Autofils,

Like you, I am quite interested in trying to pull in WNPI-23 digital here. With my current ganged PR-8800 + LNA setup on the tower, it almost works on quiet nights. I believe another 3dB or so might do it.

Now, the DBGH and ancestors are tuned to do best in the high-middle portion of the band. But we could retune them slightly for the longer wavelengths down lower.

What if we lengthened the wires slightly, say 200mm segments rather than 180mm, and bumped the spacing from the mesh up to 111mm to match.. ?

I wonder what your gain plots would say for that beast? :)

Cheers

What if we lengthened the wires slightly, say 200mm segments rather than 180mm, and bumped the spacing from the mesh up to 111mm to match.. ?

Speaking of which.. I have a related question about the current DBGH design.

My intuition suggests that the the current design should be spaced out 90mm from the reflector screen (1/2 the driver segment length), rather than 100mm. Even the original patent document suggests about 3.5" for this, or about 89mm.

Do we understand that 100mm number?

In post #114, mlord asked about the effect of reflector spacing...
http://www.digitalhome.ca/forum/showpost.php?p=715706&postcount=114

I had modeled reflector spacing for the SBGH, and assumed the same spacing for the DBGH. But I was also curious about other DBGH reflector spacings and since the cpu execution time is much faster for the 11 pair collinear rod, I ran a number of runs for reflector spacing of 95, 100, 105 and 115 mm.

Incidentally the cpu execution time is roughly proportional to the {number of segments} Squared. The 11 pair rods has 825 segments; the 30x75 screen has 5,049 segments, so if you do the math, I think you'll see why I ran with 11 pair rods.


DBGH 11 Pair Collinear Rod Reflectors - Reflector Spacing Variations

The results might surprise you, but the results indicate that 100mm is the best choice

This seems contrary to normal antenna theory with reflectors, so if anyone has an explanation, please tell me. I cannot explain this result.
Maybe I'm suffering from modeling fatigue....and can't see the errors.

My excuse, is and will always be....I'm just the modeling nerd :eek:

http://i258.photobucket.com/albums/hh262/autofils/DBGHRefSpacingInvestigation.gif

Doh ...!!! I did it again Doh...

Please disregard the photo in post 133, The plot for 95mm reflector spacing is wrong.
I knew there was something not right and I found a really stupid error in the 95mm spacing file.

So with a even redder face now, here is the correct info.


DBGH 11 pair Rods: Effect of Reflector Spacing

Now the results make sense. The gain does increase with closer spacings, as antenna theory predicts.
The reflector spacing affects the swr, but remember that the net gain calculations are the model's raw gain corrected for any impedance mismatch.

The results show that closer reflectors spacings give a slight increase in gain at the high frequency end.

The numerical gain from these results at Ch 46 are:
100mm - 16.15 dBi
95 mm - 16.21 dBi
85 mm - 16.29 dBi

Note: The channels are plotted at the mid point of the channel's 6 Mhz bandwidth; thus Ch 46 is 665MHz.

In theory a closer spacing to 85mm gives +0.14 dBi, but in practice there's no significant difference.
What it does tell us is ....the tolerance on the reflector spacing should be kept within 85 to 100 mm !!



http://i258.photobucket.com/albums/hh262/autofils/DBGH11PairRefSpacing-2.gif

Great!

Now we just need to find someone with a 1GHz bandwidth scope to compare theory with reality! My scope here only goes up to 200Mhz -- good enough for (most) VHF, but not nearly high enough for UHF.

Somebody else out there must have something suitable..

EDIT: or maybe 200MHz is enough, in combination with a suitable downconverter?

in combination with a suitable downconverter?

Like an old TV with an accessible IF circuit?

Let me note that my analysis was a total scaling; make all dimensions the same factor larger. This "progressive" Hoverman tries to vary the dimensions continuously with the same electrically connected elements. This is very different from what I was doing, and will need it's own rigorous model to see what happens.

But, you never know ...
Frank

Frank,

I really wasn't ready to post any info on this Ottawa Down_Shifted DBGH 11 Pair version, but since mlord has mentioned it, I thought I had better respond, just to make sure that incorrect assumptions are not made by readers of this forum.

In the "Calling all Nerd Modelers to push the envelope !!" in post 149, I mentioned this task as one of the many tasks that needs further modeling work. The reason I am looking at it now, is as mlord has stated. Three of us have just recently got together (last wednesday night) to find ways to improve reception on the only two american digital channels currently available to us [Ch 23 and 38]. [mlord, tvlurker, myself and a fourth GerryB is coming]

We have an extensive testing program to run and we will probably not be reporting on this activity, until it has been completed...as the plans are just starting to form, and the work is only preliminary at this time.
This testing will be on design variations of the GH(single/double); GerryB's integrated Yagi-LNA preamp and possibly a version of GerryB's preamp as a separate unit, plus other odds and ends.

I see that mlord has tried to clarify his posting about the design of the Ottawa_DS version. All I will say at this time, is that it was scaled to shift the freq down by about 15-20 Mhz. So that means; all the array elements ( lengths and spacings) are scaled upward. The reflector lengths are scaled upward, but we maintained the same reflector spacing with respect to the array and vertical position of the reflectors. The feed-gap and reflector gaps remain the same. If our field tests show "proof-of content" then as I suggested in post 149, this design should be optimized and offered as another version that favors the low end of the new UHF band.

I invite you to join my nerd call, and assist with modeling tasks. I know that your passion is down_shifting the 8 bay bow-tie, so if you would rather do that task, that's great, because I see this nerd modeling club as taking on other projects besides the Hoverman variations, with the objective of improving OTA reception for both VHF and UHF.

To that end, you could grab the CM-4221 ez file from Ken Nist's web page, strip out the reflector screen, to get a base CM-4221 array that you can then modify and test. You might find the steps I outline in post 150 to be useful.

I guess a lot will depend on DCH'ers interest in modeling. If there is little interest, then it will just flow to the waste basket, much like my idea of a contest for the 3 best builds of the Gray-Hoverman, posted on this forum during the 2008 year.

On the other hand, if there are a lot of modeling nerds, then imagine what more effort might yield, Sort of the hardware equivalent of open source Linux.

I can only judge the interest in this, by readers reply postings to post 149 and 150....time will tell if this has any wings..


Meanwhile, if you do pursue the modeling of downshifted CM-4221 and need any help stripping out the reflector screen send me a message and I could help on that score.

Regards
...autofils

Frank,

All I will say at this time, is that it was scaled to shift the freq down by about 15-20 Mhz. So that means; all the array elements ( lengths and spacings) are scaled upward. The reflector lengths are scaled upward, but we maintained the same reflector spacing with respect to the array and vertical position of the reflectors. The feed-gap and reflector gaps remain the same.


I guess I was tired when I posted this last night, and fighting a flu bug.
I didn't define the Ottawa_DS 6 pair collinear rod reflectors correctly.

What I actually modeled was:
- The reflector lengths scaled upward and the vertical spacing of the reflectors was also scaled upwards, but I maintained the same reflector spacing with respect to the array, at 100mm.

- The lengths and spacings of the array were scaled upward and I maintained the same feed-point gap (44mm) and reflector gap (20mm) used in the GPL'ed SBGH. We will experiment with the length of the hoizontal stubs.

Believe or not, this modeling does take a bit of time and I have only been experimenting with this since our Wednesday meeting, so this is very preliminary info. If we could mobilize more modeling resources, this work would go much faster, and likely yield better results. That's why I made the call for more nerds to help out.

If our test results confirm the model results, I will post the design info, and the nerd modelers can participate and help in it's optimization.

If anyone wants to try a simple modeling task to "wet their whistle", I would suggest look at optimizing the feed-point gap with the array-only to get started. ...Any volunteers?

Here is an update re the status of the Gray-Hoverman design evolution.

Short Historical Summary

This project started with modeling the original Hoverman 4 pair Collinear rod reflector design from measurements given in patent 3,148,371.

Further modeling analysis showed that significant improvements were possible by implementing a 6 pair collinear rod reflector, and this design was released under GPLv3. Included was a non-optimized DBGH design, as well as some modeling information using split-screen wire-mesh reflector on the SBGH and a full wire-mesh screen for the DBGH.

A lot of additional modeling analysis and research has been conducted for further optimizations of the SBGH 6 pair.


Today I am releasing preliminary modeling data that shows further significant gain performance in the low end of the post-February 2009 UHF band (470 to 700 Mhz ) for the SBGH design.

It is using a symmetrical design of 8 pair collinear rod reflectors. Further investigation is required to determine if a "contoured mesh-screen" can provide additional gain compared with the "standard" rectangular mesh-screen.

There are still a lot more optimization trials to conduct, before issuing the design dimensions for the GH-8 pair collinear design and the optimal mesh screen size and shape.


Gray-Hoverman Design Evolution

Here is the Net gain plots, that show the progress in the gain performance from the original Hoverman 4 pair collinear design to the current "in-progress" preliminary data for v3.0x 8 pair collinear design.

The v3.0x design version is aimed for fringe to deep-fringe reception.
The preliminary gain plots for v3.0x show the net gain for 8 pair collinear reflectors and the additional gain which would result by adding a mesh-wire screen. Looks like an added mesh screen can give up to +1dB gain...important for deep-fringe.

I would encourage those creative folks to post fabrication ideas for adding a wire-mesh screen, ( maybe even chicken-wire) using the rod reflectors as the supporting members. The reason I mention this, is because prelim data suggests that a contoured screen may give higher gain than the simple rectangle mesh screen. This is related to the "tuning-effect" seen with the different lengths of collinear rod reflectors.

http://i258.photobucket.com/albums/hh262/autofils/GHHistory.gif

The net gain curves show the design evolution.
Just a note about those strange frequency values along the X-axis. Each channel occupies a 6 Mhz bandwidth. The UHF band starts at 470 Mhz, so Channel 14 occupies 470 to 476 Mhz and it's mid frequency is 473 Mhz... and Ch18 mid-freq is 497...etc...[So now you know...the rest of the story.."Paul Harvey"]

1. The original Hoverman patent 4 pair collinear design (blue)
2. The SBGH GPLv3 6 pair collinear design (green)
3. Preliminary data for SBGH v3.0x 8 pair collinear design (red)
4. Preliminary data for the v3.0x showing the additional gain by using a wire-mesh screen (35x46in) (dotted blue-red)

The current thread concerns only the second generation SBGH2 and DBGH2 antennas, known originally inside the Gray-Hoverman project as Version 3.0x as work continues.

The first generation SBGH and DBGH are discussed in the following existent thread:

http://www.digitalhome.ca/forum/showthread.php?t=81982

Terrific work, Autofils - this is very promising!

I would encourage those creative folks to post fabrication ideas for adding a wire-mesh screen, ( maybe even chicken-wire) using the rod reflectors as the supporting members. The reason I mention this, is because prelim data suggests that a contoured screen may give higher gain than the simple rectangle mesh screen. This is related to the "tuning-effect" seen with the different lengths of collinear rod reflectors.


Autofils,
What do you specifically mean by "contoured screen" ? Do you mean like bending the edges of the screen a few inches in the direction of the TV station, like in the CM4228 ? (Ive also noticed that in some of the newer Channel Master UHF antennas, they bend the screen away from the TV station direction, go figure.) Or contoured in a parabola type way ?

And "wire-mesh screen (35x46in) ". Does that mean 35 inches wide by 46 inches long or vice-versa ?

Chicken wire overlaid and attached with nylon wires ties on a six pair SBGH also helps in other ways, like straightening, spacing and firming up the reflector rods. Of course then the question becomes, 1 inch chicken wire or is 2 inch chicken wire sufficient, considering there are already reflector rods in place. The 2 inch stuff would have less weight and wind load.

Im half way through building a second identical SBGH, which Ill attach at the 127mm spacing between the last and first rods and elements to have a DBGH. (easy to do, one of the benefits of pvc construction) Of course, this is will make it slightly different from the GPL DBGH, as Ill have an additional rod. But seeing where the rod placements are, behind the stub element, I think it will help not hinder.

Autofils,
What do you specifically mean by "contoured screen" ?

And "wire-mesh screen (35x46in) ". Does that mean 35 inches wide by 46 inches long or vice-versa ?




300ohm

The "contoured mesh screen" refers to the shape of the screen. Most mesh screens have a rectangular shape of WxH. The length of each of the reflector pairs of the SBGH-6 and SBGH-8 are not equal and have different lengths due to the "tuning-effect" seen in the modeling results. If you have ever investigated tuned RF or IF stages, you would be familiar with over and under-coupled LC circuits..., well the variation in reflector lengths shows the same response. This effect makes the modeling optimization process rather difficult...but that's another story...... well ok..... here is the story outline....

If you visualize using 33 reflector pairs, each of a different length; the modeling task is to determine the lengths of each pair that maximizes the overall gain-bandwidth. The "contoured mesh-screen" then becomes the shape that is defined by the optimized lengths of the theoretical 33 reflector pair model. (... I think this would make my old Queens Prof Penstone proud !!...)

The Gain plot of the v3.0x showing the additional gain obtained by adding a mesh-screen, was modeled with a rectangular mesh-screen 35" wide by 46" high. The size was chosen as a first-cut trial example and is not optimal.

For deep fringe reception, I plan to fabricate the 8 pair rod reflectors and then fill-in the gaps between the reflector pairs using small-mesh chicken wire, which can easily be shaped to whatever length variations that the optimized "psuedo 33 reflector pairs" turn out to be.

I think the 2" mesh would be fine, but you could experiment with adding an overlapped layer to see the effect of a smaller mesh.

The fabrication method needs to fasten and stretch the chicken-wire onto the reflector rods. With all the creative folks on this forum, we should get some really great fabrication ideas.

If there are any members that wish to participate in the v3.0x modeling optimization process, contact me for details.

The "contoured mesh screen" refers to the shape of the screen. Most mesh screens have a rectangular shape of WxH. The length of each of the reflector pairs of the SBGH-6 and SBGH-8 are not equal and have different lengths due to the "tuning-effect" seen in the modeling results.

Ahhh, contoured like a shapely woman.

If you visualize using 33 reflector pairs, each of a different length; the modeling task is to determine the lengths of each pair that maximizes the overall gain-bandwidth. The "contoured mesh-screen" then becomes the shape that is defined by the optimized lengths of the theoretical 33 reflector pair model.

Are you saying that the vertical elements in the chicken wire can be ignored in the model because we're (in general) looking at horizontally polarized signals? Can't there still be a current in the vertical element if the potential in two adjacent horizontal elements is not the same? Or am I betraying my lack of RF modelling knowledge here?

Also, if you follow Doug Lung's RF technology columns, he seems to be pushing Circular and elliptical polarization to improve coverage. (He's also pushing beam tilt, which would kill Watertown reception for us in deepest fringe Ottawa)

TVl

TvL,

You are quite right about the induced currents in the chicken wire, but my approach is to ignore those details and start on a simple basis, with the modeling in a series of "baby steps" with quite a number of assumptions. I start with 8 collinear pairs and then expand that to 33, as a means to emulate a "screen-reflector". So in these models, there are no vertical elements.

I don't intend to model the "chicken-wire" mesh-screen, but rather make field tests, with and without the chicken wire. I'm betting on the chickens...

I start with 8 collinear pairs and then expand that to 33, as a means to emulate a "screen-reflector". So in these models, there are no vertical elements.


Hmmmm, so in the model above if I understand correctly, V3.0x Prelim 0.89 X 1.16 SCR, cutting even more of the vertical elements in the screen mesh doesnt affect the model ? Which would give it even more wind resistance. Cut 2 save 1. ( As you may note, wind resistance is an important factor in the design for me. 40-70 mph winds are common sometimes around here. ) Also the .89 dimension refers to common 1 X 1-1/2 hardware cloth/fenceing, commonly known as rabbit wire in some circles ?

Hmmmm, so in the model above if I understand correctly, V3.0x Prelim 0.89 X 1.16 SCR, cutting even more of the vertical elements in the screen mesh doesnt affect the model ? Which would give it even more wind resistance. Cut 2 save 1. ( As you may note, wind resistance is an important factor in the design for me. 40-70 mph winds are common sometimes around here. ) Also the .89 dimension refers to common 1 X 1-1/2 hardware cloth/fenceing, commonly known as rabbit wire in some circles ?

300ohm,

I think you are trying to read-in much more detail than currently exists regarding the screen-mesh, as my modeling is restricted to just simple horizontal collinear reflector rods. Based on those results, I will build my reflector using 8 pair of collinear rods, and I'll fill the gap between the rods with some sort of chicken-wire and use the "tried &true field test" to determine reception, with and without the screen.

A crude measurement technique that might work, is to insert an adjustable attenuator in series with the antenna, and adjust it for the "digital-cliff" effect cut-off without the screen and re-test with the screen attached. Hopefully the signal will magically appear with the screen attached.

I have noted your requirement for low wind resistance, and that is a major reason that I am considering "chicken-wire".

RE the .89 x 1.16 Scr notation on the plot...That's simply the WxH of the screen in meters, which I converted to inches for those metric-challenged folks. :p

RE the .89 x 1.16 Scr notation on the plot...That's simply the WxH of the screen in meters, which I converted to inches for those metric-challenged folks.


Oh OK. Heh, I thought it was mesh size. :)

Ive always been fine with meters,cm,mm etc measurement and the Celcius temperature. But what throws me for a loop is that darn windspeed in meters per second, or something like that instead of miles per hour. It just never seems quite right to me.

It is using a symmetrical design of 8 pair collinear rod reflectors. Further investigation is required to determine if a "contoured mesh-screen" can provide additional gain compared with the "standard" rectangular mesh-screen.

To me, this sounds like the rods are not really needed (except for structure), and what is more important is the shape of the mesh -- after all, once the rods are covered with mesh, they really don't have much to do.

Or another way, perhaps, is to just keep filling in with more rods, with lengths contoured to whatever works best in simulation, until they're all no more than, say, 2" apart from each other. Thus, the rod equivalent of a mesh, but without those nasty non-horizontal segments.

And perhaps one could convert an existing SBGH/DBGH into an approximation of the "new" design, by simply infilling with more rods -- say, 11 pairs in total. That's the original 6-pairs, plus 5 new pairs of infill.

Makes sense -- the more rods, the closer we get to a full mesh, except this kind of mesh has only horizontal segments.

Cheers

I have been slowly making progress on models for the Gray-Hoverman Generation II and this post provides an update that shows the progress being made in achieving a top grade hi-gain UHF antenna.

In modeling the "tuning-effects" of the collinear rod reflectors, I had mentioned that the gain-frequency characteristic was very similar to previous work I had done with tuned RF-circuits, and I found that with a slight "over-coupled" circuit, you can produce two peaks in the gain-frequency curve, which are separated by about 100 Mhz. Utilizing this "Twin-Peaks" property, it is possible to make significant improvements in gain for the lower channels, and about another 1dB for the upper channels.

Since the bandwidth of the post 09 UHF band (470 to 700Mhz) is greater than the "Twin-peaks 100Mhz bandwidth", two versions of the GH-Generation II will be required to cover the 470 to 700 MHz range.

The NetGain plot shows the preliminary data for v3.1x SBGH with 8 collinear rod reflectors for UHF-Lo and UHF-Hi versions. The NetGain of the current SBGH GPLv3 design is also shown, to provide comparisons to this new "Twin-Peaks" v3.1x.


Antenna Version...........First Peak Gain (dBi).............Second Peak Gain (dBi)

UHF-Lo ........................ 14.6 @ Ch16 ...................... 15.2 @ Ch32
UHF-Hi ......................... 14.8 @ Ch 32 ..................... 15.2 @ Ch46

Comparing the "Twin-Peaks" to the SBGH GPLv3 design, you can see that there is a very significant increase in gain for the lower-end channels.

http://i258.photobucket.com/albums/hh262/autofils/GH-v31x.gif

I am working on the optimal shape for a mesh screen reflector, and then will finish this project with a DBGH design for UHF-Lo and UHF-Hi.

Assuming the Mesh screen will add another +1dB for the SBGH, and a DBGH design can add an additional 2.5dB gain, the DBGH Generation II could have an overall max gain in the range of 18.5dBi :p

.... Stay tuned for further updates ... coming soon.