Phased KAZ Antennas

[WA1GFZ]

Playing with phased KAZ to find the best approach. Dallas Lankford flips the phase of the front antenna 180 degrees then adds a phase line to the rear loop so the null is improved. Anybody have any suggestions? A simple delay on the front antenna seemed to play ok. Simulation shows both work. My approach didn't require a different delay line length when you change bands.
I'm hoping to use the antenna 160 through 40

[WA1GFZ]

I studied the phase flip Dallas does and found it gurantees a cancel off the back of the array at all frequencies since the delay line and antenna spacing are equal.
My method guarantees phase add from the front side. Since we prefer cancel off the back I think the Dallas method is superior. Good thing I didn't switch back this morning before work.

[WA1GFZ]

More testing showed the D.L. phase flip configuration only improves S+N/N with two elements when the element spacing plus delay line of the rear antenna is in phase with the front element. My case only on 40 meters with quarter wave spacing and quarter wave delay line adds up to 180 degrees matching the front element phase flip. On the other bands the null was there but the signal didn't improve with the second element, it went down.
So back to the first configuration to see if a second element is worth doing and still cover 3 bands. My configuration will always be in phase from the front but will the null still work off the back?

[K1JJ]

Frank,

I modeled the two Kaz's, alone and then phased. Be sure to phase the second and fourth nulls at least 110-130 degrees relative to the first outside side peak. You don't want the main lobe to be skewed in the direction of the reflection - even if it is matched well.  The Kaz is OK as a single, but add the second and it may get ugly without ferrite or air core matching inside the near field, thus far field reduced anomalies and field coupling in the H plane.  Maybe a 75M beacon is the only way to align your common intended insulation aspect without generating harmonics, etc. Certainly the S/N will improve on 40M with the optimized but looser tolerance.

T

[Steve - WB3HUZ]

I agree with Tom, although the tolerances probably aren't that bad. It could be simpler to consider the extra nulls as overkill and work on only the first outside side peak. But as I said before, either way will work.

[WA1GFZ]

As a three band antenna I agree phased not worth it. I'm going to try my first method one more time. Degrading the signal to noise ratio for a better null seems to be a bad thing.
My bigger problem is there is a BB noise source that covers 40M. This is new and I have a new neighbor. Or the guy next door has a new toy. Sounds like a switcher. My noise floor on the HPSDR is up about 10 to 15 dB from 6.8 to about 7.5 MHz.

BTW. My matching at each feed point is a 1:4 BB transformer with a string of common mode beads on the 75 ohm side. Load on each loop is 1000 ohms
Loop spacing is 33 feet and delay line on the rear loop is 90 degrees on 40
About 20 feet of coax VF .66.
So when you look at say 75 meters the spacing and phase delay only adds up to 90 degrees. on the front side this attenuats the signal. On 160 only 22 degrees away from null so the signal is really attenuated.
My method always feeds the front signals in phase since spacing delay matches the coax phase delay. (adjusted for angle of radiation)

[Steve - WB3HUZ]

Try flipping the phase and adding the nulls. You automatically get alignment and much simpler switching and feed arrangements. Pattern modeling shows the two are equivalent but one is much easier to implement in the real world.

[K1JJ]

Right.  You may have a good idea there. What if Frank were to use a pair of unun-isolated RF steering diodes?  Modeling does show some interaction because of ground effects, but since he wants a simple feed arrangement with minimum stray inductance, the nulls will be more difficult to balance out with the standard reversed arrangement. Good thing the f-b is not affected when the s/n is optimized on 40M, at least in this case.   Good thinking, OM!

T

[WA1GFZ]

Well, I got a break in the rain and went outside to put both loops in phase. Now when I disconnect one loop on any frequency the signal drops about 4 dB so this verifies in phase always adds. Not sure if it is worth having two loops since the signal off the back also comes up. Configuration both loops in phase spaced about 33 feet with a 20 foot coax (I suspect VF .66 RG59 solid dielectric)  delay on the front loop.
Maybe I'll flip one around and see how they play as a switchable array.
I flipped between a dipole and the KAZ listening to WBZ to east and WTIC to west. Looks like about 15 to 20 dB FB on broadcast band. 
40 meters is unusable for weak signals with the new RFI, need to DF over vacation.
So it looks like the D.L. configuration is only usable on one band.

[Steve - WB3HUZ]

So Frank, why didn't you flip the phase and add the nulls?

You automatically get alignment and much simpler switching and feed arrangements? I suspect you'll get a 6dB improvement too and less noise on the input and output. 40 Meters will come alive!

[WA1GFZ]

Steve, I tried it both ways. When the phase between loops is 180 degrees the null is added on all bands but from the front a second loop degrades the signal on all bands except for 40.
In phase 0 degrees between loops the signals always add when a second loop is added.
I'm thinking of putting up a third loop facing West as a test antenna.
Also I'm going to build a pair of magic T combiners A 0 degree and a 180 degree so I can avoid going outside every time I want to make a change.
I was doing some simulation last night and thought I should reconsider out of phase.
Need to find the source of the 40 meter crap because it is trashing the noise floor of the whole band.

[K1JJ]

Frank,

The Magic T combiner may be the key, but I would use FOUR KAZ's instead - in p-p parallel. Don't laugh, the Kaz, being an anomaly antenna similar to the W8JK in mystery probably will come alive with with a sharper beamwidth, plus the outside side nulls will be greatly attenuated as Steve suggested using 180 out.

Actually, a four trace scope would make easy work of the null alignment procedure. Can you lay your hands on TWO dual trace scopes with bug ability?  Use the bugs with the nulls and you'll find that 40M trash as fast as Irb found that jammer...

T

[Steve - WB3HUZ]

I'd recommend using a hybrid combiner instead of a Magic-T. The input and output noise will add with a Magic-T but will not with a hybrid. The hybrid also stabilizes the phase to a very small error leaving both the null and the forward gain intact, but with loose tolerance. Considering that you're using two different sources and that the input and outputs are different, you would be well advised to do this.

And you get the added benefit of only needing a dual trace scope. The two hybrid ports need to be monitored to do the alignment, even if you are using more than two inputs. You'll probably want to run the scope off of a battery supply to avoid ground loops and common-mode noise corruption in your measurements.

[K1JJ]

Why would the hybrid be better than the Magic T ? The in/out noise should cancel if aligned using Taylor hybrid theory anyway, right? The difference using a balun device would void the common mode current anyway since we need a high impedance for RF on the external side of coaxial cable to establish a baseline for the outside side nulls. So having high impedance casually for some length of coax cable (odd multiple of 1/4 lambda) thus using hybrid center. The simpler balun device is dual trace scope ready, so to speak. Some might say length-ready too if placed just below the Kaz(s) feed points.  

This shunt inductance makes the hybrid work pivotally to the magic T impedance (we mean the external side of cable shield) since it's higher so that the RF current is lower and will find a high resistance and its low value will be very low (that will not disturb Kaz1 or Kaz2 or Kaz3 or Kaz4) which flows inside the coaxial cable at a high value.      

Wouldn't the main lobe's highest sensitive selectivity reduce the lowest need for additional test measurement common mode corruption anyway?

Watsa?

[WA1GFZ]

Interesting thought on the scope. I have a couple scopes that will do 4 channels. That would be a great way to dial in the phase.
Well, I flipped the loops out of phase again and did some more testing. I used the 80 meter dipole as the reference and compared the KAZ performance with WTIC and WBZ. Solid 15 dB FB to the East.
On 75 and 160  I may need to use a preamp but 40 the loss is a lot less. I was listening in the 80 meter window and A UK station was a lot quieter on the KAZ.
The new neighbors across the street just came home and the guy next door works third shift so his place is dark also. It will be interesting to see where this RFI is coming from. Noise on 40 is cranking.
I'm going to build another PVC matching module and face another KAZ West and see what happens.
I'm just using a minicircuits splitter/combiner I'll check out the hybrid although the signals should be about the same level except for the delay line loss. Dallas tried to equalize and posted it wasn't worth the trouble in the "Dallas Files".
HPSDR has a nice accurate S meter

[Steve - WB3HUZ]

That's cool Frank. You could add a line stretcher and vary the phase to meet the prevailing conditions. On 160 about 62.5 degrees would be a good starting point. Then you could vary to raise the signals to what you need. Then flip the phase to reduce the noise and get two for the price of one.

I still think the hybrid is better Tom. The Taylor approach going to make it hard to account for input noise, unless you have a vector voltmeter to trim the tolerances. Your balun idea would most certainly help since Frank is already seeing 15 dB on WTIC. With a preamp, the reverse isolation would also add in to subtract out some of the common-mode corruption.

All of this might be a moot point of Frank's neighbor is in the near field.

[WA1GFZ]

I have common mode chokes at the antennas so maybe I should add some in the shack. I just did some more simulations and stored the plots so I could compare them. 2 elements have a lower angle of radiation and better FB at higher angles. I'm thinking both antennas should have the same signal level but the phase will be way off. So after thinking about it maybe a 3 resistor splitter y 3- 25 ohms. Maybe a summing OP amp. Steve are you talking about an active or passive combiner? I can see how a magic t might be diverting power to the to the resistor due to phase offset. 
Dallas said he just copied the minicircuits design for his combiners

[Steve - WB3HUZ]

I'm talking about a phase-matched hybrid with two-port control and balancing compensation. Over coupling will occur with a Magic-T and the first inside outside nulls will be greatly reduced.

Your high-angle F/B is probably coming from the phase differences at the feedpoints and not at your combiner. Sensitivity analyis will tell you this for sure.

[WA1GFZ]

What a bone head I am. Sure hybrid is the way to go. I have a nice homebrew I use for dynamic range testing. It has about 24 db of port isolation and I have designs for better ones if that isn't enough.
I'll try that next,
TNX, GN

[K1JJ]

"You could add a line stretcher and vary the phase to meet the prevailing conditions."


I see you've been reading up on Taylor hybrid theory. Yep, he also said that the two cannot exist or be used at the same time, thus a line shrinker is out. Too bad cuz if Frank plans to use a 3 resistor splitter, it would be perfect. But a summing op amp places it in a different category, likely much more moderate to hybrids. Taylor didn't say much about them, probably not having op amps in his day.  

BTW, the Magic T may divert some power but using a capacitive compensator during phase offset might have the desired effect.  Or do you think the stretcher is a better idea in lieu of Frank's Op amp, Steve?

T

 

[WA1GFZ]

Hybrib combiner is my next test. You want isolation between sources due to the phase shift. A magic t will divert out of phase energy to the load resistor.
You want to sum them at the output. Last night I had the delay on the wrong coax when I flipped the loops out of phase. I actually flipped the pattern but didn't measure the loss. I'm thinking a 2 pole multi position switch could be used to select different delay lines. An LC delay needs many sections to be broadband.
Tom, ever work on LC delay lines for RADAR systems at LaPointe?
I will also try the resistor combiner which is very broadbanded but has more loss. 

[WA1GFZ]

Dallas has a new article worth reading. He got a close spaced loops system working. He is at 29 feet and I am at 33 feet. He is using magic T combiners??

[Steve - WB3HUZ]

Or do you think the stretcher is a better idea in lieu of Frank's Op amp, Steve?

The op-amp will induce it's own phase-shift. So matched and/or compensated units are required, otherwise the line-stretcher is out. Bi-phasing could be used instead, or Frank is back to a four-channel scope and adjusting everything manually at the feedpoint(s) and again in the shack.

Taylor was op-amp agnostic. His main concern was increasing the SNR with the optimized but looser tolerances.

[WA1GFZ]

Steve what are you using for your phased beverages? I wonder if you tried different delay lines on different bands. I don't want to reinvent the wheel here.

[K1JJ]

HuzMan said:

"Taylor was op-amp agnostic. His main concern was increasing the SNR with the optimized but looser tolerances."


Good point about the optimized but looser tolerances. Only Taylor theory could pull that off with a hybrid.

I'm still wondering if the line stretcher and summing Opamp could be used. How about using the Taft-Starsky approach where the input(s) are run at unity and the output goes into a splitter for processing? Then the Op amp summing would be controlled with GUI circuitry. Though I don't think Frank wants to get that complex with the design. Still it sure wud be nice to see a 50db f-b with DSP-type performance.  Remote steering can do wonderful when hands-off tailoring is electronically non-corrupted and accepted, huh?

T