SOME FIELD INFORMATION ON THE DOUBLE-DELTA (D-KAZ) ANTENNA

060714

About the Double-Delta/”D-Kaz” Antenna

Performance consistency: Real-world observations

Chasing the measurements for some meaning

Practical pattern-reversal

The “Broadside D-Kaz Antenna”

Roll your own D-Kaz: Some ideas about deployment

By Mark Durenberger

DX-ing near N 45 48 56 W 94 34 46

About a year ago I was introduced to the Double-Delta antenna, as modified by the brilliant

DX-er Neil Kazaross, to become the “D-Kaz” antenna. The Delta and its Double-Delta variant

aren’t new; the DHDL may be the grandfather of this approach. Kaz has suggested we can

expect as much as 40 db front-to-back nulling with the array. However, judging by DX-er

reports, the various methods used to deploy the Double-Delta array might sometimes result in

less-than ideal operation. So we wanted to spend time taking a look. We’re still climbing the

learning curve, so all ideas need to be considered. If you’re new to the Double-Delta read on,

about a simple, small, reversible wire antenna that can deliver this sort of pattern:

We carved out a few days in May and June 2014 to measure D-Kaz MW performance with

different methods of connectivity. We’re fortunate to have an ideal landing site for this

antenna, behind my lake home in a remote rural area, away from MW pests and in a region

where electrical power is underground. While our observations and conclusions are not

classified as “scientific,” enough data points were collected (and repeated) to make this

information reasonably satisfactory. On the other hand (here comes the disclaimer) YOUR

MILEAGE MAY VARY…these are simply the conclusions of one DX-er who measured these

antennas under specific conditions.

Here’s the basic delta configuration:

And this is the basic design of the D-Kaz antenna for MW (and well above). The “self-

impedance” of the D-Kaz is around 1000 ohms, as seen at corners “A” and “B:”

Each of the two D-Kaz wires composes two/thirds of a triangle and then continues as the base

of the other Delta. The crossfire phasing causes a 180- degree phase shift so the resultant

pattern is almost the same as two loops in series fed with something close to a 180-degree

phase difference.

Here’s an interesting way of contemplating this antenna…from the mind of Jim Snowbarger;

a DX-er blind from birth:

“If you lay out a rectangle of wire, short vertical sides, and long horizontal sides, and just pick up the right hand side of the rectangle and flip it over, trading top and bottom ends of the right hand vertical edge, the rectangle is now twisted, and the long horizontal members cross, you can bend the wires as needed, even giving up the 90 degree corners, to form it into this dual triangle shape, where the wire that forms the base of one triangle, extends to become the side of the adjacent triangle. . “It was that mental exercise that pointed out to me that these two loops are out of phase. And, they have a certain amount of physical distance. So, there is phase delay due to that (distance) as well. The pot just balances the effects on the receiver of the two loops, to get cancellation.”

The pickup pattern of the D-Kaz is similar to that of phased loop antennas; the phasing might be

thought of as eliminating the “rear” half of the “Figure 8” of a typical loop.

It’s the simplicity of the arrangement that’s appealing. Cancellation of signals to the “rear” is

accomplished by simply adjusting the null pot opposite the receiver. (By the way we’ve put

“rear” in quotes since the antenna can be reversed by swapping null pot and receiver. Thus,

“rear” is an abstraction. The receiver ((“front”)) side is the end facing the desired signals.)

Now: Astute viewers have noted the cancellation components are a resistor balanced against

what’s likely a complex impedance. That “almost-but-not-quite-a-match” is part of our story.

Experimenters have tried various Double-Delta antenna lengths. It was determined early on

that shorter versions caused some degradation in low-end MW performance (below maybe 700

kHz), while the longer versions didn’t produce good front-to-back nulling at the high end of the

MW band. A compromise length is 120 to 140 feet, and we worked with these two lengths.

Taken by itself as in the diagram above, it’s a “darn good antenna” as Nick Hall-Patch might say.

But if you’re not interested in standing sentinel outside right at the antenna, it’d be nice to

‘extend’ the ends to the shack, for comfort and convenience. It has been posited that both the

null-pot and the received signal from the other end can be extended for a distance from the

antenna without serious degradation. Here’s the elemental form of this extension:

An analysis of this arrangement might lead one to wonder if such extensions could upset the

electrical geometry of the antenna and thereby degrade the nulls. This is what we wanted to

determine. (Referral to this diagram will be useful throughout our test results and in a

following discussion of “reversibility.”)

A DETECTIVE STORY

DX-ers are having good luck with this antenna. Our work for the DX Audio Service found it to be

exemplary for specific purposes where a broad region is to be vacuumed for MW signals. But

we were bothered by some inconsistencies in operation and decided to see what we could

learn. We set up the antenna where we could leave it in place for a few days.

TEST LOCATION:

Summer cabin in north-central Minnesota. By the whim of geography (the existing road-bed),

our D-Kaz was oriented at 330/150 degrees. While we made measurements in both north-

facing and south-facing directions, the preponderance of reliable ‘reference’ stations that might

be used as null candidates was south of us. So most of our data was gathered from a north-

facing set-up.

TEST OBJECTIVES:

i) Determine which null-pot and receiver connectivity arrangement delivered the deepest and

most-consistent nulls on the “deaf” side of the array. Could we arbitrarily extend the “ends” of

the antenna into the shack? Or did the null-pot and radio have to be at the antenna?

ii) Determine whether the null-pot could be replaced by a fixed “average” resistor value. If we

could eliminate the need for an adjustable null, one could further simplify the set-up.

TEST CONDITIONS:

Weather was sunny and dry for all five test days (5/23, 5/24 and 5/29, 2014 and 6/4 and 6/6,

2014. All observations were recorded mid-day, from the signal-level display of our Perseus

SDR. QRN was absent until after dark and QRM was below the noise floor of the Perseus.

TEST METHODOLOGY:

We first plotted the performance of a 140-foot antenna under various connectivity alternatives.

We then shortened the antenna to 120 feet and repeated this work.

For our null measurement references we selected a number of stations to the south whose

ground-wave signals made for reliable nulling. These stations are identified below, with their

directions plotted against the EZNEC predicted pickup pattern for this antenna.

Measurements were made by: 1) Noting signal strength in dBm with the null end of the

antenna “open” (“MAX” below); then: 2) Connecting the null-pot and adjusting for best null

(“MIN”); and: 3) logging the resistive value of the null-pot setting (“R”) for that best null:

We conducted a significant number of measurements. For most tests, the 9:1 transformers on

both ends coupled to various lengths of Cat-5, extended respectively to the null-pot and

receiver. Null-pot and receiver were tried in diverse locations.

The 9:1 step-down at the ends of the antenna meant the native antenna impedance of 1000

ohms was taken to just over 100 ohms, and that was a nice match for Cat-5 (110 ohms). (Cat-5

cable and its successors is well-balanced twisted-pair data cable, and in some applications it

outperforms coax, so it’s a good choice.)

With this impedance step-down and transfer by a 110-ohm Cat-5, the null pot at the extended

end would now need to be by definition a 100-ohm device. Interestingly…it’s been shown

among DX-ers that a 250-ohm pot is needed to provide enough range. This is a significant clue

that maybe we’re not doing a good job at accurately stepping down the antenna impedance. In

fact, this was the first clue in our detective story.

Since the receiver was also being extended by Cat-5, a further step-down (2:1) of the extended

feed to the receiver matched the Cat-5’s 110-ohm impedance to the RF amp and SDR.

RF amplification was placed in front of the Perseus SDR (Kiwa amps). 10 dB of amplification

was suitable for the Perseus, with all candidate signals appearing at least 25 dB above the noise

floor of -114 dBm.

The initial data gathered was meant to determine the effect of various lengths of Cat-5

extensions of the ends of the antenna. After a few hook-up variations it was becoming evident

that there was no real predictability to the results! For example, Cat-5 extensions of equal

length would yield one set of readings…while simply reversing these extensions to opposite

sides of the antenna might yield a different set. (?)

Unequal lengths proved interesting. An extension of the receiver end transformer all the way

back to the null pot end (that’s 150 feet of receiver-extension Cat-5) and a short extension from

the null-pot transformer to the null-pot provided one set of readings…then…moving the

measurement point to, say, 15 feet farther from the null end (thus 15 feet closer to the receiver

end) provided different numbers. Null-pot resistances were also somewhat unpredictable.

We began wondering if there was a magical “sweet ratio” of Cat-5 extension lengths. That

thought about inconsistencies turned out to be our second clue.

Because we wanted to examine the feasibility of a “Broadside D-Kaz” (see below) we also

tested the array with extra-long (240-foot) Cat-5 extensions. Now we discovered that null

depths started to shallow out…AND we hit a mid-band spot, around which readings changed

dramatically. This was our third clue.

At this point we had more questions that answers. From one of Kaz’s suggestions we

considered whether the long Cat-5 extensions might be picking up extraneous signals and

impacting the results. So we moved the RF amplifier to the receiver end of the antenna so we

were pushing a hotter signal through the Cat-5. Nulls got slightly better and, interestingly, nulls

on the low end got a lot deeper! Though we didn’t know it yet, this was our fourth clue. (We

also contemplated the possibility of external-signal pick-up in the long null-pot extension.)

The other objective of our measurements had been to determine if a fixed resistor could

replace the variable null-pot. From our measurements to this point, this didn’t appear

practical, unless one were willing to accept a D-Kaz with an average front/back ratio of less than

15 db under various extension lengfths. So the next arrangement was removal of the 9:1

transformer and direct connection of the null-pot to the antenna (the 250-ohm pot now

became 2K). Immediately the nulls were dramatically deeper, averaging better than 30 db and

the resistance values varied less than 10% across the band.

Next we began varying the lengths of the receiver extension Cat-5…and found less variance.

That was the final clue. Attention turned to the 9:1 step-down transformer.

From extensive measurements with the null-pot right on the antenna we now knew we could

qualify the 140-foot d-Kaz to work with a fixed null resistor, delivering antenna performance

with average null values of 30 dB or so. Receiver-extension length was less relevant. This is

pretty much what Kaz predicted would happen

HOWEVER: That wouldn’t satisfy those who wanted to null specific frequencies from the shack,

reaching for far deeper nulls than the 30 db or so provided by an average resistor value.

But extending a null-pot for manual nulling in a shack would required a step-down transformer

that wouldn’t make things go hooey. What was going on to cause all the inconsistencies when

the transformer was inserted? Here’s my semi-informed theory:

When you consider the Double-Delta as a candidate for phasing, the nulling components on the

ends should approximate each other. In this case, at the receiver end of the antenna you find a

lumped impedance consisting of the 9:1 transformer, the Cat-5 extension, and another 2:1

transformer. On the null-pot end we’re ‘balancing’ this with a pure resistance. That pure

resistance is actually pretty good for fairly deep nulling…but it’s not perfect. And if we want

shack-based control we need to make that null-pot extension look more like the other end.

In our early observations we were using a Mini-Circuits T9-1 step-down transformer to feed the

null-pot. This transformer, followed by a run of Cat-5 to a pot somewhere else was, in total,

presenting a weird impedance to the antenna not as close to a match as was the pure

resistance of a null-pot.

The differing lengths of Cat-5 extensions were changing that impedance as it was reflected back

to the antenna through the 9:1 transformer. THAT idea seems to account for most of the

inconsistent measurements.

The next move was to see if we could come up with a better step-down transformer that was a

better friend to the Cat-5 extension, and would present to the antenna something that looked

more like the impedance at the receiver end. We experimented with three different toroid

core sizes and materials, as well as various turn-counts that would keep us around 9:1. (This is

the sort of challenge the late great John Bryant would have relished!)

After a good deal of experimentation, we settled on a 21-to-7 turns ratio, wound “SS” style on

an 82-75 core. And THAT did the trick…at least for our situation (remembering YMMV).

Now we could remotely null some stations well down into the noise floor. (One of the locals

presents a signal of about -40 dbm, and the noise floor was below -115 dbm. Do the math.)

During this break-through test we still had the amplifier right at the receiver end, fed thru the

Mini-Circuits T9-1 at that end and, after amplification, on down the Cat-5 to the 2:1 transformer

and into the Perseus SDR. Putting an RF amplifier outdoors at the antenna may be impractical,

so as a practical matter we extended the amp back near the remoted null-pot. Nulls may have

been degraded a couple of db. But it was certainly more than acceptable.

A quick wrap-up and we’ll get to further D-Kaz applications:

With the change in transformers at the null-pot end we had a solution that would allow remote

nulling a lot deeper than a null-pot at the antenna. This we feel is because our lash-up looked

more like the receiver end’s lumped impedance than did a pure resistive null-pot. We also

confirmed Kaz’s expectations that a fixed average resistance will work if you’re willing to accept

less-than-deepest nulling.

The toroid we wound may or may not be best for your use so experimentation in this area may

be useful. An idea for the next time we do this is to substitute another 9:1 toroid for the T9-1

on the receiver end. Apparently these are very good:

http://www.qsl.net/k1fz/flag_antenna_transformers.html

We felt kinda dumb at this point, knowing better transformers were routinely being used and

we had to wonder if many of our experiments were time wasted.

But there’s no better way to learn something than to try it!

It may be that we simply got lucky with our home-brew toroid. All we know is: it works. The

two-station rows indicate a co-channel signal that came up after nulling:

By fine-tuning the null-pot it was possible to take some stations right down into the -115 dBm

noise; the data above is for a fixed (average?) null-pot setting.

One thing that was predictable (even early on) was that, as expected, the 140-foot D-Kaz was

somewhat better on the low end of the MW band than the 120-foot version. What wasn’t

anticipated was that null depth average and null-pot resistance vales were also a bit smoother

with the longer antenna.

TEST QUALIFICATIONS:

a) Through most tests, the nulls were generally repeatable over the several days of

observations.

However, due to the effectiveness of the antenna’s “backside” null, on some channels

there was a good deal of co-channel interference when the ‘candidate’ signal was nulled

and a station on the opposite side came up. So the null readings under co-channel

conditions were not necessarily accurate and readings varied with the beat of the

multiple station’s frequency offsets. (Near the end of this document you’ll find a list of

co-channel signals that came up simply by reversal of the antenna.)

b) Except in the few cases where the nulls were extremely sharp and deep, it was possible

to land the null-pot slightly off its perfect sweet spot. Thus, some resistance readings

may not reflect the exact pot position actually required for the best null. However we

made enough measurements over several days to identify and average the resistance.

We didn’t find time to measure the effect of unequal lengths of Cat-5 once we had the

transformer problem in hand. We’re assuming that for best balance in our situation, equal

lengths was probably best…but maybe not, since the toroid wasn’t identical to the T9-1

remaining on the receive side of the antenna…so here we go again!

Finally…the fast and dirty way to deploy a D-Kaz has been to tie the wires off to the end stakes,

putting the wires under tension to keep the deltas in shape. We now added insulators on the

wire ends…and gained a couple more db in performance. Yeah…insulators are just basic

Good Engineering Practice <g>

SOME ADDITIONAL APPLICATION INFORMATION

PATTERN-REVERSAL. A significant appeal of this antenna is its reversibility. This is done by

simply swapping receiver and null-pot ends. A handy remote-switching approach is espoused

by Mark Connelly and others. It assumes that you will be extending the ends of the antenna to

a ‘central’ location, whether it’s at the mid-point of the antenna, a nearby “shack” or your

vehicle (don’t park too close to the antenna). Quite simply, it reverses the antenna extensions

so “front” becomes “rear” and vice versa.

On the next page: a chart showing how well this works (because of the comparatively sparse

radio-station population within a few hundred miles.) This should not be confused with the

data just above. In the chart above the co-channel station appeared as the residual after

nulling. The list below displays the co-’s when a deliberate antenna flip is made and the so-

called “hot” side of the antenna is actually looking for signals from the opposite direction.

This group of stations was recorded about 100 miles south of the cabin where the others were

captured. (This kind of “flip-copy” probably wouldn’t be as easy in the Eastern U.S.) But this

should serve to give you an idea. Perhaps you can see why we like this antenna!

Here’s a diagram of that remote-switching system:

This has been combined in a mini-box we unimaginatively call a “Reverse/Null-box.” Here it is,

along with other transformers (and the null-pot used when directly connected to the antenna):

Note on the Reverse/null-box there’s a “vernier’ screw-driver-adjust put for deepest null.

When the amplifier is located at the receiver it’s connected to one of the BNC outputs on the

bottom (they’re labeled “main” and “friends”). The two Cat-5 extensions are connected to the

top of the null/reverse box. Now pattern-reversal is as easy as flipping the switch.

THE BROADSIDE D-KAZ

This variation takes advantage of the directive properties of the D-Kaz and, by adding a parallel

element, narrows the beam-width of the array. The resultant pickup pattern begins to

challenge the Beverage antenna…and without the Bev’s unpredictable side-lobes. The electric

set-up for such an antenna might be similar to this:

Here’s the proposed physical layout for a Broadside D-Kaz:

And now you see why we wanted to qualify those 240-foot Cat-5 runs during our tests. Of

course if we settle on an average fixed resistor for the null, we can simplify this.

DEPLOYMENT OF THE D-KAZ; SOME IDEAS LEARNED THROUGH SET-UP PRACTICE

You can set up one of these antennas in a half-hour or so if you’ve prepped the components.

Here are the parts; before we close we’ll offer a suggested step-by-step deployment procedure.

Have on hand:

Two 9:1 step-down transformers: See discussion.

One 2:1 step-down transformer.

One 250-ohm pot.

RF amplification of 10 to 15 dB.

Insulators for the wire-ends.

Two support poles. The 22-foot “fish-pole” is great.

http://store.kittyhawk.com/22-Foot-Heavy-Duty-Telescoping-Windsock-Pole-P1432.aspx

“Stands” for these poles: Two 2-foot galvanized pipes. Kaz says: “I use 1" ID pipes. 1" ID pipe has

an OD of about 1.31" and the masts slide over that very nicely for a good fit.”

Two pre-cut lengths of stranded wire: #18 or larger is best. If you’re building a 140-foot D-Kaz

the wires can be cut to 155 feet each; that’ll give you a bit extra. We use different wire colors.

Six wire-support stakes. Home Depot sells these in 12-pack bundles; 3-ft, pre-cut with points.

Drill a couple of holes near the top. Use them as end-posts and as wire supports.

You can also use one of these as the center-support…but we do something a little different.

Our center-support pole: This is where each of the delta ‘triangles’ comes to earth and is

extended as the base of the adjacent delta:

We drop a short, small-diameter pole into the ground and slide a 1-1/4” PVC pipe over it with

holes drilled at the 2-foot level for the wire-crossings. The PVC can ‘ride’ up and down and by

its weight puts a soft tension on the wires so the deltas keep their shape. We used a four-foot

PVC for enough weight to do some good without placing undue tension on the wires. Note the

PVC is slightly above ground, indicating a small amount of tension. (Or just use a stake.)

The Double-Delta antenna is really quite simple to construct and uses few parts. While we’re

not certain anyone has actually verified the pickup pattern’s conformance against the predicted

EZNEC patterns on these pages, there’s no doubt it provides excellent front-to-back ratios.

This fall we hope to deploy a Broadside D-Kaz in the desert near the Colorado/Utah border for

the purpose of chasing sunset skip…watching it approach from the East and following it to the

West. The D-Kaz coverage pattern provides a really nice sample of stations over a defined area,

and hopefully we’ll be able to publish a report of our success.

And that leaves one item: how to build the antenna. No doubt you’ve mentally constructed this

array (if not actually done so). Since we have one page left in this binder, we offer one fellow’s

version of how the antenna can be deployed in just a handful of minutes.

We hope you found something of interest in these pages and that you’ll take the time to

respond with questions/comments/challenges/suggestions. 73 and good DX!

Mark4 (at) Durenberger.com