Comparison of off-axis gains for multisatellite Ku reception

[pendragon]

There is a very interesting thread running in this forum concerning linuxman's project to capture multiple satellites:

I got the bright idea the other day to make an attempt at getting as much of the Ku arc viewable from North America as possible using fixed Ku dishes and multi-lnbfs.

I've also been contemplating this for awhile to reduce the competition of our family for the motorized dishes in our farm. With all the choices available, my interest lies in analytically determining a feasible design for a given set of birds/transponders before I pick up the tools in the shop. Rather than hijacking the original thread, I thought it better to start a new one summarizing what I have measured and learned. This is a work in progress as I've found a lot of myths, snake oil and much less hard science than I would like. I will be happy to correct any of my errors that are pointed out.

My first step was to consider what could be done with ordinary parabolic dishes. Parabolic dishes show very little loss (< 1 dB) for a feed off-axis by a few degrees. After that the drop-off accelerates. I didn't find a lot of useful data, but afcsat.com had some info I was able to spot check against a few other scattered points across the web. The agreement was also close for a few measurements I made for a rather unrelated project I am working on for my 3m. I've done a crude data fit and calculated the off-axis gains for several dish sizes in the Ku band. To correlate the relative performance, I've compared several vendors' specs and direct measurements on my dishes. The differences were mostly on the order of 0.5 dB and shouldn't have much effect on the bottom line.

My examples were:

1. 1.8m Fortec prime-focus
2. 1.2m GeoSatPro offset (also compared a Fortec 1.2m)
3. 1m Winegard DS-3100 offset
4. 90cm GeoSatPro (also compared a Fortec 90cm)
5. 76cm Winegard DS-2077 offset

The off-axis gains are plotted on the attachment as five of the six lines. The on-axis gain is that stated by the vendor, and checked for 1,2 and 5 as I have those dishes. The off-axis loss was calculated from my fit to the paucity of data available. The curves are only rough, but I think they are representative out to about 10-15 degrees. Realize that a parabolic antenna can only point focus on-axis. Any off-axis source will exhibit coma, and this will increase as one goes further off-axis. After that the defocussing will become messier and less subject to analysis; I wouldn't be surprised if these curves are on the fairly optimistic side.

In summary the fit predicts the gain will be 3dB down at about 8 degrees off-axis, and 10dB at about 15 degrees. That means even a 1.8m dish would be only be the equivalent of a 76cm dish (on-axis) at around 13 degrees. I didn't find this encouraging because the bulk of the FTA channels we watch are tending towards high SRs, high FECs and DVB-S2 over time. I would like to have enough margin for that.

Toroidal dishes represent an alternative geometry that offers the possibility of a broader arc of possible reception. After reading Iceberg's extended review thread of the Wave Frontier T-90, I located one of these for a reasonable price. I was a bit dubious of WF's claim of 40 degree arcs, so I decided to run a series of detailed tests to find out how it actually performs.

According to WF, the gain of a T-90 is comparable to a 90cm parabolic. This is possible because it has considerably more surface area, but I decided to check it. Trying to measure this within fractions of a dB seemed a bit onerous, but after thinking about it I decided to hold as many variables constant as I could. We're also DN subscribers and I already had several 76cm pointed and peaked at 61.5, 110, 119 and 129. The problem with comparing the T-90 is that as it has two reflective surfaces, circularly polarized signals are flipped compared to a single reflector. I could have used linear for the comparison, but CP signals average better and I wouldn't have to move the 76cm dishes.

My plan was to peak the T-90 on a DN bird, measure several transponders on my spectrum analyzer and then move the LNBF and cable over to a 76cm dish. To factor out any frequency dependent gains/losses, I wanted a LNBF I could quickly convert from normal CP polarity to reverse CP polarity. So I popped a DP dual LNBF apart and proved to my satisfaction that rotating the dielectric plate could give results accurate to 0.5 dB or better as long as I used the same output.

After running the tests, the T-90 measured about 0.7dB gain over the 76cm dishes. That's not amazing, but WF's spec is only 1dB more than Winegard's, so given my measurement errors, I'll give them the benefit of the doubt.

As it got dark I then ran the LNBF from 0-22.5 degrees offset in steps of 2.5 on the T-90. At each point I skewed the dish to the predicted center and then peaked the elevation and azimuth on the same bird. It would have been nice to repeat the measurements several times and at more points, but I only wanted numbers good enough for design purposes and I also couldn't hold up dinner. So the data I plotted on the graph for the T-90 may be a little ragged. Nevertheless a real-life gain profile is not going to be a smooth curve.

Eyeballing the results indicates the T-90 does not exhibit the rapid fall-off of a parabolic. In fact at 12 degrees off-axis and beyond, it beats a 1.8m. Still at 20 degrees it is 3dB down and well below a 76cm on-axis in terms of performance. Although WF doesn't directly say this, one can imply it is from their spec of 40 degree arcs.

One positive that didn't immediately occur to me but was borne out as I moved a linear LNBF across an arc of a tuned T-90, was the angular beamwidth is tighter than a 90cm dish in-track. I saw very little spillover from adjacent 2 degree spaced satellites on the spectrum analyzer. Probably this is because the in-track dish dimension is nearly 1.1m. The cross-track beamwidth probably isn't as good, but who cares? This bodes well for nailing a series of Ku birds at 2 degree separations. I am testing modified Superdish 105 and 121 feeds to accomplish this without having to grind down the LNBFs, but that is another project.

While I had hoped for a little better performance on the toroid, my measurements do validate the specs. I'll probably restrict its use to around 30 degrees of arc. I am also considering using an extra 1.2m I have lying around, or possibly picking up another 1.8m, but the useful arc for me would only be around 20 degrees.

I hope these data will prove useful to others looking at multi-satellite reception.

 

[Wescopc]

Wow! Very useful, thanks for your work. I have a T-90 with 8 LNBs on it and like the convenience of it. My experience with GeoSatpro 90cm is that they drop off more rapidly then your graph would indicate. I'll have to set up a test - you have perked my interest.
Thanks,
Bob

 

[pendragon]

My experience with GeoSatpro 90cm is that they drop off more rapidly then your graph would indicate. I'll have to set up a test - you have perked my interest.

That would great if you could run that test; I'd be happy to incorporate your results on the plot(s).

My gut feeling is the same as yours, but there is very little in the way of hard data. What I found is for larger dishes and expensive commercial feeds. The plots for parabolic dishes could be nearly impossible to attain using standard FTA components. The T-90 plot is real, however.

 

[linuxman]

Very interesting and valuable data!

My preliminary findings on my 1.8M are following your findings. Hopefully I will know more later today.

Thanks for the hard work!

 

[B.J.]

A couple years ago, I did an experiment with a 90CM Fortec, which I later partially repeated with a 10' orbitron. What I saw convinced me that any off axis feed should suffer a lot more than 1db of loss, although I don't have the equipment to evaluate that electronically.
What I did was to put little mirrors on the dish surface, and aim the dish at the sun, to observe the pattern of little spots from the various points on the dish. What I observed, was that when you move the dish off aim by even very small amounts, that the spots from the various points on the dish focused all over the place. As an example, look at:

which shows the spots from 4 of 5 mirrors on a fake lnbf, when off aim by only a very small amount (actually this was the best pattern I could get with the factory position of the LNBF holder, which wasn't quite at the focal point).
But what really impressed me, was watching these spots dance around in all directions as I moved the dish very slightly. Move the aim one direction, and each spot would go off in a different direction. After seeing this, I'm amazed that people have the success that they do with multiple lnbfs, unless the Fortec dish is much worse than other dishes.
More pictures from the experiment are at the end of this page: FC90-sg2100.html

I did a similar experiment with a 10' orbitron, and it showed even worse results, however it was difficult to stick the little mirrors onto the mesh in a way to have any confidence that they were parallel to the actual surface. I had a hard time getting more than 2 mirrors to focus within 10" of the feedhorn.

Anyway, if anyone has a dish they aren't using, try this mirror experiment. It will show you what a compromise it is when you start moving off axis.

 

[pedrogarcia]

Pendragon excellent data on a really interesting subject, great job. Many years ago (late 60s) for the military I was developing off axis satellite dishes and quickly realised the advantage of parabolic and off set dishes. Prime focus were dismissed almost immediately due to the loss of useable surface area when using an off axis LNB, a 3 metre dish became equivalent to a 1.8 metre. From there I developed a multi focal offset 90cm dish with 3 tx/rx lnbs which looks unbelievably similar to the current Dish units of today. Should have patented it 30 years ago!

 

[Cadsulfide]

Great Data!Thanks!

I'm going to re-evaluate my array, as I should be able to get a lot more birds for the buck!

 

[Zorak]

Thanks for the very interesting and useful info. I have some overdue dish work I hope to finish soon and will try to add to the pool of data.

 

[pendragon]

My experience with GeoSatpro 90cm is that they drop off more rapidly then your graph would indicate.

I didn't get a lot of dish time today between the Met Opera and the symphony. However I snuck up on the roof and waved a LNBF around my GeoSatPro 1.2m looking for hot spots offset from three known dish positions. I only had time for a few DN birds and had to use my squawker for the measurements as I was lying on my back overhanging the roof. However I had calibrated the squawker against my spectrum analyzer recently, so while the results are rough, they should be in the ballpark. I then popped the normal LNBF out and measured the peak level for each position with the same LNBF I had used for the offset measurements.

The attached plot shows the earlier results for the T-90 and the curve fit for large parabolic dishes with commercial multi-satellite feeds, plus today's measurements on the 1.2m. As wescopc and I suspected, the commercial results appear optimistic for our purposes. It may be that shallow offset dishes are not ideal, but I'll have to ponder the optics analogue some other time. I would like to try this on my 1.8m and 3m prime focus dishes, though that will be more time consuming.

If more accurate measurements bear out these results, the case for a toroid looks a little better. For example a 1.8m dish with a similar fall-off would do worse than the T-90 after about 10 degrees of offset.

 

[pendragon]

A few more thoughts:

Because a parabolic dish exhibits coma with off-axis signals, the shape of the focal cloud becomes less obvious, and trying to find the 'hottest' spot with a circularly symmetric feed is more complex and possibly non-intuitive.

As a diversion, OMI: Technical Articles: About Coma has some interesting illustrations and an explanation of coma; after all a Newtonian telescope uses a parabolic mirror just like a prime-focus dish. The twist to keep in mind for an offset dish is we are using an off-center chunk of a parabolic reflector. When the article talks about annular rings, we have to imagine the effect of generating an energy distribution with a series of truncated rings. That will make the pictures shown a little more evil.

I have spent most of my investigation simply looking at gain. The effects of the aberrations on cross-polarization performance is also likely to be a substantial factor in many FTA environments. Locking difficult signals at larger offsets may not be tea with the queen.

 

[pedrogarcia]

In Europe the torroid dishes had beeen around for some time and in fact they were a scrapped 80s design due to inaccuracies. The understanding then was there was little loss from the primary reflective surface area, and they would be cheaper than motor dishes( they never were)
Whilst it is accepted it can attain up to 48 degrees of arc spectrum where located well within all the sat footprints. Signal results beyond + or - 12degrees have been discredited on a number of trials as it was found the horizontal focal point which increased with the offset to be wide enough to receive spillage from adjacent sats.
I have to say the amount of RF loss through scatter and aberration of reflection on 2 surfaces must take its toll.
Of course all this does not mean a thing considering the Dish signal but on some of the Ku FTA it does.
I believe motor with all of its faults is still the bettersolution.

 

[B.J.]

Because a parabolic dish exhibits coma with off-axis signals, the shape of the focal cloud becomes less obvious, and trying to find the 'hottest' spot with a circularly symmetric feed is more complex and possibly non-intuitive.
.....

When I did my mirror experiments, "non-intuitive" would be an understatement. When moving off center, the spots from different parts of the dish moved in different directions. Even with small movements, it looked like bugs flying around a light on a summer night. I was really surprised.

I have spent most of my investigation simply looking at gain. The effects of the aberrations on cross-polarization performance is also likely to be a substantial factor in many FTA environments. ...

I was about to say that looking at gain alone can be VERY misleading, because no matter where you move your lnbf, you will see contributions from virtually ALL satellites within range, and there will be hotspots that will have major contributions from other sats. I would think that quality or S/N based on the locked signal would be a better measure, but even there, you might not really find the best spot for the satellite you're looking for, but the point where the interferrence is least.

However my real reason for responding to this is that I was interested in the "cross-polarization" comment. I'm not positive of what you were commenting on here, but over the years I have seen posts by several people, who contend that different polarities reflect differently off different parts of a dish, and since an offset dish is selectively using just one quadrant of a dish, I am curious what people on this forum have for opinions on this topic.

Personally, I never believed that the polarities would prefer different sections of the dish, however I have changed my mind to some extent, or at least I am not as convinced as before. I have 2 3' dishes, which spend most of their time aimed at AMC-21, which is fairly low on my western horizon. One of these dishes is motorized, and the other is fixed. Ie the difference between them is that one is facing the sat standing straight up, while the other is more or less on it's side. Both dishes are aimed perfectly enough that doing the pull down lift up thing in all directions only results in the signal getting worse. However the motorized dish is very poor on horizontal transponders while getting the vertical transponders fairly strong. The fixed dish, with rotated feed gets both polarities equally well. I've tried rotating the feed on the motorized dish, and again the signal only got worse.
Anyway, I've been wondering about what could possibly cause this. One issue, is that the motorized dish is looking through some tree limbs, and it's possible that the tree limbs are selectively attenuating the horizontal polarity. I thought that the other possibility might be that maybe the posts I've seen claiming that different parts of the dish might be best for one polarity might actually be true.
However a third possibility occurred to me, and that was that I KNOW that the feed on my motorized dish is not exactly at the focal point, but is off by about 2", and the best focus it gives is illustrated by the picture I posted above of the 4 sun reflections, all of which would have missed the feed throat. I found a better focal point 2" away, but there was so little difference in "gain" at that point, that I didn't attempt to bend the LNBF arm to move the feed. However NOW, I'm wondering if by in effect using an off axis "hot spot" as you say, I might actually be selectively attenuating one of the 2 polarities.
Anyway, I'm just curious what other people's opinions are relative to polarity with respect to different quadrants of the paraboloid, and/or off axis reception.

 

[pendragon]

In Europe the torroid dishes had beeen around for some time and in fact they were a scrapped 80s design due to inaccuracies. The understanding then was there was little loss from the primary reflective surface area, and they would be cheaper than motor dishes( they never were)
Whilst it is accepted it can attain up to 48 degrees of arc spectrum where located well within all the sat footprints. Signal results beyond + or - 12degrees have been discredited on a number of trials as it was found the horizontal focal point which increased with the offset to be wide enough to receive spillage from adjacent sats.
I have to say the amount of RF loss through scatter and aberration of reflection on 2 surfaces must take its toll.
Of course all this does not mean a thing considering the Dish signal but on some of the Ku FTA it does.
I believe motor with all of its faults is still the bettersolution.

I agree that a properly aligned motorized dish is generally the right technical solution and will deliver the best signal. It is not always the right sociological solution for six people. I started with a 3m (C & Ku) on a HH motor, added a 1.2m (Ku) on a HH motor and finally a 1.8m (C) on a HH motor. This is on top of the four fixed 76cm dishes for DN I threw up to deal with rain fade and their crummy 129 bird. It's not that we watch an excessive amount of television, but fights and domino effects often arise over who gets which dish when, and the headaches of non-blocking switching of a lot of LNBs to a lot of receivers.

There are other impracticalities I'm sure I'm not alone in facing. We have plenty of land and the city isn't terrible about restrictions. However I have to pull a permit to put a pole in the ground, for which I had no problem on the 3m. Strangely they will let me put up any number of any size antennas on any existing structure so long none exceeds 5' over the top of the roof-line peak. No permit required. I guess my strategies are pretty clear.

I have an extra 1.2m lying around and was planning to put it up with another HH motor on a new pole. But after working out the typical usage scenarios that play out here, and the fact that I am running out of roof locations with HH visibility, I began wondering about the technical possibilities of multi-satellite dishes. All are compromises, but as long as they work well enough there will be less load on the three HH dishes that can then be employed for the more difficult reception situations.

The T-90 is a partial solution for me. It cost roughly the same as my 1.2m plus a motor. It is horizontally wide enough that I get clean separation of adjacent birds pretty much out to the end of its LNBF mounting limits. My 1.2m is only marginally better. Sufficient CNR restricts me to about 15 degrees of offset. I had hoped a toroid might free up pole space with the 76cm dishes, but the DN birds are in locations with fairly sparse FTA Ku needs. There's plenty of signal on the toroid at the ends of arcs to catch them, but I would be giving up the rain fade margin I have with the 76cm dishes.

At the moment it looks like I can take down two of the 76cm dishes and use the extra 1.2m to catch a 10-15 degree arc, including two DN outliers. The T-90 could then receive 30 degrees of prime Ku FTA real estate. For the trouble I will largely solve the dish contention issues, and have the same number of poles as I do today.

 

[pendragon]

I was about to say that looking at gain alone can be VERY misleading, because no matter where you move your lnbf, you will see contributions from virtually ALL satellites within range, and there will be hotspots that will have major contributions from other sats. I would think that quality or S/N based on the locked signal would be a better measure, but even there, you might not really find the best spot for the satellite you're looking for, but the point where the interferrence is least.

Signal quality and 'lockability' have a lot of meaning to a specific receiver. Unfortunately these aren't objective measurements that can be translated to someone else's environment. Witness all the bizarrely different SQs people regularly call out on the forums. The numbers mean a lot to the authors, and possibly anyone with the same receiver, but my eyes simply glaze over whenever I run across them. Gain, beamwidth and cross-polarization are more to my liking, because they help gut estimate CNRs for a given reception requirement.

However my real reason for responding to this is that I was interested in the "cross-polarization" comment. I'm not positive of what you were commenting on here, but over the years I have seen posts by several people, who contend that different polarities reflect differently off different parts of a dish, and since an offset dish is selectively using just one quadrant of a dish, I am curious what people on this forum have for opinions on this topic.

Cross-polarization is a distortion that causes one polarization to appear in form as another. For example some of the energy of a H polarized signal may appear as V. If this is happening, so may the converse. This will not only lower the level of the desired signal, but has the potential of introducing interference in the form of the opposite polarization. Even a perfectly transmitted signal has to run a gauntlet of trouble, starting with the atmosphere, moisture, reflections off the dish, feeds and LNBs, each of which introduces more cross-polarization. There is no doubt off-axis reception deteriorates cross-polarization. I haven't looked at how a well designed toroid does in this respect, because I didn't find any readily available specs. I would like to measure this sometime, but it isn't going to be particularly easy. Unfortunately I have more compelling projects with better ROI.

Personally, I never believed that the polarities would prefer different sections of the dish, however I have changed my mind to some extent, or at least I am not as convinced as before. I have 2 3' dishes, which spend most of their time aimed at AMC-21, which is fairly low on my western horizon. One of these dishes is motorized, and the other is fixed. Ie the difference between them is that one is facing the sat standing straight up, while the other is more or less on it's side. Both dishes are aimed perfectly enough that doing the pull down lift up thing in all directions only results in the signal getting worse. However the motorized dish is very poor on horizontal transponders while getting the vertical transponders fairly strong. The fixed dish, with rotated feed gets both polarities equally well. I've tried rotating the feed on the motorized dish, and again the signal only got worse.
Anyway, I've been wondering about what could possibly cause this. One issue, is that the motorized dish is looking through some tree limbs, and it's possible that the tree limbs are selectively attenuating the horizontal polarity. I thought that the other possibility might be that maybe the posts I've seen claiming that different parts of the dish might be best for one polarity might actually be true.
However a third possibility occurred to me, and that was that I KNOW that the feed on my motorized dish is not exactly at the focal point, but is off by about 2", and the best focus it gives is illustrated by the picture I posted above of the 4 sun reflections, all of which would have missed the feed throat. I found a better focal point 2" away, but there was so little difference in "gain" at that point, that I didn't attempt to bend the LNBF arm to move the feed. However NOW, I'm wondering if by in effect using an off axis "hot spot" as you say, I might actually be selectively attenuating one of the 2 polarities.
Anyway, I'm just curious what other people's opinions are relative to polarity with respect to different quadrants of the paraboloid, and/or off axis reception.

All things being equal, I would expect an offset dish to experience more aberrations than an equivalent aperture prime-focus, but this isn't based on a theoretical analysis. Off-axis and non-focal point positioning of the LNBF will also increase these effects, and may cause the issues you see between the polarities. However any object in the signal path, like tree limbs will also cause cross-polarization, diffraction and other hideous effects. I have seen multipath from trees not even in the signal path cause minor signal deterioration, and this was on my deep 3m dish. A chainsaw quickly proved this assertion Warped dishes may also attenuate one polarity over another.

I find your mirror experiments interesting, but I'm not sure how accurately they can be placed. Even a tiny angle with respect to the dish surface is going to translate to a major defocussing. I had thought one time of gluing foil onto a smallish dish, but the prospect of peeling it off didn't have a lot of appeal. I read on the net that someone plated a pizza dish, but unfortunately their main line of research had more to do with burning holes in objects than examining aberrations.

 

[Anole]

let there be light:

I had thought one time of gluing foil onto a smallish dish, but the prospect of peeling it off didn't have a lot of appeal.
I read on the net that someone plated a pizza dish....

A generous application of a Clear Gloss should do the job, without destroying any existing color nor logos.
After the experiments with light, a layer of Matte Clear should restore the dish to satellite use.
And in the process, provide a fine protective layer against decades of harsh environment. -

Or for a lower tech idea, what about a slathering of oil or grease? -
That comes off with some soap 'n water. - :up

 

[pendragon]

A generous application of a Clear Gloss should do the job, without destroying any existing color nor logos.
After the experiments with light, a layer of Matte Clear should restore the dish to satellite use.
And in the process, provide a fine protective layer against decades of harsh environment. -

Or for a lower tech idea, what about a slathering of oil or grease? -
That comes off with some soap 'n water. - :up

There is also reflective spray paint. One still has to track a moving target and worry about burning holes in the focusing plane. It could also be a little tricky to meaningfully measure the energy distribution of the focal cloud.

In the end we have to consider that a LNBF is part of the equation, so I'll probably stick to RF. It would be a fun project for someone with time on their hands.

 

[pendragon]

While I was contemplating putting up my 1.2m offset to catch a 10-15 degree swath of the Ku arc, I realized there would be little additional effort to substitute a 1.8m prime focus, and of course that would yield another 3dB or so of gain. It then occurred to me that a prime focus dish might exhibit a slower off-axis gain drop-off than an offset dish because of the geometry. So it was another tryst on the roof vs. paperwork. Today was nice so I decided on the fresh air.

This was a comparative test between two dishes on motors I already had up, so I ran between a 1.2m GeoSatPro offset and a 1.8m Fortec prime focus using the same LNBF and cables. All LNBF positioning was by hand as I only wanted to establish which performed better. This hurts the prime focus a little because my arm had to reach in front of that dish, and this is not an issue for the offset dish. A couple of spot checks on the 1.2m offset proved similar to what I measured over the weekend. I therefore decided it would be sufficient to make relative measurements and do the data reduction later.

I established three points for the 1.8m prime focus at 5, 10 and 19 degrees off-center. Compensating for the dish gain difference, I added the prime focus results to my earlier plot (attached). This allows one to compare the different dish types directly, given equivalent on-axis gain.

The 5 degree result is a bit iffy because the C-Band feed and the support arms made it difficult to adjust the Ku LNBF for the best focal position and pointing. Nevertheless it seems consistent with the other two offset points. I wanted to test my hypothesis that the prime-focus would suffer less off-axis, and at least this test showed it would be a better choice, all other matters being equal. My 1.8m prime focus still falls a little short when compared to large commercial dishes with custom multisatellite feeds, but I'm satisfied.