The cross-pol dilemma

Ku-band VSAT field technicians have a tough job.  Not only are they expected to aim a small, wide-beamwidth antenna exactly between neighboring satellites 2 degrees away without accidentally illuminating either one, they are then expected to adjust the rotation of the feed within 1 or 2 degrees of some angle they can’t directly measure, with the objective of suppressing cross-pol power by 30 dB or 35 dB as observed on a transponder they can’t see.

A little math shows how touchy this is!  For angles close to nominal, the cross-pol discrimination (XPD) is  -20log(tan(θ)).  So 5° pol error gives XPD = 21 dB;   2° error gives 29 dB, and 1° gives 35 dB.  You can see that in the diagram: the blue curve is co-pol EIRP vs. feed angle, and the red curve is cross-pol EIRP.  The green vertical bar represents a nominal acceptable tolerance.  The noise represents a possible C/N measurement limit.

Co- and cross pol response vs. feed rotation

In our online VSAT courses, we teach students to preset the polarization with an inclinometer, which should get the feed aligned within about 2-5°, but most satellite operators need much better than that — usually 30 or 35 dB — to keep cross-pol interference under control.

Of course, the most reliable way to set cross-pol is to transmit a CW test signal and adjust the feed while someone is watching the opposite-pol transponder and telling you when you have nulled the carrier on it.  Usually that person is a technician at the NOC, watching the downlink from a larger antenna on a spectrum analyzer, and talking to you on your mobile phone. And in fact, we teach students to always do an “uplink cross-pol alignment” if at all possible.  But what to do if there is no mobile phone service where the VSAT is?  Isn’t that the point of having a VSAT – to establish communications where there isn’t any?

Let’s take a look at the options.

1. Peak the co-pol signal by rotating the feed.

This is the worst possible choice!  Consider that the repeatability of any signal level measurement is on the order of 0.2 dB, due to scintillation, psuedo-random noise-like signal spectrum, limited C/N, and other effects.  An uncertainty in level of X dB is equivalent to a rotation of  arccos[10^(-X/20) ] radians.  So a 0.2 dB co-pol uncertainly translates to 12° rotation error, which would give an XPD of only 13 dB!  In order to get XPD of 30 dB, rotation accuracy must be 1.8°, so co-pol peaking repeatability must be 0.004 dB — clearly impossible for any practical field measurement.

2. Mechanically pre-set the feed rotation with an inclinometer.

Field technicians will usually have an inclinometer in their tool box.  Even a simple hardware-store inclinometer can read surface angles with about 1°-2° accuracy, provided that the surface is fairly level in the orthogonal plane (otherwise the needle will drag).  So a simple and effective technique is to first lower the antenna elevation so the feed system is physically horizontal, place the inclinometer on the edge of the OMT waveguide or other surface that can be relied upon to be accurately parallel to or normal to the polarization plane, rotate the feed for the desired reading, then raise the elevation back to the target.  In our GVF certification training course, we teach field technicians to always use this method to pre-set their feed pol as accurately as they can, even if they will later be doing an over-the-satellite uplink alignment.  In most cases it should result in an XPD in the order of at least 20 dB, and maybe 25-30 dB if the technicial is very careful (and lucky).  But it has several limitations, including:

  • Mechanical accuracy of BUC or WG surface vs. the electrical transmit pol plane axis
  • Spacecraft’s own skew (for example, Eutelsat satellites)
  • Residual rotation as the elevation is raised, for example if the mast pipe is not plumb

3. Automatic cross-pol systems (ACP)

Some VSAT networks offer options in which the hub is equipped with a receiver to automatically measure the level of a cross-pol test signal and report the result back over the outbound TDM carrier to the specific remote VSAT, where it can be shown to the field technician in real time as he/she adjusts the feed.  This is a very elegant solution, especially if the result is displayed on a meter or other device at the antenna.  Some systems even couple this test with activation, thus preventing VSAT’s with inadequate XPD from joining the network.  The main disadvantages of ACP systems is cost and complexity at the hub.  It is not trivial to calibrate a cooperating earth station to accurately measure XPD, and upgrading a hub antenna with the required RF options and processing electronics and software for ACP can be costly.

4. Receive beacon null method

Satellites generally have a CW beacon with a fixed downlink EIRP on each polarization (V and H).  In theory, you can measure the beacon on the opposite pol and rotate the feed for a null.  In practice,this is not often a workable solution for several possible reasons:

  • A spectrum analyzer is required.  Many field techs have only a meter or simply use the VSAT modem to measure levels.
  • Beacon EIRP’s are quite low.  Even with a high-performance spectrum analyzer (narrow resolution bandwidth, good stability, and low phase noise), with a small VSAT antenna it is unlikely that you can get sufficient C/N to null the beacon to a depth of more than 30 dB.
  • Sometimes satellite beacons are not available on both polarizations.  In that case, the feed must be rotated back 90° which compromises accuracy.
  • The antenna feed transmit and receive polarization planes may not be perfectly orthogonal.  Keep in mind that all antennas actually transmit and receive elliptically polarized signals.  Only in a perfect linear cross-pol feed/combiner will the ellipses have both infinite eccentricity and exactly orthogonal major axes.

4. Receive balance and null balance methods

In certain circumstances you can take advantage of the steep symmetrical shoulders of the null in the response curve to set cross-pol with only a standard meter or VSAT modem.  In the example in the diagram, we see that the technician has measured a peak signal reading of 11 dB as the feed is rotated.  (We’ll assume the measurement is C/N, but it could equally be Eb/No, or any other related metric such as IRD units.)  The feed is rotated until the signal almost drops into the noise – about 4 dB C/N.  The technician notes the feed scale reading (93°) then continues rotating until the same signal reading is again 4 dB (125°).  Therefore the co-pol null must be at the mid-point (109°) and if the feed is rotated exactly 90° from there, it will be at the cross-pol null(19°).

Using the receive null balance method

You can see that this method could be quite accurate, but it has some drawbacks of its own, such as:

  • The clear-sky downlink signal must be very strong.  A C/N of at least 6-10 dB is needed.  This is often not the case with modern FEC schemes.
  • The feed must have a rotation scale with a resolution of 1°-2° across at least 180° of rotation.
  • The technician must be trained in the method.
  • The antenna feed transmit and receive polarization planes may not be perfectly orthogonal.

So which is the best technique?  

Even if they are trained in the importance of preventing cross-pol interference,  field techs are in a very difficult position if  the system does not have ACP and mobile phone service is not available at the site.  At a minimum, every field tech and reasonably accurately pre-set polarization with an inclinometer, but there is no universal answer.  Perhaps VSAT network and equipment design engineers will integrate better solutions in the future.  If you have a suggestion, please post a reply!

5 Responses to The cross-pol dilemma

  1. Ralph Brooker says:

    That’s exactly right! In our GVF training courses we require the student to first preset their pol, then accurately point in az and el using the beam balance method BEFORE they do a final cross-pol alignment. Ralph

  2. P Lozano says:

    May be it is useful to add that the antenna must be previously adjusted very accurately in Az and El before proceeding to the Xpol isolation adjustment. Although the Xpol adjustment has been carefully made, the value of X-pol isolation will not reach the minimum required (say 30 dB) if the adjustment of Az and El is not nearly perfect. This effect should not be ignored.

  3. high accuracy inclinometer says:

    Creative article and very useful
    information in this
    high accuracy inclinometer
    high accuracy inclinometer
    Thank you.

  4. Dave Bookham says:

    Expanding on the beacon method described above, If you are fortunate to have some wide carriers on the opposite polarity and they co-inside with the guard band between the co-pol carriers, then you can slowly rotate the feed until the cross pol goes to a minimum.You don’t need an expensive spectrum analyser for this if the wanted carrier is high power and the cross pol carrier is fairly wide-band since it can be observed on a low cost sat finder with a basic inbuilt spectrum mode. I have used this method many times and have always passed the TX cross pol test without any re-adjustment. Remember, the test carriers that you are observing don’t need to be your own, just look across the entire satellite bandwidth and you will normally find a few DVB carriers up there.

  5. andy hope says:

    If there is no comms method back to the NOC, using above methods 2 and then 4 gets a pretty good result especially if the “limits” of the rotation are carefully noted, and the point precisely half way between the marks is used – using a satmeter or the modem carefully will give the points where the signal drops to noise at each of the rotation limits. But best method is of course to carry a portable satphone when operating in remote areas – a good satphone is cheaper and easier to carry than a quality specanalyser. Ask the tech at the NOC to call you back, to save your costs – they usually will. Then the polarisation will be approved by the NOC so liability on the field tech is lessened. If using the modem to set the received quality is not possible then there are some pretty good sat aiming test sets out there, which include “signal quality” bar graphs that are pretty good – like EbNo these bar graphs are subject to scintillation so taking ones time and using the eye and brain to average out the noise is worthwhile.

Leave a Reply to high accuracy inclinometer Cancel reply

Your email address will not be published. Required fields are marked *

HTML tags are not allowed.

31,809 Spambots Blocked by Simple Comments