Analyzing the Wrong Things

I expect that everyone knows the following old joke. One night a passerby sees a drunk on his hands and knees searching the pavement at the base of a lamp post. He asks and is told that the drunk is looking for his keys. The passerby sees nothing on the ground and asks if he's sure that he dropped the keys there. The drunk answers, no, but the light is better there.

The joke resonates because, even cold sober, we have all done something similar. We look for answers where we are comfortable searching even if we strongly suspect the answer lies elsewhere. Worse, we may decide the answer in advance and search for affirmation.

Hams often fall into the same trap since, despite this being a technical hobby, many of us don't always have the required technical skills or tools. Simple and wrong answers abound, and in many cases you can still get by despite getting it wrong.

Probably the best known example is SWR. Decades ago when I was young and new to amateur radio, you tuned your tube transmitter for maximum "smoke" and hoped for the best. There was no easy way to determine if the antenna was resonant, presented a good match or met the advertised specifications. 

The only affordable and easily employed tool was the SWR bridge. A low SWR became synonymous with a properly working antenna system. That would be nice were it true, but it isn't. It's just easy to measure, like the drunk under the lamp post. So we convince ourselves that an SWR of 1 is the ultimate objective.

There are many similar examples that I can think of, and probably so can you. That's not the purpose of this article. Nor is it to ridicule hams for a lack of knowledge; no one can know everything. The objective should be for all of us to learn and, where appropriate, to teach.

Certainly we have the resources to do diligent product research and the money to buy good transceivers and test equipment. What makes them so good and how can we tell? If we don't have the knowledge to measure and understand, are we to trust marketing brochures or the opinions of other hams? What do we analyze and how do we analyze?

Let me give a personal example. Around 1990 I purchased my first PC, a 286 or 386 running DOS. Although I was a software professional with access to mainframes, I made the purchase for one reason: ELNEC. I had an abiding interest in antennas from the 1980s and the public availability of MININEC brought the ability to design antennas from first principles. I was eager to get started. Until then I would write my own software and spreadsheets to produce variations of standard yagi designs using formulae from W2PV ("Yagi Antenna Design") and others. My knowledge extended no further.

My most elaborate models had so many segments that runs would take an hour or more. That was no problem: I would start it before the evening meal and read the results afterwards. Although ELNEC/MININEC had its limitations that I regularly ran into headfirst, I still learned a lot and used it as the basis for 40 meter wire yagis and other antennas that were compatible with my modest station.

One of those antennas was an 80 meter half-sloper. I built and tuned it based on myth and lore, and I was pretty happy with the DX I could work. At the time (recently transplanted from VE4) the novelty of working DX on 80 meter coloured my view of how well the antenna performed and how it functioned.

I went ahead and built a model for it in ELNEC, hoping to gain insights that would help me to make it better. But this is not an antenna that models well with the MININEC engine. I tried to get around those limitations with load objects to simulate ground interactions, the coax run up the tower and the effect of yagis on top. Eventually I got an SWR curve that closely matched what I measured -- with an SWR/power bridge, not an antenna analyzer -- and performance that kinda sorta reflected my experience with it.

Years later when I fed the same model into the more accurate and comprehensive EZNEC and NEC2, it produced completely different results. My earlier trial-and-error approach to antenna design was exposed as an abject failure. Stripping off all the crud from the model resulted in a more truthful outcome that showed me just how poor the antenna really was. I had no idea what I was doing. The DX in my log didn't disappear but perhaps I could have done better with a different antenna, and by avoiding analysis errors.

When hams get excited about downloading the now free EZNEC (or one of the other free modelling programs) with stars in their eyes about the great antennas they were going to design, I politely try to inject a little reality. Software is not judgmental: feed it garbage and it produces garbage. EZNEC will warn you away from some common stupidities but most will pass without notice. It's a tool, not an engineer or a mind reader. 

Picking up a saw does not make you a carpenter, and antenna modelling software does not make you an RF engineer. Choosing the right tool and using it effectively requires knowledge and skill. Without those you will find yourself analyzing the wrong thing. But you probably won't recognize the error until the result does not meet the objective. At that point some will blame the tool or ignore the flaws and declare victory, as long as the SWR is survivable.

Learn the basics of antenna theory -- you don't need a university degree! -- and you can use modelling software more effectively. The resulting antennas will be more likely to match the designs. Do not ignore warnings and helpful suggestions that the software authors build into their products: they are trying to help you avoid mistakes due to the constraints and limitations of the underlying calculating engine, which they all have.

Example include: acute angles between wires, perfect capacitors and inductors, ground characteristics and connections, radial system, poor wire segmentation, common mode, interactions and much much more. If you can't learn the fundamentals, at least thoroughly read and follow the documentation. It can also be enlightening to talk to a fellow ham with the requisite expertise. 

Another suggestion is to download a reliably physical model that is similar to your idea to see how it's done and then carefully alter the design to get what you want.

An indispensable tool is a single-port VNA, better known as an antenna analyzer. Get one for which the accuracy is known and is easy to use in the field. I use a Rig Expert product since it's accurate and takes some rough handling for tower and field work. There are other products that work well. 

Accurate measurements of both R and X will help you find problems and make improvements, and will unearth model errors. Don't buy an analyzer just because it's affordable! Too many are inaccurate, especially as the impedance get further and further from the nominal 50 + j0 Ω. That flea market bargain can be like a hammer with a splintered handle on which the worn steel head wobbles. You'll only end up hurting yourself.

Another failure of analysis is reception; that is, using readability as a measure of antenna (antenna system) performance. For example, you conclude that since you can hear the DX that everyone else can hear your antenna system is doing well. It may seem reasonable but it is not a reliable measure of performance. Insert a -6 db attenuator or 6 db gain (pre-amp) into your HF signal path and, in most cases, readability won't change. The reason is that signal path gain affects received signal and noise and therefore does not affect SNR (signal to noise ratio), which is the ultimate measure of readability.

A more directive antenna -- typical for a higher gain antenna -- will improve SNR, but only when the noise is not in the same direction (azimuth and elevation) as the desired signal. You may hear a station as well as the big gun, but you will not have the same opportunity to work them; the big gun will have a stronger signal at the DX station.

Analyzing readability is the wrong approach since it does not measure how well the other station can copy you. An extreme example of this is a directive low band receive antenna such as a Beverage. You will copy stations well but you'd never use it as a transmit antenna because it is exceptionally inefficient, with a radiation resistance of perhaps 1 or 2 Ω. Its high feed point impedance is almost all due to ground loss. 

Antenna system issues that can reduce your signal's readability include: transmission line loss, antenna loss (e.g. thin wires, transformers, coils and traps) and radiation pattern. Some will dismiss these concerns because, say, the difference is only 1 db. One is a comfortably small number. 1 db is also a small fraction of an S-unit. But 1 db is close to 25%, and that does make a difference at the other end when their noise level is high or you are competing in a pile-up. It came be a grave mistake to make conclusions based on an analysis of S-meter readings and small numbers since they employ a logarithmic scale.

Contesters may be tempted to analyze their logs with a variety of free and paid tools to try and identify their strengths and weaknesses, or to track their performance from year to year. The trend towards public logs allow comparison to the logs of casual and elite participants. There may be insights to be gained. That is, if the analysis compares the right things, and knowing what those things are can be quite difficult.

This is a Cabrillo extract from my recent activity in the LZ DX contest. You can see that I was operating 2BSIQ at the time on 10 and 15 meters; that is, running on two bands. You can analyze my log all you want and all that you'll learn is that I had a good opening to Europe and that I am moderately proficient at 2BSIQ. The analysis won't teach you how to do 2BSIQ, only that there may be value in learning the skill.

There is also no indication of my multiplier hunting ability or strategy from this snippet since I was running at the time. Although running with a big signal is itself a strategy (since it attracts many multipliers and reduces the time devoted to hunting) you must still develop a strategy based on your station capabilities and propagation, which is rarely the same for two contests. Log analysis will also not tell you how to find the mults and quickly get through the pile up.

You could compare my log to that of another SOAB CW HP station to make a guess at my multiplier strategy and 2BSIQ ability. Does that analysis help you to do better? I suspect not. A more sensible analysis would lead you to conclude that you also need a good path to Europe and/or a big station, SO2R capability and that you ought to practice SO2R and 2BSIQ. 

Once you have those you can more effectively compare yourself to the best contesters when analyzing their logs. Don't analyze the logs of casual operators or those with far different stations since you'll learn little of value. Don't overlook error rates if they're available. Accuracy is another skill you won't learn by analyzing logs.

Rather than log analysis you might be better off reaching out to other contesters and asking them about their station design, training and tactics. You might think those are closely guarded secrets but not really. Most elite contesters are pretty friendly and happy to answer questions when they have the time.

Data gathering and analysis are at the heart of the scientific method. But you must analyze the right things to reach valid conclusions. That requires an understanding of the subject along with accurate data. Throwing a tool at the problem may or may not deliver results since without knowledge you might take the wrong lesson from the exercise. 

It also requires a healthy dose of humility. Both amateurs and overconfident experts are prone to measurement and analysis errors. Learning is a lifelong activity.