The 5/8 and .64 debate!

[wavrider]

Hi Marconi good to see you on the boards

 

[Marconi]

Hi Marconi good to see you on the boards

Thanks wavrider.

 

[Marconi]

I used standard L networks tuned to optimal for these antennas.

DB, I modeled NB'rs ideas at 72' and 67.5' feet to the radials. I didn't expect as much of a gain difference for my models.

I used Eznec's ground description over Average Earth, but I don't think that explains the big factor noted by the gain differences.

My models show an SWR in the range of 18.0 - 23.0:1 SWR. Does your use of the matching device make a SWR difference for your model results? If so, do you think this might account for the gain differences we get?

I also have a wire attached from the base of the different antenna models running to ground. The lower antenna heights showed slightly more currents on this wire than the higher antenna heights, but none of the models here had significant currents on this wire. The radial system seems to be doing the job of isolating currents admirably.

The currents on the mast for my model at 67.5' feet is much lower in magnitude than my model at 72' feet, the opposite of your results. I have no idea why we get this difference.

I also note that my .625 & .64 patterns at 72' feet do not resemble your patterns, they are not even close. My models at 67.5' feet are very similar to your models.

I can post the models if you think it will help explain these issues for NB'r.

 

[The DB]

Hey NB, sorry on the delay on the time I thought I would have my version of the updated models out, what can I say, stuff came up...

Eddie, good to see you back...

DB, I modeled NB'rs ideas at 72' and 67.5' feet to the radials. I didn't expect as much of a gain difference for my models.

I used Eznec's ground description over Average Earth, but I don't think that explains the big factor noted by the gain differences.

My models show an SWR in the range of 18.0 - 23.0:1 SWR. Does your use of the matching device make a SWR difference for your model results? If so, do you think this might account for the gain differences we get?

The matching network does reduce gain some, and I've noticed that correcting for the R component in complex impedance generally generates more loss than the X component.
Because of the matching networks inserted, SWR is very close to 1:1 in the models I posted above. Just a note, the matching network does not change the pattern the antenna generates, it just attenuates it some...

Another thing that would have a potential effect here is the fact that I used T6 aluminum elements instead of the theoretical perfect conductor that NEC likes to use. I've noticed that aluminum will have a similar effect on overall gain as using the default theoretical perfect conductors.

The currents on the mast for my model at 67.5' feet is much lower in magnitude than my model at 72' feet, the opposite of your results. I have no idea why we get this difference.

I also note that my .625 & .64 patterns at 72' feet do not resemble your patterns, they are not even close. My models at 67.5' feet are very similar to your models.

I can post the models if you think it will help explain these issues for NB'r.

The wire running to ground in my models was made out of copper, which will have a different electrical length than the theoretical perfect conductor NEC likes to use. A change of velocity factor in run of near two electrical wavelengths (varying by model height of course) may be enough to explain this variation between our models. I'm sure we also used different thicknesses of wire which will also have an effect.

I am currently thinking that just running a wire to ground may be unreliable when determining the maximum potential strength of common mode currents on an antenna system, which was the point of me running said wire to begin with. The reason I think this is the common mode currents present will change based on this length, and the fact that a straight run to ground isn't always, and is in fact rarely the case in the "real world". I am thinking a single wire of 1/4 electrical length would be better suited for such determinations as in most cases this length seems to like to attract RF currents. I will likely toy with other lengths as well just to make sure though. For me the intention of running such a wire is to ensure that the antenna design is sufficient to control said currents, not to determine the pattern with said currents. Such a pattern based on such a model would be unreliable at best as most people don't have a wire running straight to ground, instead their coax line runs in various directions, and around unpredictable curves, and near metal and other such materials that will have other effects on it. So when designing for a real world simulation, the maximum amount of current on this line is of more concern to me than effects they generate which might not even be accurate for a given installation.


The DB

 

[Marconi]

The matching network does reduce gain some, and I've noticed that correcting for the R component in complex impedance generally generates more loss than the X component.

Because of the matching networks inserted, SWR is very close to 1:1 in the models I posted above. Just a note, the matching network does not change the pattern the antenna generates, it just attenuates it some...

I agree, and this is generally what matching devices are said to do...add losses. In modeling however, I suspect that within reason the higher SWR I see, without using the matching feature, tends to show some increase in gain for the model. This is what I figure accounts for our differences in gain. I'm still curious why our patterns at 72' feet are so different.

Another thing that would have a potential effect here is the fact that I used T6 aluminum elements instead of the theoretical perfect conductor that NEC likes to use. I've noticed that aluminum will have a similar effect on overall gain as using the default theoretical perfect conductors.

I used T6 aluminum too, and I don't see much of a difference either.

The wire running to ground in my models was made out of copper, which will have a different electrical length than the theoretical perfect conductor NEC likes to use. A change of velocity factor in run of near two electrical wavelengths (varying by model height of course) may be enough to explain this variation between our models. I'm sure we also used different thicknesses of wire which will also have an effect.

I will check this out, but I suspect the wire thicknesses will tend to show more difference than varying the wire material will.

Here are my patterns at 72' feet and 67.5' feet.

 

[viking63]

Have at it, document your results ,come back and post your findings

i needn't do the experimentation. as so many antenna companies have made, them tested them. they have everything it takes to do what your asking of me. hollow words. as i 5/8th wave needs added inductance to equalized the extra capacitance that a 5/8ths wave is inherent with to make a 50 ohm load. and as we all know the longer the radiator the stronger the received signal. and if this is not true, the why do long wires work at all?

 

[The DB]

Seems to be an error on my side, I made both of the vertical elements for my 72 feet radial height models a little to short. I corrected in the plot above.


The DB

NOTE, it won't let me edit that post anymore to note the update...

 

[bob85]

viking63,
The reason a 5/8wave usually outperforms a 1/2wave when mounted on the same mast is twofold,

Its easier to make a low loss matching transformer for a 5/8wave than a 1/2wave ( nobody makes a low loss 1/2wave )

a 5/8waves current maxima is higher above ground than current maxima of a 1/2wave when mounted on the same mast,

Build a low loss 1/2wave and mount it at the same tip height as a 5/8wave and it has the edge over the 5/8wave,

The extra portion of radiator the 5/8wave has is out of phase and not adding to performance other than raising the tip higher,

Extending the radiator length to more than 5/8wave causes worse performance at low angle were we need it..

 

[Marconi]

DB, since our 72' foot models were not similar, I went ahead and made my model at 58.5' feet like NB requested. This pattern too does not resemble your model at 58.5' feet.

The currents on this model below at 58.5' feet also shows very low common mode currents flowing on the mast. My mast is set at 1" diameter and the material used is set for T6 aluminum as well.

As you noted earlier, this idea about currents is not to suggest these particular antenna heights are anything special for real world installations. I also think these ideas may be difficult to explain or consider regarding modeling, and I look forward to your ideas in further consideration for what it may be telling us.

 

[Marconi]

viking63,
The reason a 5/8wave usually outperforms a 1/2wave when mounted on the same mast is twofold,

Its easier to make a low loss matching transformer for a 5/8wave than a 1/2wave ( nobody makes a low loss 1/2wave )

a 5/8waves current maxima is higher above ground than current maxima of a 1/2wave when mounted on the same mast,

Build a low loss 1/2wave and mount it at the same tip height as a 5/8wave and it has the edge over the 5/8wave,

The extra portion of radiator the 5/8wave has is out of phase and not adding to performance other than raising the tip higher,

Extending the radiator length to more than 5/8wave causes worse performance at low angle were we need it..

Bob, I hope I find you doing well.

I tend to agree with most of your comments here. For reasons I can't easily explain, I tend to favor my real 1/2 wave antennas over my 5/8 waves. Right off it might just be the size of these big 5/8 waves in my case, plus if I can get the 1/2 wave tip up to the tip of the 5/8 wave I don't see much of a difference that really matters.

Thanks to DB he has confirmed for me that when he uses the matching feature correctly in modeling...it can and will correct the matching results for the model, and in a guess I would expect no less. However, I was puzzled to question this issue...mostly because very few modeler's and the guru's seldom wrote, talked, or used the matching feature in their articles.

I have not been able to understand the application for this feature in Eznec, and thus I could not use it to prove the matter one way or another. Again, it made sense to me that matching matters, but I could not work out the details, so I was stuck.

I have suspected however, that the higher SWR results noted for my EFHW models may be falsely elevating the gain results. It seemed to me that over time as I perfected the dimensions of some models I often saw the gain drop as a result. At some point I began to consider this to be what should likely be happening with modeling. There is a distinction to be made however, this is not to say that correcting a gross error made in a model would or could not show improvements in gain.

It would be interesting to see if DB has tested an EFHW with and without matching and accounted for the difference in gain he noted.

 

[bob85]

Its good to see you back Eddie, im well, just busy with other things and not much time for radio,

I don't understand the effects of the matching feature either, its not something i see guys that model use, Maybe you and DB can figure it out, IM watching with interest.

 

[The DB]

When it comes to the matching networks, I create a plot for the unmatched antenna, then open the matching screen, which looks like:

When choosing for comparison purposes I tend to stick with L networks, and choose whichever one for a given antenna produces the least amount of loss. If I used a Pi or a T network I could artificially manipulate the Q reading and to an extent manipulate the output, which is not what I want to happen. I could also use the stub match, although I generally don't tend to.

Comparing plots for matched and unmatched antennas, I'll start with an end fed half wavelength antenna.

As you can see, the plots are the same, except one is attenuated slightly. Here is the data for the plots shown above.

With these screens you can see the effects of the matching network compared to no matching network. The impedance is much lower, as is the SWR reading. Other changes, such as the efficiency readings are adjusted as well.

Comparing various antennas in a given environment, some antennas will be affected more by matching than others, for example, an end fed half wavelength antenna vs a center fed half wavelength antenna. The end fed half wavelength antenna will have far more matching losses than the center fed half wavelength antenna, which likely has none as matching should not be needed. In the end it is just one additional factor to consider to get more accurate results...

A note on using this matching feature, it assumes a low loss L network. This can be a problem as many antennas (the a99 for example) don't have a matching network that even resembles such a low loss matching system, so matching for said antennas will always be a bit optimistic. I have yet to find a sure way around this...


The DB

EDIT: Odd, in one of my browsers the pictures don't show but can be right clicked and downloaded, and in another browser they show fine...

 

[Marconi]

DB, thanks for your exhibit of 4Nec2 matching feature. I still don't have the necessary understanding to apply the matching feature using Eznec, but this may help me get started into the issue of antenna matching in modeling.

However, like I have claimed many times in the past...in the minds and ideas of many CB'rs all around the world, you might agree that your results are like most of the works of radio guru concerns, reported effects, and other scientific issues published for the radio hobby...the differences are generally minor and manufactures generally blow the differences all out of proportion in their Puffing in advertising. A reason they should not be believed.

There are truly differences to be noted in your example application above...using an antenna with probably the worst matching issues a CB operator will ever experience...unless he just forgets to connect a suitable coax to his favorite antenna and his transmitter...THESE DIFFERENCES ARE MEAGER AT BEST.

In the past, I've made frustrating type claims for this hobby group about the very small differences proclaimed as important by the real science of radios and antennas...I've noted many times that most CB'rs probably get what is said in science all wrong, and/or just misunderstand, as they consider all this business as very important to their success in 2-way radio or the lack thereof.

With the results difference you note here for a simple antenna we all have likely dealt with...don't you agree with me...WHAT DIFFERENCE DOES ADDING A MATCHING DEVICE TO OUR ANTENNA MODELS REALLY MATTER...IN THE BIG SCHEME OF THINGS?

Sorry, but no one I know could ever realize this very small .06% difference in performance gain just using their radio. This also verifies another consideration that I have made over time...that in modeling I tend to see improvements in my personal models manifest as small decreases in performance difference rather than improvements in gain for example.

As you can see, the plots are the same, except one is attenuated slightly.

This consideration does not include gross errors however.

With your help here, you have helped me understand better why we probably see very little guru commentary about antenna matching in modeling.

Thanks,

 

[Marconi]

DB, I hope you don't take my comments above in the wrong way. I appreciate the good work you did for us here on matching.

You told us your post edit feature timed-out on you, will you be updating your models.

I have added below an overlay of the models that I made for NB's ideas so we can more easily compare the antennas in groups at the 3 different heights noted.

I also hope NB returns with his report soon.

 

[The DB]

Eddie, you are completely missing the point of why I add matching networks to my models. You see it as an ends, and see a minor change in my demo models and quickly jump to calling it irrelevant, or nearly so. I, on the other hand, see it as a means to an end that I am working towards. I am trying to use it as a stepping stone to move beyond something I see as one of the many limitations of NEC based modeling, and make my future models even more accurate simulations of real world antennas. To me it is a step in the process of further improving my modeling technique.

We are obviously at a point of disagreement. Due to our past on such discussions, I have no interest in discussing this with you further.


The DB