Towertalk
[Top] [All Lists]

Re: [TowerTalk] Modeling vs Experimenting Crowds

To: towertalk@contesting.com
Subject: Re: [TowerTalk] Modeling vs Experimenting Crowds
From: David Gilbert <xdavid@cis-broadband.com>
Date: Wed, 11 Sep 2019 21:45:39 -0700
List-post: <mailto:towertalk@contesting.com>

I think most of that is fairly misguided advice when it comes to antennas.

Here's what I don't understand.  Of all the things we as hams have under our control to improve our ability to communicate, antennas are at or near the top of the list.  Up to a certain point (as was discussed here recently), the effectiveness of an antenna has more bang for the buck than pretty much anything else we have a handle on.  So why on earth do some hams relegate it to "just put something up and see if it works"?  That can certainly be fun, and I have personally built multiple dozens of different antennas just for grins, especially over the several decades I've operated Field Day or gone camping.  But if you aren't trying to optimize your results just admit that you don't really care.  Which is fine, of course. There are lots of aspects of ham radio I have zero interest in, but antennas isn't one of them and I would have thought that anyone subscribed to this list was here to learn how to make better ones.

If you are indeed trying to optimize your antenna(s) , trial and error is a terrible way to learn how to do it because you can't control enough of the variables and some variables are even pretty hard to measure.  Multiple iterations are slow, as well as being imprecise (because of the variables) if you're trying to improve things like gain or pattern or the ability to be heard in general.

It used to be that we were either stuck experimenting with antennas or doing laborious mathematical calculations by hand, but even basic modeling programs have turned that completely around.  It isn't that the models are infallible (they are not) .... it's our ability to quickly and easily learn from them that is important.  As I said above, I have built many, many antennas over the years and I played around with most of them to see what would happen, but I've learned far more from looking at the current distributions along the elements in EZNEC and observing the effects on gain bandwidth, pattern, and SWR than I ever did with various experiments.  The rest of what I've learned came from smarter folks than I here on TowerTalk who either understand antenna theory better than I do or were better at modeling that I am.  I can honestly say I haven't learned much of anything worthwhile from somebody who put up something and claimed "it worked."

You say no one doubts the results of the experimenter.  That maybe true of the chemist or circuit designer who can control his environment, but I ALWAYS doubt the results of the antenna experimenter who can't explain his results with either theory or a suitable model, or both, because I know in most cases he wasn't able to control or allow for important variables (propagation, proximity effects, noise variability, etc) and probably didn't even understand enough to do so if he relied only on his experimentation to teach himself about antennas.  I have a few of my own examples of this, the most notable being an antenna I tried for Field Day one year.  I wasn't able to do my normal preparation so I modeled a flat elongated vertical wire rectangular loop fed in the middle of one of the short vertical segments.  The model said it would perform marginally at best, but it was easy to put up and could be fed on both 20m and 40m, so I went with it anyway.  The night before Field Day I gave it a try and worked an FR5 (FR5DN, I think) first try using 5 watts from here in Arizona with Q5 CW signals both ways (somewhere I have the QSL card to prove it).  I thought hey, maybe this thing is better than I thought. It wasn't.  I didn't work a single other DX station that night and the antenna turned out to be one of the worst I ever built for Field Day as well.

The bottom line is NOT that antenna modeling will always give you the right answer, but it will make you a heck of a lot smarter more quickly than simple experimentation will, and in most cases you will at least know in which direction to make changes if you're trying to make improvements.

One last example.  I recently built a 5 element 6m yagi using dimensions posted by DK7ZB on his website, and he came up with those dimensions using modeling software.  Most 5 element yagis have a feedpoint impedance down between 12 and 15 ohms, but his design (others have done the same) natively gives a 50 ohm feedpoint at a very small sacrifice of forward gain.  It does so by making the first director actually longer than the driven element.  How much time and how many iterations do you think it would take an experimenter to stumble across that?

By the way, modeling has given me FAR more "eureka" moments than any antenna I ever built.

73,
Dave   AB7E




On 9/11/2019 6:45 PM, Don wrote:
Well put by Mr Fox, Shawn. Reminds me of a little picture frame with the following which hung on the wall of one of our top engineers in the test and instruments company I worked for. Smart, learned, well educated and a dedicated experimenter, determined to defy the 'it has to be so' crowd. Quite successful.

"No one believes the results of the computational modeler except the modeler, for only he understands the premises. No one doubts the experimenter's results except the experimenter, for only he knows his mistakes'. Beneath was two handwrittenlines on a strip of paper. "Modeling is not as exciting as experimenting where the outcome can be an Eureka moment".

I had copied that and tucked it away in my old company history files I left with.

Don T W7WLL

On 9/11/2019 2:08 PM, Shawn Donley wrote:
I occasionally teach a class on modeling certain mechanical systems using Simulink.  The second slide is a quote from the British mathematician George E.P Box.  I think it may apply to this discussion as well.


"All models are approximations. Essentially, all models are wrong, but some are useful."
_______________________________________________



_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk


_______________________________________________



_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk


_______________________________________________



_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
<Prev in Thread] Current Thread [Next in Thread>