Multipath solutions at 900Mhz licenced

[RFburn]

Hi all
I have some licenced 900Mhz digital links that randomly fade or cut in and out, usually during the night/early morning. I'm thinking its multipath from atmospheric conditions. Sometimes the signal level will fade, other times there will be good signal but the SNR will drop.
I have reflector antennas at both ends (15dbd) all vertical polarization.
I've read about the advantages of Circular polarization, but can't find any source for them at 930-950Mhz. Not sure if that is a fix or not. I can't do diversity due to tower loading at the hub and the remote end are on water towers.

Any ideas?

Thanks!

 

[Higgler]

RFburn,
I don't think it's atmospheric caused fading. How far are the towers spaced and at what height and what's in between them. Described the antenna feed? is it rotatable? Is this a big issue or just a slight annoyance?

I suspect one of two options,
1) the temperature change makes your rf bounce add in and out of phase by changing the height of your towers relative to the ground. You don't need much height change to make the signal change.
2) moisture caused change, moisture on your antenna or ground.

If it's multipath then usually circularly polarized antennas help since the bounce is the opposite polarization (or crosspolarized). Even though you can't add antenna diversity, you probably can modify your feed horns and add polarization diversity, i.e. if you're mechancically inclined, add a small motor and rotate your feed horn. More complicated is to add a polarization convertor (back to back circular polarizers) in front of your feed horn and rotate the outer one.

I would personally try to rotate one or both of the antenna feed horns just a little. Place them part way between vertically polarized and horizontally polarized. It could be a quick solution.

kch

 

[RFburn]

Hi
Thanks for you reply

There are actually 6 links in star formation, each at various heights at hub depending on the lay of the land to the remote sites. The remote sites are mostly on water towers (about 40+m high each) since this is a municipal owned radio system.

Some link antennas are Sinclair sinclairtechnologies.com/catalog/resources/pdf/SV460-DI.pdf
and some are Katrien: kathrein-scala.com/catalog/MF-950B.pdf

Both antenna types do it. I think i'd have to rotate the whole antenna, not just the feedhorn. The antennas are currently vertical polarized.

 

[Higgler]

If the reflectors are round, and you can rotate the feed horn, technically that works. I have a feeling it's easier for you to rotate the entire antenna.

There are simple "polarization twisters" to just hang in front of your feed antennas.
These change your V pole to other linear polarization that are just a series of slanted wires spaced by foam

If you take wire or metal tape and two spacers of 3 inch thick x 12"x12" foam with some scotch tape you can make one in about an hour. The sandwich is just wire,foam,wire,foam,wire.

Let's say you use metal tape (any hardware store) cuz it sticks to the foam (or glue paper to the foam and stick metal tape to the paper). Cut metal tape strips about 0.1 inch thick and space them evenly on the paper 1 inch apart (+/- 20% tolerances).

Orient the metal tape wires that are closest to the feed horn "horizontal" (i.e cross pole to your vertically polarized antenna), have the middle metal tape layer at 22 degrees relative to horizontal and the third layer (closest to the reflector) at 45 degrees relative to horizontal (accuracy not too critical on the angular placement), this will make your polarization at 45 degrees to vertical. The polarization is 90 degrees opposite your 45 degree wire/metal tape orientation (not aligned with the wire). The metal tape removes that polarization, it doesn't create it. Locate your foam/metal tape item front of your feed and try it out.

That may be less work than rotating your antenna. You will lose 3 dB due to the polarization twist of 45 degrees unless you put another identical twister on your receiving antenna. A well made unit will have less than 1/4 dB inherent loss. I'm used to making a 6-18 GHz unit requiring 5 to 7 metal wire layers.

kch

 

[RFburn]

Hi
I could mount the entire antenna at a 45deg angle at both ends if it would help, but I'm not understanding why giving the antenna polarization a tilt would help.

I understand that circular polarization will help since the mulitpath is likely to be in the opposite direction, however I can't seem to find a source for circular polarized antennas

Could you elaborate on why the tilt to the polarizarion will help.

Thanks very much

DBH

 

[VEBill]

If you were going to change the polarization of the SV460, then you'd probably want to turn the entire antenna including the reflector. The SV460 webpage (I didn't bother downloading the pdf) mentions that polarization can be vertical or horizontal, so I assume that the instructions describe how to make this change (probably the mounting bracket at the back is probably designed to fit onto the mast either way).

The webpage also mentions that the Horizontal beamwidth is 16 degrees, while the Vertical beamwidth is 30 degrees. Although they don't define which axis is which on the webpage, the narrower beamwidth is typically aligned in the wider dimension. Therefore, to reduce ground reflections, you'd probably want the antenna installed with the tallest axis up & down.

Problem is that if your path length is long (many miles, and I assume it is...), then even the narrower 16 degree beamwidth would still include a ground-bounce path in the beam.

Crazy idea: tilt the antennas up so that you don't lose much gain compared to the bore-sight, but reduce the gain aimed at the ground. I don't really like this sort of ugly kludge, but it might be useful for experimentation (keep stats to see if it helps - proves ground reflection is the problem or not).

If the path-length is REALLY long, then you could have all sorts of atmospheric effects refracting your signal off to the next state. More power can help.

Since multipath is frequency dependant, try changing channels during a fade. Or move one antenna by roughly 0.25 wavelength (a few inches).

Also worth checking to make sure it isn't interference or desense from external sources. Sometimes it helps to actually listen to the signal and maybe you'll hear some obvious interference.

In summary: it's probably time for some experimentation to confirm in detail what the problem is.

 

[Higgler]

The reason to try the tilt is;
When a huge ground bounce occurs from a nice flat surface it creates nulls and peaks vertically in space. Hence if you were in a balloon rising upward you'd get for each polarization
V pole = peak, null, peak, null
H pole = null, peak, null, peak

hence the bounce for V that's hurting you at night can help you instead.
Lots of variables in the real life scenario.
For circular polarization, the ground bounce is cross pole, so your dips in space as the balloon rises should be zero for a perfect world.

I'm certain that a particular rotation of your antenna will help, but if you need to do alot of them to find the optimum, then the little polarization rotator may be easier, or not. depends on how tough it is to rotate your antenna.

How big (dB) are the changes at bad times?

kch

 

[RFburn]

The fades can be over 25db at times and are worse in the summer and almost non-existant in the winter. The longest link is 37kms and the shortest is 22km.
Of course with digital you can't just "listen" to the signal like go ole analogue.
These are also licenced fixed frequency (full duplex)so I can't just change the frequency at will.

It doesn't look like I can attach files here otherwise I could put up some the path profiles.

Problem with trying things is that this randomly happens usually during the night and also having to hire riggers to go make the changes up the tower.

So if I were to put a 45deg tilt at each end (in the same tilt direction) it could reduce the amount of multipath? or should I be leaving one end verticle and just tilt the other and put up with some lower signal in order to have a lower null?

There are some times when the signal levels stays constant but the SNR drops.

Trouble is also by the time i'm aware its happening, its gone before i can get there

Thanks for all your ideas so far!

 

[Higgler]

25 dB is quite large, and your beamwidth is very wide, so that rules out any large antennna pointing motion as the cause. (I know, sounds very obvious, just thinking out loud).

No fade in the winter is very strange, where are you located? It's usually wetter in the winter and water vapor changes would be a thought as a major change in the rf bounce from structures. Water is very dense.

Does this happen on all your links, or only on one of them?
Can you tell if it's a two way fade or only loss in one direction, hence my next statement if it's a one direction loss or if you can't answer that question.

If one direction fade is occuring, what if part of your hardware is intermittent with summer humidity and temp change (am and pm) and dryness in the winter keeps your intermittent connections stable. Anyone who's worked with environmental testing fears the intermittent connection with temp and or humidity tests. It's so hard to find when circuit card via connections go bad. I've had antenna wire center conductors move slightly and put 25 dB nulls in the antenna just due to a temperature change.

Here's a reasonable reference talking about link fading,

http://www.ictp.trieste.it/~radionet/2000_school/lectures/carlo/linkloss/INDEX.HTM,

there's a section on digital communications too.

Tricky problem, good luck.

kch

 

[VEBill]

"...can't just "listen" to the signal..."

Assuming that it isn't a wideband signal such as video or VERY high rate data, then listening can be as easy as connecting a Radio Shack scanner to the antenna lead and punching in the frequency. Make sure that the scanner actually covers the assigned frequency because many such V/UHF scanners skip the "800 MHz" cellular band and perhaps adjacent.

Listening helps to rule out silly causes (*) and should help to confirm that the signal is actually dropping out and being replaced with white noise during the fades. Just because your frequency is assigned, it doesn't necessarily mean that someone's wireless headphone (or whatever) hasn't drifted onto your channel.

(* Once upon a time, someone was having terrible difficulty getting their dial-up Internet connection connected. Many frustating days of endlessly 'Format C:' later, the modem speaker was finally enabled so that one could hear what was going on. "This is a recording. The number you have called has been changed..." Ah ha, maybe that's the problem.)

Random thoughts:

You can make CP with two crossed linear antennas fed with the appropriate phasing.

You can stack two or four (etc.) antennas (probably at the water tower ends) vertically using a phasing harness to increase gain and narrow the vertical angle of the pattern.


Based on your description, I suspect that temperature inversion layers and similar daily atmospheric effects are sending your signal off into interstellar space. Most likely solution is more power one way or another.

25dB fades aren't a problem if your system has 30+dB of margin to begin with (not at all unreasonable). If you're starting with 10dB of margin (for example), then it's a marginal system to start with. A lot of interesting things can happen over 37km.

How much RF cable is between the transceiver and the antenna? How much loss in the cable? Can you relocate the transceiver to be remote-mounted next to the antenna so that you gain several dB?

Is it a new system, or a system that has always had this problem? Or has it deteriorated over the years?

Is it a one-way system? Can you put a REALLY good (super low noise figure) LNA on the receive side? At 900 MHz, you should be able to achieve less than 0.5dB.

You mentioned tower loading at the hub. Are other systems on the same tower? Have you checked for desense? Even pretty good receivers are susceptible to desense from nearby transitters. Digital receivers (the point to point stuff) are often more digital than good (radios designed by computer weenies...). Might need a set of helical filters at the front end to filter out the rubbish.