuff. Each half of an element is a cantilevered beam fixed at one end. The equations show that the droop of an element under its own self-weight rises as the cube of the length. So if the tube diameter or the wall thickness is inadequate, you can easily get to the point where as you extend the element, the tip is moving rather more down than out. It is how to make an inverted U. 
Anything even approaching this is a disaster for achieving the frequency span target, let alone mechanical integrity. It is why the element diameter is tapered down as we approach the tip.

We also need to find the optimum wall thickness since increasing it is a trade-off between adding strength but also increasing weight. After that, any ice loading just increases the weight and windage. At this stage I made the decision that I was only ultimately concerned with stresses on a parked PICaYAGI. That is, with the elements fully retracted. For visual impact reasons, I already have the discipline of parking the elements whenever it is not in use for any protracted period and certainly overnight. And my operating habits are such that if we have gales or any evidence of serious icing then I simply forgo the pleasures of the lower frequency bands.

The practical approach I use is to wind the elements out and if they start flaying about, I wind them back in and move up a band until they are not. You quickly learn to correlate the movement of tree branches with what is reasonable. You may consider this approach to be very amateur - and I would proudly agree. It is no more onerous than asking a sailor not to raise all the sail in strong winds and if gale force, to stay in the harbour. It happens very rarely. 
The upshot of all this is that the parked elements are comparable dimensionally with the design of the 10m Yagi in the ARRL Antenna Book, falling between their heavy duty and medium duty designs. The medium duty design will handle “wind speeds of 96 mph with no icing and 68 mph wind with ¼’’ of radial ice.” That will do for me! To balance the rotator wind torque, I also need to add a small torque compensator plate to the boom. It is so small, I haven’t yet bothered.
Finally, in order to reduce friction on the sliding tube, it is important that the fixed tube remains as straight as possible as the elements extend. To this end some vertical cord bracing is essential.

However this bracing has no other function when the element is parked than to withstand any ice loading. So that is indeed the icing on the cake!

RF PERFORMANCE MODELLING.

First we need to define some terms. I both regret and resent having to do this here - but 84 years since Yagi and Uda and despite heroic efforts by the ARRL, there appears to be little consistent practice out there. The three parameters we typically want to tune for are Gain, Front-to-Rear ratio (F/R) and SWR and guess what? These and the definition of the environment are all open to abuse. 
Environment. We can define our performance in hypothetical free space or at a specified height over specified ground. The merit of free space is that nobody has access to it so we are all equal.

Thus we can validly compare one configuration to another or one antenna to another on the basis of calculated performance. The merit of using real ground is that you can make real measurements and it is generally possible to determine if you have improved your antenna - but it is desperately difficult to define the measurement circumstances such that you can validly compare it with others. 
All PICaYAGI performance numbers used to evaluate concept feasibility were calculated in free space. All subsequent measurements of achieved performance were made under some precise circumstances I will carefully define in detail later. Gain. This is the performance in the main lobe relative to either a dipole in free space (dBd) or relative to an isotropic point source in free space (dBi). Either will do fine and one can convert using the formula dBd + 2.15 = dBi. The confusion creeps in when one unit is maliciously used in free space and the other over real ground and you end up comparing apples with oranges. Back and Rear. Wikipedia (and others) say Front-to-Back and Front-to-Rear ratio are the same thing. Not here they are not. 
PICaYAGI at full element extension illustrating modest element droop.

© G3XJP 2011 PICaYAGI - Part 1 Page 4 v10 of April 6, 2011

I use the definition in the ARRL Antenna Book. Front-to-Back ratio (F/B) is the gain ratio between the main lobe and the lobe at exactly 180° off the back. By contrast Front-to-Rear ratio (F/R) is the gain ratio between the main lobe and the worst lobe anywhere to the rear. Both are measured in the same plane as the gain. F/R is the more strenuous criterion to meet and is always the one used here. A plot for a typical 3-element Yagi is shown in Figure 2. This antenna has been tuned for good F/B and it achieves a very attractive sounding 30dB. This tremendous performance straight out the back is always accompanied by unacceptable rear lobes in other directions. The F/R is a much less

exciting 18dB. Except for specialised applications, quoting only F/B is misleading and unhelpful. Not least because QRM rarely chooses to be on exactly a reciprocal bearing.

SWR. The other modelled performance parameter that must never be taken for granted is SWR. One behaviour that modelling any antenna teaches you is that if you allow the SWR to drift up even slightly then it is possible to produce significantly better apparent numbers for the gain and rear performance. Provided that the beam can indeed be tuned for unity SWR at the chosen feed impedance then I don’t accept anything over 1.01:1 as a mandatory maximum. If it’s more than that I just keep on optimising until it isn’t - while typically watching the gain and rear performance fall way. That way the results are validly comparable with others. In real life, I’m perfectly happy with modest SWR on the feeder but in trying to optimise a design, the practice of letting the SWR drift up is one best confined to the production of doubtful advertising copy.

IS MODELLING EASY?

In his penultimate In Practice column [3], Ian White GM3SEK said that one (of many) aspects of an antenna design that we need to consider is "Ease of computer modelling and whether a design can be converted into real-life hardware with a minimum of uncertainty." Although Ian was referring at the time to VHF/UHF long Yagis, the sentiment is equally true for any Yagi.

However at VHF/UHF, the task is more straightforward because you mostly use straight untapered elements. There tends to be a lot more of them of course, but that only increases the scale of the modelling task, not its validity. These elements don’t fall into the grey zone of what the modelling software will validly handle. This got me thinking. Could it be that most of the popular HF Yagi designs are what they are only because they could be easily modelled? Is that why they are so unadventurous? To advertise with the ARRL [4] you need either certified results from an antenna range or instead, performance figures derived from either YO (Yagi Optimizer, for Yagis only) or the latest version of NEC. One can only applaud their desire to eliminate misleading advertising claims.PICaYAGI can be validly modelled with either of these packages but easy it is not. Why? Because literally every unusual aspect of this design falls outside the scope of what these packages will handle without significant creativity. 

I confine this discussion to the use of NEC2 since that is what I actually used. It is free and later versions (which are not) offer little if any incremental benefit for Yagi design. The trap for the unwary is that NEC itself will not prevent you from entering pretty well any complex structure you like. All it needs is the radius of all the tubes and the co-ordinates of both ends. You can add reactances and a feed point and other forms of discontinuity. Having entered all those physical parameters, NEC will then do the sums and give you the answer. This is great, but in the case of PICaYAGI or anything else with “unusual features”, you would be well advised not to believe it.

This is not a criticism of NEC in any way. It is great - if used carefully and within its specified limits. All mathematical models depart from reality in some ways and the NEC documentation goes to a lot of trouble to point them out. The problem is that if you model PICaYAGI literally, you break most of the rules and produce some spectacular numbers and in many cases, no warnings. So regrettably it doesn’t meet GM3SEK’s criterion for “Ease of modelling”. It is rather difficult and this is why:

PICaYAGI MODELLING HAZARDS.

These are the circumstances where the physical dimensions can’t be entered into NEC literally. Element cross section. 
As shown in Figure 1, the centre of the elements has an oo section. This is simply not allowed and you have to replace those two clamped tubes with one round tube of somewhat larger diameter. How much larger? Tapered elements. 
NEC is notoriously inaccurate in modelling any discontinuities in element diameter. 

PICaYAGI has a large % step in diameter where the sliding tube enters the fixed tube. In general, there is an excellent way to handle this called Leeson’s Correction. It replaces all the tapered tube lengths with one of the same total length but with one equivalent diameter. But Leeson’s Correction has several restrictions and infringing any one of them prevents us from using it.

Needless to say, PICaYAGI infringes the whole lot. Specifically, you are not allowed any reactive loads on the element and the elements must be perfectly straight. No droop!

Element droop. This causes a problem in its own right since if you model it as a series of wires at increasing angles, the model rapidly becomes unstable producing wild swings in output for small changes of droop.

Feed point:
My tentative idea for the feed arrangement has a fine and detailed structure compared to the elements - and uses differing diameter tubing at apparently arbitrary angles to the driven element. There are precautions you can take with careful segment alignment but you are taking a big chance.

So how to get around all this? The only way I know is to take the real element, put it up in the sky, measure its self resonant frequency - and then enter it back into NEC as tubing of the same length but with one net diameter that resonates on the same frequency. This is in effect manual Leeson’s Correction but without the restrictions. The photograph shows some early measurements being made on my driven element. These were later refined at the intended install height.

You will have spotted the catch. You have to commit to the materials to build the antenna to build the model to prove the antenna will work. In reality, you creep tentatively up to that point with an ever increasing sense of confidence. Finally, having built the model you can and must run validity checks to test a) if you have got adequate segmentation and b) that the sum of the power radiated in all directions is equal to the input power. These tests give you significant confidence that the modelled performance results can be trusted.

135, 120, 105, 90, 75, 60' 45, 30, 15, 0, 345, 330, 315, 300, 285, 270, 255, 240, 225, 210

195 180 165 150 -3dB -6 -10 -15 -20 -30 Back -40 Rear arc 180°

FIGURE 2: Performance plot of a 3 element Yagi tuned for best F/B  at the expense of F/R. Scoping the frequency span of the driven element.

© G3XJP 2011 PICaYAGI - Part 1 Page 5 v10 of April 6, 2011

CHOOSING A MODELLING PACKAGE.

There are many PC packages around which provide a user interface to the ubiquitous NEC core - with much variation in price and facilities. I am indebted to Ray, WB6TPU for pointing me towards 4NEC2 by Arie Voors [5] which is an absolute delight to use. Arie has devoted countless years to getting this package right and he makes it freely available to us in the true spirit of amateur radio. The critical 4NEC2 feature from this project’s perspective is the inclusion of a genetic optimiser. Most traditional antenna optimisers use a hill climbing approach to optimise your specified variables (eg element lengths) against your chosen criteria (eg Gain, F/R, SWR). They will indeed get you to the top of the hill. But what they don’t tell you (simply because they don’t know) is that it may not be the highest hill in the mountain range. By contrast, a genetic optimiser behaves as Charles Darwin decreed and it throws off random mutations (aka sports) to see if they lead anywhere useful. So your chances of ending up on the summit of Everest and not K2 are greatly improved.

The only practical downside is that this does take much longer. I have spent literally months of nightly runs on the PC in search of the very best answers. Having posed the right questions, at least you don’t have to actually be there while it finds the answers!
But exactly what are the right questions? Because if you don’t ask the right ones, the answers will mislead you. It would appear there is some significant ambiguity out there.

TUNING FOR PERFORMANCE.

The crunch question is, what exactly do you want to tune for? Very rarely do best Gain, F/R and SWR coincide. What I want is my Yagi optimised for good performance in every direction other than that of the station I am having a QSO with. It might sound strange put that way, but I know that there is almost nothing I can do which will influence the strength of the signal I can receive or transmit in the main lobe off the front. But off the rear, that is another matter. On 15m for example, I could indeed trade 0.05 dB gain improvement off the front for 11dB degradation off the rear.

Contemplate that horrendous and typical tradeoff. It is inherent in any and every 3 element Yagi.

I live near the NW edge of a small rural village. When beaming away from the village to the USA, my receiver noise floor is some 9dB better versus beaming into the village. This is far more important to me than any fractional dB gain improvement in the main lobe. Further, much of the literature refers to the wonders of beam performance on transmit but mentions nothing about the mayhem you can cause to others if you tune for maximum gain at the expense of rear performance. And for what? Next to nothing. 
I regard it as precisely analogous to tuning your linear for 0.05 dB more output at the expense of 11dB increase in IMD products. Am I missing something here? I claim no expertise and I really would love to know. There are indeed occasions where the facility to listen off the rear and even at times to transmit off the rear would be useful. But that is a story for later.

In fact, PICaYAGI can store up to three tuning solutions per frequency so it is entirely possible to retain Best F/R and Best Gain as alternatives. The only time I use the latter is if the band is closed behind me and I’m working a station who is having trouble copying me. I can’t say it gives much obvious improvement and generally I would prefer to suppress my local noise floor which is there 24 hours a day. And I have a quiet location.

Finally in this context, there is one significant advantage we have over most of the antenna manufacturers. They cannot predict the customer’s installed height nor their ground characteristics. Since most of these offerings are designed to be used essentially “out of the box” they need to adopt a relatively low Q approach so that the antenna will deliver acceptable performance over a reasonable range. This is a compromise we don’t have to make since ultimately this antenna is tuned from the shack at your installed height and over your ground. So we can indeed design for better mono-frequency performance. Which is not only better, it is easier!

PICaYAGI MODELLING PROCESS.

For a multi-band design the process is somewhat iterative. The first task is to find the best values for the relatively uncritical variables. That is, boom length and driven element spacing - which are indeed uncritical by comparison with the element lengths. This entails modelling all the variables over a plausible range of boom lengths and spacings for at least a few representative frequencies in each band. And certainly including the bottom end of 20m and somewhere near the top of 10m.

If building a mono-band design, there are some well established guidelines for boom length.

So a good starting point is to choose a 3 element boom length which is about right for 17m, use it somewhat short on 20m and increasingly long on 15/12/10m. I finally settled for 4360mm (approx 14.3 ft) mostly because I already had a suitable pole of that length. The other observation from the modelling is that the position of the driven element on the boom is not critical and anything from about 1/3 to 2/3 from either end is fine. The classic issue with a short boom is the narrow SWR bandwidth. But with the element lengths adjustable, this is a non-problem in our case. On the other hand, a boom that is too long gives poor F/R and there is no obvious way round this.

Much trawling of the web looking at other solutions gave nothing helpful. For example, the FAQ on the SteppIR website [6] quotes 10 to 20dB F/R on 12m and 10m for 3 el and their brochure quotes F/R of 15dB on 12m and 11dB on 10m.

This is with a 16ft boom even longer than mine. Because this is nothing like the sort of performance I wanted off the rear, I decided to add a fixed length element four (known as the ELF) with switched stubs for 12m and 10m only. This makes my 4360mm boom just about perfect for these two bands as well and what’s more, it delivers 4elements

performance. 
Having chosen the best boom length and element spacings, the final task is to determine if the elements would tune from the bottom of 20m to somewhere near the top of 10m and all the modelling indications were that it was distinctly marginal. So a switchable linear resonator was added to both the main director and to the reflector to increase their effective maximum lengths by about 1.5%. That may not sound much but it is 200kHz on 20m. I don’t like linear resonators for large changes in resonant frequency because they radiate and potentially upset the pattern; and they have high circulating currents which can increase the losses. 
But a small one is harmless enough. And by luck (not by design) the ability to switch the linear resonators On/Off gives some valuable instant pattern switching options. The final configuration is shown in Figure 3.

This also defines the names of the elements which I‘m going to use from now on. (For the User Interface I need some distinctive and intuitive three letter abbreviations with a unique first letter.)

You can also see the linear resonators on the REF and the DIR, the anti droop cords on the three main elements and an early glimpse of the matching arrangements. More detail on that and the electronic aspects next month.

REFERENCES / WEBSEARCH

[1] http://groups.yahoo.com/group/picaproject

[2] http://uk.groups.yahoo.com/group/PICaYAGI

[3] RadCom, June 2010

[4] http://www.arrl.org/advertsing_opportunities# Acceptance_Policy

[5] http://home.ict.nl/~arivoors/

[6] http://www.steppir.com/FAQ.html

REF

FED

Boom 4360mm

Spacings from REF :

FED 1524mm

ELF 2823mm

DIR 4320mm

DIR

Not to scale

ELF

Weight 16.8kg

Turning radius:

Parked 3498mm

Max 6355mm

FIGURE 3: PICaYAGI general mechanical overview and significant mechanical specifications.