Im optimizing the 8X8 KISS4P for max low channel gain over the range of 470-518 mhz now Cool. Yeah the 4f's are harder than most on the SWR of the feed. Don't bother with to much in depth study (of the softkiss). I don't think it beats the 4fX36 (~36 sq ft) by much. It is actually a little slice of two cones base to base.

I don't think we'll need to use a form to bend the mesh.
It is an attempt to model the kiss build with no folds or creases. Just overlap, stitch and hang on a mast.

It is a little to sharp at the Y-axis, and has some more minor discrepancies.

I don't think can get much closer without going from measurements of an actual build.

The vertical pattern needs a little work but check out the horizontal/azimuth views pattern.
But
I might explore what is going on at 2.4 and 10 GHz (re-segment for high freq.) see if it's an error or really 69 dbi ?

Most gh models are to narrow in beamwidth and to high gain to perform well as a feed.

Was thinking of using only the GH6 reflector rod design ... and keeping the LPDA feed. (not using the GH zig zag receive elements).

So that idea was - four (4) GH6 reflectors, placed and angled like the KISS planes - reflecting towards the LPDA feed (receive element).

I see as a reflector not a feed.
Correct me if I wrong (I don't know the GH models as well as I should)
gh6 has a reflector spacing of ~6-8 inches. It'll work but 8" at uhf is large
You could rip/delete 2/3 -3/4 of the rods out of the kiss4p.nec (http://ghz24.freeshell.org/antenna/models/newermodels/parabolas/kiss4p.nec) model and see what happens

I see you posted again: I'll have to carefully read it again but I dead- ended a model called wings
edit
Looking at post #340 It seems you would like to increase the number of horizontal straight sections from two to three \__/ (top view) like this?
If so then - at 6 feet diameter the difference between a full (~perfect) and all horizontals changed to two straight planes/rods is 0,6 dB AVG over 473-695 MHz. and the optimizer claimed the position of transition from flat to angled did not effect gain.
At larger sizes ( 8+ to 12) I'm sure the winged models would have showed their value.
The vertical curved approximated by a straight line cost ~1 dB avg.
Also the lpda feeds pattern is higher than it is wide.

Not exactly.

I'm thinking of the Chevron-Trough Reflector something like this:

< Top view: Chevron (angle exaggerated)

/
| Side view: Trough
\


And if the LPDA illumination pattern is Tall and Narrow, then the reflector should be made similarly. Taller and more narrow.

So one needs to "invest" reflector material wherever / and maybe more densely, where it will give the more "returns" - i.e. where the bulk of the LPDA feed pattern is. Center, and taller - up and down, where the LPDA pattern is.

Other idea:
(possible for *any* reflector design, KISS 4-plane, and/or Chevron-Trough)

Use Closer pitch mesh in the Trough / center area.
and Wider pitch mesh top and bottom on the outer edges or "wings".

So, making your reflector more Dense, where the strongest part of the LPDA's pattern lies - the center ?

So it's the vertical curve you want to make closer to a true parabola.
That is where more gain was lost (by reducing vertical curves to a straight lines) and as the height increases the straight line becomes a limit (to gain increase)
Yes more closely approximating the vertical curve with more planes helps not lose as much gain and allows taller reflectors to be a benefit.

then the reflector should be made similarly. Taller and more narrow.
that's why we were discussing 6x9 foot and 8x12 foot sizes.

So, making your reflector more Dense, where the strongest part of the LPDA's pattern lies - the center ? Increasing the density (of wires) is really easy on the surface patch models.
it would increase the density across the whole reflector .
But that would tell how much could be gained from tighter mesh fence or doubling the fence. I don't think there is much extra gain to be gotten by this strategy and most of that will be at the higher frequencies.
But at ~ 24 lbs weight for a 8x12 version doubling up parts may be an option.
When I get back to my modeling computer I'll do a quick comparison if 300ohm doesn't beat me to it.

BTW the models I referred to as chevron are full curve parabolas in the vertical (only horizontal curves became straight lines)

Edit: Increasing the number of wires over the whole model (horizontal only) from 36 wire (every 2 inches) to 72 wire (every 1 inch) gives ~ 0,04 dB difference (up to 800 MHz.) IMHO that's not worth the extra weight.

But at ~ 24 lbs weight for a 8x12 version doubling up parts may be an option.
Another option is just stacking two 6X6's for a 6X12. Add about 2.5 dBi to the 6X6 gain chart and it may exceed the 8X12 version in a narrower package.
(That seem logically because smaller spread out parabolic radio telescope arrays are much higher gain than one large parabolic radio telescope )
If the KISS4P follows the parabola aperture shape, ie the aperture is less than the diameter of the parabola, then the best practical stacking distance is the ends of the reflectors next to each other.

I just ran a scan (step 12 cause it's big) on a 8x12 chevron -12 gauge wire model
It beats or matches the wade double 8 foot big boy the PB-82-BB :D

21.92 dBi at 473 and a nice flat slope up to peak at 24.31 dBi at 677
and it's still rougher that the intended build.
I also have played with a scale model of 1/2 inch hardware cloth 3'x3' .
I'll try and post some pics
It did remind me of kite building.
similar to the way those diamond KITES are made.
a lot.
Good idea. If using 4'X4' panels or lighter gauge 3'X3' panels, each panel could be stiffened up by using light fiberglass driveway poles in an X formation and wire twisted to the mesh.
Stiffeners could be helpful and may be necessary but I didn't need any.

Regarding the mast mounting of the non-moving panel KISS reflector, something like this five point mount (or is it an 8 point mount ? heh) would be pretty easy to do and sturdy.

I like the spars (yellow) and the feed support continuing to support the apex.
I gave up all remote adjustment with the multi-overlap versions as they are to rigid and to close to a perfect parabola to get much benefit.

I don't seem to be able to uplaod any pictures (even previously "valid" ones)

So for now I put some pics of the scale model (half scale of a 6x6 ) in a zip (http://dl.dropbox.com/u/55862093/para3.zip) It is hard to photograph and show the shape shadows can completely obscure and confuse the shape.
The tilt at the max. Z value is about 3 inches and at max. Y it's way to big at about 9.5 a spreader (instead of the green cat5 strands) helps get this to a better value of about 5 inches and tilts the corners in more. Overall better shape.
This might be a good place for stiffeners or spar like structures 300ohm has suggested previously.
the vertical cat5 wires are not tight and don't change the vertical curve much.
The diagonal wires are needed to keep the opposing corners (upper right and lower left )from springing back and forth from (higher x value) than the other two corners and lower x values than the other corners.

This might be a good place for stiffeners or spar like structures 300ohm has suggested previously.
Like in post #324 ?

I just ran a scan (step 12 cause it's big) on a 8x12 chevron -12 gauge wire model
It beats or matches the wade double 8 foot big boy the PB-82-BB

21.92 dBi at 473 and a nice flat slope up to peak at 24.31 dBi at 677
But what was the SWR ? Ive been running the Evolver on the 8X8 KISS4P , 470 - 518 mhz, for about 5 days straight using the same criteria that I used on the 6X6, and cant get the SWR down to acceptable levels, ie > 2.7 max, let alone 2.7 average.
http://img207.imageshack.us/img207/1926/kiss8x84pevolverpic.jpg

Either Ill have to weigh for something like 100% SWR/25% Gain or develop a lpda feed thats closer to 100 ohms.

But what was the SWR ? 2.1 max.
I ran the 8x8 model with a set of the variables I saw in the pic you posted you forgot to change the impedance back to 75 ohms. I recognized it quick cause it happens to me all the time.

The SWR is so stable I've been not weighting it at all (the optimizer ignores it) and just hoping for the best. And mostly I haven't been to disappointed.
So better results than I've posted may be available by weighing swr net gain wise.
But at ~2 max, raw and net gain are real close.

edit: like 324 yes.

2.1 max.
I ran the 8x8 model with a set of the variables I saw in the pic you posted you forgot to change the impedance back to 75 ohms. I recognized it quick cause it happens to me all the time.
Opps, I did the same, heh. I had started with it set at 75, but then a couple of days later playing with another NEC file, changed it to 300. Setting it back gives an acceptable 2.45 SWR max for me on the 470 - 518 mhz model.
The Raw Gain for me is pretty smooth at 19.11 min to 20.75 max, 470 - 698, but SWR is relatively bouncy bouncy 1.54 min to 2.45 max, 470 - 698.

Trying to show the vertical (seem) curve, view along back side.
http://dl.dropbox.com/u/55862093/para2.JPG

http://dl.dropbox.com/u/55862093/para3.JPG

I cut the horizontal (green) and vertical (blue) wires with little change in shape and added a spreader to force it open more. The spreader is just held there by the structure pressing in.
http://dl.dropbox.com/u/55862093/para6.JPG

You can see at the corner this is reclaimed wire not factory flat and smooth or uniformly curved like new fence would be (hopefully)


Setting it back gives an acceptable 2.45 SWR max for me on the 470 - 518 mhz model
Should probably re-optimize with it set to 75 ohms it was probably giving away gain to get good swr relative to 300 ( i think you inadvertently asked for higher swr, closer to 300)
I'll bet it optimizes overnight in 38 steps.

I don't seem to be able to uplaod any pictures (even previously "valid" ones)

Yes, me too early this morning.
Unable to upload pictures to my album (on a totally different subject).

Tried to upload 2 pictures to my album, then tried them 1 at a time, as suggested and I got an "Invalid File" message.

Both are 800x600 jpg less than 100K in size and should be O.K.


I put a message for the Forum Admins In this section:

Digital Forum > Digital Forum Help and Administration > Member Feedback

In an earlier / existing Thread called something like "Uploading Pictures to Album failed"
(might be related to recent change of server for the DHC forum ? Not sure.)

Thanks. Just wanted to corroborate ... "me too".

EDIT: Working now ... just uploaded 2 pictures to my Album.

This model is of a 8x12 (could be 8x8 with deletion of the wires above 48 z value ) showing "what if" three vertical planes and what angles/distances.

http://dl.dropbox.com/u/55862093/8x12Build3varchine.jpg


8x12 three chine model (http://ghz24.freeshell.org/antenna/models/newermodels/parabolas/8x12chined.nec)


6x6 three chine model (http://ghz24.freeshell.org/antenna/models/newermodels/parabolas/6x6chined.nec)

That 8X12 looks good, and actually looks like a bit more beamwidth on it than Ive seen on other 23 dBi models (maybe as much as 20 degrees ??). As Martha Stewart would say, thats a good thing, heh. What freq was that and what was the SWR ? (I wish Arie would include MHz and SWR in those 3D screens in future updates as the data is available and theres enough room for them in the top left hand corner IMO, ie for two lines of text)

I see you went back to rods and GX card combo. The SM/SC card and GX card combo does really seem increase the compute times on large models. Im not quite sure what going on. From what I understand, the GX card stores the coordinates to flip around in a matrix. I using the max memory on XP 32 bit > 3.2 GB of memory, so the matrix has got to be huge, unless the NEC2 engine is restricted from using over X amount of memory core, which it may well be because that was a common programming practice in the early days.

As long as the folds are on a 2" interval for 2X4 mesh, making more bends is quite easily doable on the KISS reflector and still making it very easy.

It's 23.86 dBi @ 585 MHz with 1.27 SWR ~3 dB beamwidth is 12 deg. wide and 8 deg. high. (actual gain at 6 deg off center is 21.29, at 8 deg. it's 19.21)
But It was run at 2 degree resolution ( I do 5 for optimizer, might go to 10 deg. like you do )so it is closer to 13 or 14 degrees. (12 degree is actually only 2.5 drop in gain)
The extra height helps squash/flatten the pattern.
That's getting pretty narrow though, and extreme gain is worthless if you can't hold it on target.
I may be at (or even past) a point of diminishing return.
Maybe we should give a little gain back in trade for a wider sweet spot?
I've heard "high gain" antennas can be hard to aim. Sloppy rotors?
But I think these are even more challenging.
I only have experience with my present rotor the Armstrong Direct Observation model. Slippers and robe not included!
Yeah the main reason I went back to wires is it's easier to pull dimensions from the model. The SM/SC combo is useful and eloquent but I don't like the fact geo. editor can't convert those models into strictly numbers only models.

I was perusing the beginning of this thread and saw post #5 about the wade parabolics that I ignored before ( I wasn't considering building/coping one then). I wont quote it all here again but here's a link.
http://www.digitalhome.ca/forum/showpost.php?p=755156&postcount=5
stampeder I see you posted yesterday and I'm hoping you or someone can elaborate on post #5 cause I'm not getting it. You say not much more gain than a cm 4228 ??? :confused: 24-21 dBi vs 15 dBi seems like a lot to me.
Then you call the cm 4251 much smaller? I guess 7 foot is ~23% smaller than an 8 foot parabolic (in surface area). Then "dangerous voltage levels" as in consumer grade equipment could be damaged.
I have a "deep fringe" all band antenna that regularly picks up UHF signals at -80 dBm (from tv fool) yet I can point it at a local station at -15 dBm with no overload let alone damage.
How is higher voltage not higher gain? How is higher gain not higher voltage?
When I build/deploy a 24+ dBi parabolic how strong of a signal can I point it at safely?
Last question I've heard that the better pre-amps use input attenuation to protect themselves wouldn't this also protect the tuner from excessive voltage?

You say not much more gain than a cm 4228 ??? 24-21 dBi vs 15 dBi seems like a lot to me.
He was referring to the lower channel gains. Even when uhf channels went up to 83, the bulk of most uhf channels for 99.9% of people were below channel 30 because broadcasters could have wider coverage using less power. Plus the higher freq parts were more expensive and installation was trickier in the old days.

But It was run at 2 degree resolution ( I do 5 for optimizer, might go to 10 deg. like you do
Yeah, anything to cut down calculation time. At 2 deg 180 points are analyzed. At 5 deg 72 points, 10 deg 36 points. I may start to use 18 deg 20 points when optimizing.
Since azimuth 90 and 270 are the only two points Im really interested in when optimizing, maybe use 180 degrees and really make the evolver zing along ???

Since azimuth 90 and 270 are the only two points Im really interested in when optimizing, maybe use 180 degrees and really make the evolver zing along ???
Yep, that seems to do the job fine. But oddly, it seems like its only about 4 times faster, I was expecting more. :(

I remember trying 180 deg (earlier in the learning curve) but I did something wrong and got obviously wrong data I think I messed with the pattern window ( F-4) and left it at some odd angle, then when I opened the optimizer widow it imported that odd angle instead of the stock 0 90 180 90 angles. And at 180 deg the "pattern" is just a line, not very helpful to an inexperienced trouble shooter. So I settled on 5 (at least it gave some pattern) and never really went back and bumped it up even though I figured out the angles.
But these models seem slow to run at any setting.