Question on antenna power dividers

[K1JJ]

Hola -

Let's say you have three vertically stacked 50 ohm horizontal Yagis.   The modeling antenna pattern looks best when you divide the power between them as:  8 watts, 10 watts, 8 watts.  (10 watts to the middle antenna)

Using coax transformers/ dividers, etc, how would you split the quantity of power between them like this?    It wud be easy to make a divider to feed equal power, but how about an 8-10-8 power split while still maintaining 50 ohms each?


I read that a binomial divider gives 1-2-1  power division, so would not be suitable.


Or would a toroidal transformer with three taps do it?  I don't know how it wud handle the 50 ohms each Yagi requires while feeding less power to the outside antennas..

Any ideas?

T

[Ralph W3GL]

Tom,

You might want to direct a PM to Walt, W2DU on this subject...

If anyone can come up with  a good solution, it would be him...

[WA1GFZ]

sounds like less than 1 dB difference What is the point of that.

[WD5JKO]

> sounds like less than 1 dB difference What is the point of that.

Yep, if it really matters, make a 1 db 50 ohm attenuator for the upper and bottom antennas, and feed 10-10-10 watts from your splitter. Is this a case where we calculate to 8 digits to the right of the decimal, and then cut with an ax? 

Jim
WD5JKO

[WA1GFZ]

I would think you would play with the phase to shape the charge

[K1JJ]

sounds like less than 1 dB difference What is the point of that.

Yep, it doesn't sound like much, but it causes the pattern to become sharper and cleaner by tucking in the rear lobe and reducing some minor high angle lobes.  The result is an almost perfect pattern. 

Though, in the real world, I doubt the pattern would be that clean or I would even have use of the 8-10-8 splitter. But ya never know when it comes time for testing and I will usually try things like that in fine tuning mode.

Changing the phase has an effect on the vertical lobes, but not the effect I'm after. The power distribution quantity is the needed solution.


Yes, an attenuator would work for sure, but I have an aversion to burning up antenna power IF there is a possibility of another solution. (A beverage termination is an exception... )


BTW, this project is the newest insanity...   it's a  2 over 2 over 2 for 40M.   Three full-size 40M 2 el ROTARY Yagis stacked at 190', 115' and 65', homebrew of course.   The 7el works good, but I'm looking to bring the takeoff angle down from about 22 degrees to 13 degrees.


T

[w3jn]

If this brings the end of the world by setting fire to the ionosphere we'll know who the guilty party is 

[K1JJ]

If this brings the end of the world by setting fire to the ionosphere we'll know who the guilty party is 

    Hey, don't laugh - if the timing is right and you guys come over here for Hoss Traders, YOU may be doing involuntary servitude and later be held by the authorities for knowledge of an insane, unnatural and perverse act on mankind.... 

T

[w3jn]

Like you did to us the last time we were there?  It was OK except for that being kept prisoner in your basement part.

[WA1GFZ]

Does this mean Yaz gets to share the electric collar......

[WA1GFZ]

Tom,
How about 3 linear amplifiers, one on each ant.

[WD8BIL]

Tom, if you direct feed at the center antenna and TEE off there to the upper and lower ants, you may inherently get the power division your looking for if you work the phasing lines correctly.

[K1JJ]

Tom, if you direct feed at the center antenna and TEE off there to the upper and lower ants, you may inherently get the power division your looking for if you work the phasing lines correctly.

Hi Budley,

Let's see...  we have three matched 50 ohms Yagis.   I want them all fed at zero phase.   By changing the feedline length, all we get is a phase angle change (not desireable) and not a power division change.  Unless you mean something else?

If there was a way to maintain 0 degrees phase and vary the power division to a 8-10-8 split, that's what I need.

The only thing I have so far (using coax) is to feed the middle Yagi directly and run a full-wave coax to the top and bottom Yagis. This will produce 0 degrees for them all PLUS give a small loss in the 90' of coaxes going to the top and bottom.  Whatever the loss is for RG-213 on 7 mhz when matched.... maybe 0.5 db?  I could use 2 full wavelengths, but then we are burning up power like an attenuator idea - something I don't want to do if I can help it.

T

[WA1GFZ]

Tom,
Do a search on directional couplers. Two cores are used  and there are a number of different power ratios possible.
 Another easy way would be a wilkenson splitter (a "Y") with two quarter wave lengths (electrical) of RG11 driving each load. Then build another and hang it off one of the first splitter outputs. This will give you 3 ports with two of them about 3 db below the first. So one output of the first splitter will go to the center beam and the second the input of the second splitter. The two outputs of the second splitter will go to the top and bottom beam. Everything will be at 50 ohms. Note the top and bottom beams will also be delayed 90 degrees but you can add 50 ohm coax to the center beam run to get the three phases equal.

[K1JJ]

Tom,
Do a search on directional couplers. Two cores are used  and there are a number of different power ratios possible.
 Another easy way would be a wilkenson splitter (a "Y") with two quarter wave lengths (electrical) of RG11 driving each load. Then build another and hang it off one of the first splitter outputs. This will give you 3 ports with two of them about 3 db below the first. So one output of the first splitter will go to the center beam and the second the input of the second splitter. The two outputs of the second splitter will go to the top and bottom beam. Everything will be at 50 ohms. Note the top and bottom beams will also be delayed 90 degrees but you can add 50 ohm coax to the center beam run to get the three phases equal.

Thanks for the info, Frank.

Using your splitter coax idea, what would 3db down be in the equivalent of the "8-10-8"  division on MiniNec?  I think that is voltage MiniNec is using, so not sure how 3 db relates.

T

[WD8BIL]

It wud end up 5-10-5, Vu.

[K1JJ]

It wud end up 5-10-5, Vu.

That what I thought, but maybe that's voltage -  is it a 20Log relationship  to muddy up the db calculations?   Or is the MiniNec input power, and if so, it IS 3db.    I suppose I shud read the MiniNec Text file...

T

[W1RKW]

You're talking watts so 3dB down would halve the power making 4-10-4 from 8-10-8 yes/no?

[KA2QFX]

Tom,
Interesting problem.  Here's some ideas...
In the style of W2DU I was thinking how you could do this using tranmission line tranformers (TLT). I like the low loss, broadband effect they provide. However, now matter how I thunk it you keep needing odd fractional transformations which don't lend themselves well to nicely balanced TLTs.
Then it occurred to me that you could easily achieve the power division by shifting the impedance of the center antenna to 62.5 ohms.  This should be achievable by play with the match and element lengths. 
Then, feeding all the antennas from a common (voltage feed) point would yield a 17.8 ohm load to be transformed back to 50.
Anyway, changing the load rather than changing the driving voltage/current seemed the easier way to achieve the small difference you want (-.96dB).   A way to provide a common voltage feed to the combined antennas still seems elusive without the fractional TLTs. Which can be done but it's not "elegant". Shifting the antenna Z down and doing a series current feed seems less likely.  I'll continue pondering the matter. Just sharing brain waves here.

73,
Mark

[Steve - WB3HUZ]

-1.9 dB

[KA2QFX]

Watts

dB = 10 x log P1/P2

Volts or Amps
dB = 20 x log V1/V2 or I1/I2

10 x log 8/10 = -.96dB

[Steve - WB3HUZ]

I assumed we're talking voltage since Tom referred to source info from Minnenec.

Which is it Tom?

[WA1GFZ]

3dB power you multiply voltage by about 1.41 and getting back to zero phase by adding another 90 degrees of 50 ohm cable will end up slightly lower.

Here is another option. Build a lumped 1:3 splitter like you just bought for EB104 but with 3 ports. On each output do a broadband transformer with different turns ratios to get the power split.
check out Mot AN749 I can send you one if you don't have it. CCI might have it also.
A 1:3 splitter each output will be about 150 ohms so you will need a step down BB transformer to drive each feed line. You might be able to play with the turns ratios to get the power split you want and still be close to 50 ohms on the input. For QRO the class e beads maybe 4 or so for each of the 3 sleeves. Then a couple stacked 2 or 3 inch type 61 or 64 cores for each of the three transformers. 150 ohms down to 50 ohms.

[K1JJ]

I assumed we're talking voltage since Tom referred to source info from Minnenec.

Which is it Tom?

Yes, it's MiniNec and voltage.

OK on the suggestions.  Sounds more complicated than I first hoped.   I'll do some more reading and thinking and see what jumps out.

After modeling the 2X2X2  40M stack further, that little power division of 1 db improves the f-b by about 10db and the higher angle lobe by about 5db.  These lobes are already tucked in to about -20db, but this makes it even better.

I've been finding that tuning 40M Yagis in the real world for f-b's of 40 db+ is possible. (at exactly 1/2 wavelength above ground)   I just put up a revision of my 3el 40 M wire Yagi to the west. It was previously measured at about 35-40db f-b. Right now I'm seeing at least -45db on certain stations around 7155. The sweet spot window is only about 50kc wide for this extreme performance.   When listening to a strong station in Europe, when switching west, I can't hear him anymore.  So I'm hoping for some extreme performance with the new 2X2X2 stack. Since it's using 2el beams, the f-b is usually not better than -18db for ONE beam. When using a stack, the f-b improves as good as a 3el beam becuz of the "curtain" effect.  This power division technique appears to be just one more trick to a higher f-b.

T

[WA1GFZ]

So the double wilkenson.
2 volts into the first divider the output of the first divider would be 1.41v and the second divider outputs would be 1 volt each.
So you would get 8V, 11.3V, 8V.
Then you need to add a quarter wave section to the 11.3 to line up the phase so the voltage will drop a tad. Is that close enough?