# Multifeed installation and exact calculation of LNB position

A

#### a33

##### Specialised Contributor
I have some questions about multifeed installation and multifeedbrackets, wanting to understand the principles behind it/them better.

Q1.
What is the best form of a multifeed bracket in the horizontal plane: straight, curved, … ?
Of course in the vertical plane a multifeed bracket should mirror the Clarke Belt, and thus be curved. But what about the horizontal plane?

I read somewhere about parabolas as optical mirrors, that the focal line is a straight line.
However, there are many multifeed brackets that are curved towards the dish.
So this confuses me: What is the better fit?
And if the focal line is curved, what is the radius of the curve? Is it the same as the focal distance, or twice the focal distance?

For small separations of LNBs this is a minor problem, but for bigger distances the effect might be noticeable. And I would like to understand the theory behind it.

Q2.
What should theoretically be the aiming point of the LNBs?

On a straight multifeed bracket you often see them pointing straight towards the dish, on a curved multifeed bracket you often see them pointing towards the middle of the dish.
Thinking about the G-spot of an offset dish: shouldn't the LNBs also in this case split the opening angle in half, and shouldn't the LNBs point to their own horizontal G-spot?

Q3.
What is the most exact way to calculate the distance between LNBs?

I've seen 3 different ways to calculate the distance between LNBs for multifeed installations.
All are dependent on the focal distance (or distance to the G-spot for offset dishes) and on the difference in Azimuth of the two satellites:
a. distance = fG x sin(dA)
b. distance = fG x ( (dA x PI) / 180) (that is: in radials)
c. distance = fG x tan (dA)
[focal distance = fG, Azimuth difference = dA]
In this order they go from smallest to biggest number.

Till about 15 degrees difference in azimuth they have similar outcomes, but with bigger angles the differences become bigger. Which is the best formula for bigger azimuth differences?

Q4.
Extra question: Would the answers to the questions above be different for multifocus dishes, such as Visiosat (Big) BiSat or Maximum E-85?
For purely toroid dishes, I know that the focal line is curved away from the dish, and that the LNBs look away from the middle of the dish. So is a multifocus dish somewhere between a 'normal' dish and a toroid dish? What kind of focal line, where to aim your LNBs?

This has become a rather long post, but I hope to get some good answers, and to understand the principles of multifeed reception better!

Greetz,
A33

D

#### davemurgtroyd

##### Regular Member
Unless you know exactly which portion of an elliptic paraboloid the dish uses (and manufacturers use different ones as witnessed by the differing focal points) then any calculations you make are going to be mere approximations and guesstimates and empirical measurements are required.

Again a curved multifeed arm designed for a particular make of dish is going to be the optimum solution - but again with a lot of lnbs they can be moved nearer or further from the dish so as long as they are aimed correctly at the dish.

General calculations may well give you approximate positioning but adjustment of lnbs will always be required on installation in situ. #### RimaNTSS

##### Super Moderator
Staff member
I have never made multifeed calculations, however have made several multifeeds and used signal measurements to find best position for multifeeded LNB. There are couple of pictures of multifeed on my SMW-1600 dish, curve in horizontal and vertical planes are clearly visible.

A

#### a33

##### Specialised Contributor
@RimaNTSS:
That is a purely toroidal dish, I believe? Nice design.
The multifeedbracket for a toroid dish is curved to the outward, that is clear. My question is primarily about a "normal" parabolical (offset) dish, though.

General calculations may well give you approximate positioning but adjustment of lnbs will always be required on installation in situ.
@davemurgtroyd:
I tend to disagree with you, here. We know and can calculate the focal point, the focal distance, the G-spot, the G-spot-distance (= effective focal distance) of a normal parabolical dish.
Exact distance calculations are given on the internet for Wavefrontier dishes (T90, T55) and Visiosat Big Bisat dishes.
On the dutch forum sat4all.com one member calculates exact LNB-distances for any parabolical dish, using G-spot-distance and the radial-calculation (see Q3 in my first post).
Manual adjusting afterwards usually gives deterioration of signal, not better signal.
Only LNB-skew needs adjusting, sometimes, due to not-exact placing of the H and V antenna in the LNB.
So: If you follow theory and your theory is sound, then you know what data you need to calculate the exact distance.
What dish-characteristics or dish-measurements do you think we need more than G-spot-distance, then, to be able to calculate the exact distances?

My big question remains: Is the focal line (or focal plane) in principle a straight line (plane) or a curved line (plane)?
And if that line/plane is affected by "parabolical abberation": what corrections are needed when you go more and more off-axis? (As we all know: a spherical mirror gives spherical abberation for parallel beams, a parabolic mirror gives parabolical abberation for off-axis beams.)

I am adding two pictures to show the difference of multifeed positions: totally straight placing of LNBs, and totally curved.  Greetz,
A33

Last edited:
A

#### a33

##### Specialised Contributor
Well, I've done some googling myself in the meantime, and this is what I found out and what I now assume to be true:

Q1
For subjects as 'mirror' and 'focal line' or 'focal plane', google gives quite a few hits; most of which concern optical mirrors though; both spherical and parabolical.
For both is stated, that the focal line/plane is a straight line(plane), perpendicular to the main axis of the mirror.
As for instance in the attached drawing (source: Telescope Optics Tutorial ): So, assuming that what is valid for optics is also valid for satellite beams, a multifeed bracket for a satellite dish should best be straight, in the horizontal plane.

As the focal line is straight for both spherical mirrors and parabolical mirrors, this would then be the same for normal satellite dishes and for (spherical) multifeed-dishes.
(But not, of course, for purely toroid dishes. See this topic: SMW OA 1600 dish .)

Q2
I think my assumption about a 'horizontal G-spot' has much chance to be correct.
The message is: Spill-over losses should be avoided as much as possible!
So aiming at half the aperture angle, at the 'horizontal G-spot', would be best.

Q3
Looking at the drawing of the focal plane and the fixed main axis, another way of calculating LNB-distances comes to mind, using exactly those given lines. I would think a proper way to calculate LNB-distances along those lines is to always calculate LNB-distances relative to the main axis (or 'central LNB', so to say), using the tangens-formula:
distance between LNB and main axis = fG x tan (dA)
[ fG = focal distance, dA = Azimuth difference ]

The distance between two LNBs somewhere on the multifeed bracket could then be calculated, as the difference between the two calculated distances to the central LNB (in the case that they are at the same offset-side of the main axis; otherwise [at opposite offset-sides] it would be the sum of course instead of the difference).

If this way of calculating the distances is in fact correct, then the other ways of calculating I gave in my starting post are still pretty good approximations for little azimuth differences to the main axis.
The further away from the central LNB, though, these ways of calculating give too low an outcome.

For wide-offset calculations, I guess the above formula would be more appropriate and more precise.
I would love to see practical confirmation, though, that this way of calculating is actually right!

Personally, for simple calculations, I prefer to use the 'radial-calculation' (method b), because it is easy to calculate with, and can be written as distance between LNBs = 0,01745 x fG x dA

So far for assumptions and theory now.
Would love to discuss if this proves to be right in practice!

Greetz,
A33

Last edited: D

#### davemurgtroyd

##### Regular Member
Well, I've done some googling myself in the meantime, and this is what I found out and what I now assume to be true:

Q1
For subjects as 'mirror' and 'focal line' or 'focal plane', google gives quite a few hits; most of which concern optical mirrors though; both spherical and parabolical.
For both is stated, that the focal line/plane is a straight line(plane), perpendicular to the main axis of the mirror.
As for instance in the attached drawing (source: Telescope Optics Tutorial ):

View attachment 89602

So, assuming that what is valid for optics is also valid for satellite beams, a multifeed bracket for a satellite dish should best be straight, in the horizontal plane.

As the focal line is straight for both spherical mirrors and parabolical mirrors, this would then be the same for normal satellite dishes and for (spherical) multifeed-dishes.
(But not, of course, for purely toroid dishes. See this topic: SMW OA 1600 dish .)
Firstly a dish is a 3 dimensional paraboloid and its focal point will be on a straight line for any particular source and as the sources in this case are in an arc the focal points for each source (satellite) will also be on a curved plane - think of car headlights - the "main beam" filament is at the prime focal point and projects a beam straight ahead and level whereas the "dipped beam" filament is offset to the side and upwards and thus projects a beam downwards and to the side (by your logic the "dipped beam" would also be level and merely pointed to the side)..

You are only considering 2d parabolas and not 3d paraboloids.

A

#### a33

##### Specialised Contributor
@battenfan:
Thanks for the link to the pdf. I've got something to read again! Hope it is not too difficult.
Quick question though: I hope aiming point should still be somewhere around what we call g-spot in this forum? Or does this pdf give still another aiming point?

@davemurgtroyd:
Yes you're right I'm considering 2D in this topic.
In fact I wrote in my first post here (and also later) that my concern is the horizontal plane of the multifeed bracket. I wrote:
Of course in the vertical plane a multifeed bracket should mirror the Clarke Belt, and thus be curved. But what about the horizontal plane?
So I am well aware that there should be a curve in the vertical plane, as you mention. In fact I think it is strange that many multifeed brackets don't have such a curve...

My concern is that some multifeed brackets are curved towards the dish, and others are straight, perfectly perpendicular to the main axis of the dish. See photos, above.
I now think the brackets curved towards the dish are not a good fit to the focal line/plane.
I think maybe they are curved to make aiming of lnb's in a direction somewhere in the middle of the dish easier. I can think of no other reason for making a non-fitting bracket, but still think it's strange.

Greetz,
A33 #### battenfan

##### Regular Member
G-spot is below center of the dish. Max gain, according to the pdf, is (point P) above center of the dish (point C) (see figure). Which one is better for you in reality, you have to find out for yourself.
Iirc, Rima has reported improved reception aiming for the G-spot.
I have no strong opinion on aiming point.

D

#### davemurgtroyd

##### Regular Member
Aimed at a satellite on the main axis the focal point is a certain distance from the dish along that axis, with off axis satellites is that distance still the same? If so the mounting bar needs to be curved and not perpendicular to the main axis or if it were straight then the lnbs would be further from the dish.

Also do lnbs not need aiming at the centre of the dish (and not parallel to the main axis) - simple angle of incidence equals angle of reflection

Last edited:
A

#### a33

##### Specialised Contributor
G-spot is below center of the dish. Max gain, according to the pdf, is (point P) above center of the dish (point C) (see figure).
Well, I studied the artice by Marco Terada quite a bit, but I couldn't find that he concludes that point P in the picture (above the center of the dish) gives highest gain.
Point P in the picture, I believe, is introduced as a variable point to the question where to aim the feed. He writes:

“If the feed remains pointed at the apex of the parent paraboloid (i.e., psi-f = 0 degrees), negligible XPOL is generated (15). However, this leads to large spillover and associated gain loss. Therefore, in practice the feed is tilted to direct its pattern toward the reflector, resulting in the introduction of high XPOL.”
“It is worth noting that cross polarization (in linear polarized feed, A33) arises from the reflector curvature and from the tilting of the feed.” (the higher the tilt, the higher the XPOL; See figure 9 of the pdf, A33)

“In many systems, the feed pointing angle psi-f is set equal to the angle that bisects the reflector, psi-B , or to the angle pointed toward the center of the projected aperture, psi-C .”
[Point C, the center of projected aperture, would by the way also be below the center of the dish, if I project correctly, A33]

As I read this article, the recommendation to have the least spillover and not too much XPOL would be: aiming below the center of the dish; to point B (G-spot) or point C in the picture!

So as far as I'm concerned: the G-spot holds its preferred position.
And also the 'horizontal G-spots' (as I called them) for aiming multiple LNBs on multifeed brackets!

Thanks for the link to this interesting though not easy article.

Greetz,
A33

A

#### a33

##### Specialised Contributor
Well, following the questions of @davemurgtroyd, I thought maybe a discussion would follow between a few members here, about these questions.
As there is no discussion, I think I can answer these questions, as I think I lined out the answers already in post #5, above.
Aimed at a satellite on the main axis the focal point is a certain distance from the dish along that axis, with off axis satellites is that distance still the same? If so the mounting bar needs to be curved and not perpendicular to the main axis or if it were straight then the lnbs would be further from the dish.
The answer to this question and assumed implication would be NO.
Also do lnbs not need aiming at the centre of the dish (and not parallel to the main axis) - simple angle of incidence equals angle of reflection
The answer would again be NO.

Similar questions were in fact part of my questions in post #1, though I had no pre-assumption of what the right answer would be (except for the horizontal G-spot!).
In fact the implication of my findings for the way of calculating the distance between LNBs highly surprised me, as I have seen that way of calculating nowhere before on any satellite-forum! But I think my conclusions are pretty sound .

Greetz,
A33

A

#### a33

##### Specialised Contributor
Coming back to this topic:

After finding the following topic on a russian forum, I find that I am partly wrong in my conclusions about multifeed brackets: Google Translate

My assumptions about a multifeed bracket best being straight in the horizontal plane, and the aiming of an offset LNB to a "horizontal G-spot" seem to be correct.

My assumption about the calculation of LNB-distances for straight brackets seems to be highly incorrect, though.
In the above topic it is found that that LNB-distance is
a) linear to the azimuthdifference, and
b) smaller than the above mentioned formula fG x ( (dA x PI) / 180) [= 0,01745 x fG x dA] .
That is: @strannik (wanderer) finds a factor of about 0,85 for his dishes, so the formula would be, if I understand the google translation correctly:
distance between LNBs = 0,01745 x fG x dA x 0,85 for straight brackets.

However, if I interpret the topic correctly, the formula for a curved bracket would still be (about) 0,01745 x fG x dA; and a curved bracket would give only 3% loss compared to a straight bracket.

These are astonishing finds for me. I'm not sure if the value of 0,85 applies to all (normal) dishes, and if my interpretation of the difference in distance-calculation for straight and curved brackets is true.
I hope I'll discover that in future , but thought I would mention these findings here already.

greetz,
A33