A Modifed AntennasDirect DB8

The motivation here was really just to prove the concept; that the reflector of an N-Bay design could be improved upon.

The common reflector of most N-Bay UHF antennas is known as a "Planar Reflector".  In contrast the reflector implemented here is described in "The ARRL Antenna Book, 21st Edition, p 18-17" as a "Trough Reflector".  It is a modification to the more typical "Corner Reflector" (as seen on the 91XG antenna).  To quote the ARRL book, "To reduce the overall dimensions of a large corner reflector the vertex can be cut off and replaced with a plane reflector".

Here are a couple pix of the modified unit in service:
 

 

Modifications
There were three separate changes implemented.  These include a change of harness, the addition of end stubs to close the dipole loops and form a fully triangular outline, and the trough reflector.  Each of these are outlined in the sections that follow.

Stock Unit Response
Before detailing the results, it might serve to get calibrated with a capture of the starting point.  I.e., what I'll call the performance of the "stock" or unmodified unit.  The capture below provides a view of the response at the test site for the normal horizontal polarization, as well as with both Tx and Rx antennas rotated 90 degrees to a vertical polarization.

Note that the vertical scale is 1 dB/division, and thus all pertabations are somewhat exaggerated.


From this you can see that that there are numerous imperfections at the antenna site, but the important aspect will be the difference between varous captures, not absolute levels.

The main things to note about this starting point include [1] the null in response at about 480MHz, and [2] that in general the vertically and horizontally polarized responses are within a few dB of each other, and demonstrate the same overall trends.

Combiner
The first changes made were inspired by a number of Ken Nist's comments and observations at An HDTV Primer on the DB8 and CM4228HD harnesses.  These sparked me to wonder if the harness could be improved upon.

I replaced the stock harness with 2 identical 19" lengths of Belden 7916A RG6, (one of the lowest loss RG6's I've come across to date), and a Pefect Vision PV23-233 Splitter/Combiner (see Combiner Comparisons).

Here is the result.  (For this and all all remaining comparisons only the stock horizontal capture is included for reference).

The result was not 100% positive across the entire band.  From about 530 to 580MHz, channels 24-31, it resulted in about a 1 to 1.5 dB loss.  However, there was almost a 2dB improvement over the lowest 20's channels and at about 44, all of which were of keen local interest.  In addition, the change generally produced a more consistent (slightly flatter) response across the band.  I decided to incorporate this change.

I suspect that by experimenting with different lengths of cables from the baluns to combiner (other than 19") , and/or with different combiner models, it would be possible to move response peaks and dips to various other points in the band to optimize for other local needs.

Closed Triangles
The second change was an experiment inspired by readings of the "Antenna Engineering Handbook", Johnson & Jasik, section 29-5.  This section, titled "Triangular-Dipole Antennas" states "A dipole formed from two triangular sheets of metal is sufficiently broadband with respect to gain and VSWR (with respect to 300 ohms) for all-UHF channel reception".  It goes on to state "Further simplification of the triangular dipole to a wire outline of the two triangles results in a significant degradation of its broadband performance.  However, the metal triangles can be approximated with wire mesh ...".

I subsequently experimented with 1/8" and and 1/4" mesh screens, cut to appropriate triangular shape and size, then fitted them around each dipole section.  Alas, I did not observe any improvement.  To the contrary, there was significant degradation at the high end.  I am still pondering why this contrasted with the Johnson & Jasik writings; I have some intuition that it may be related to a shift of impedance resulting in a poorer match with the built in baluns.

Nevertheless, on impulse, I picked up some 3/16" tubing from my local Ace hardware and fashioned small end pieces (described on the Construction Details page) that fit perfectly over the existing wire, and which have the effect of "closing the loop", thus forming a complete outline of the triangle.  Installing 16 of these over the ends of each of the dipoles, along with the replaced combiner yields the following result, compared to the stock unit.


The most dramatic improvement is obviously at the low end, for channels 14 through 18.  But there were other approximately 1dB improvements across the band.  The following capture illustrates these by comparing the response for both the combiner and triangle stub modifications versus just the combiner modification.

Above about 680MHz, channel 49, its clear the closed triangles begin to have a negative effect.  Locally, however, there were no channels of interest above 44 so this was not of concern, and I ultimately kept this modification.

Trough Reflector - the final result
Finally, the most mechanically dramatic (and laborious) modification, the addition of the upper and lower reflector sections to form the trough reflector.


The combined result was quite satisfying.  Even discounting the deep dip at the low end as a test bed anomoly, the improvement is about 3 to 4 dB across the band up to about 685MHz, from channel 14 up through about 50.

Just in case you're interested in the incremental effect of adding the reflector, the following capture compares the response of all three modifications to just the combiner and triangle stub modifications..

This indicates that the reflector, by itself, contributes a pretty uniform and consistent improvement across the band, starting at an average of about 1.5 dB at the low end, and tapering off to about .75 dB at the high end.

Field Measurements for locally available UHF Channels provide some supporting real world data.
Construction Details are documented with a few useful photos.

Summary
The experiment was quite satisfying from an academic perspective; a good bit of empirical knowledge of the effects of various tweaks and adjustments was acquired.

I kept this unit in service for about 6 months; the exact same unmodified DB8 had previously been in service for about 18 months.  Over the service period, one aspect I "felt" but did not quantify, was a slight improvement in the stability of the digital signal "quality" over time.  I.e., a lower variance.  Not huge, but perceptible.  The comparisons on the Field Measurements page between the modified DB8 to both the stock DB8 and a 91XG Yagi continue to fuel a theory that N-Bay designs may provide a slight long term reliabilty advantage over Yagis under certain adversarial scenarios where specific types of reflections are involved (multipath).  It is based on the observation that this modified design has a dramatically larger "footprint", or "capture area", than the 91XG.

While Yagis like the 91XG can often provide slightly higher level signals, there is some empirical data here that hints the spatial diversity of the planar multi-element N-Bay may yield better long term stability of signal quality, albeit as a small sacrifice in signal level.  For adequate margin, the reality is you need both: quantity and quality!

In the fall I took the modifed DB8 down to relocate it 85' up in a neaby Pine tree; but winter set in it before the re-install could be completed.  It is presently on standby.

On reflection, I may or may not re-install it.  While I'm convinced that signal levels can be improved, my reality is that there is adequate signal level for several local stations, but they are significantly impacted by reflections related to the local mountainous terrain.  The addition of a reflector such as described here does increase the capture area by about 80%.  However, in contrast, vertically mounting two stock units doubles the capture area; a 100% increase for a lot less work.  Further, like the trough reflector addition, vertically mounting identical units also has the effect of reducing the vertical aperture, thus reducing sensitivity to nearby ground reflections.

So you could argue that a slightly better (and much less labor intensive) route to take would be the vertically ganged off-the-shelf 8-Bay scenario, and I would go with the CM4228HD if the weakest channels are at the low end, or the (new) DB8 if you need to favor the high end.