First the basics of multipath
Normally the problem is substantial signals reflected from objects bringing in other signals from just off the side (off axis) or from behind the receiving antenna. The reflected signal arrives at the antenna later than the main signal with information that not in phase (at the same time) with the main signal. Analog TVs showed a ghost because the reflected image arrived later and the horizontal sweep (left to right) of the TV's displayed it after the main image. In digital it just trashes the decoding with bits that are so far out of order error correction can't compensate.
{Advanced: Since the horizontal rate for
NTSC was 15.734 KHz, and time equals the frequency divided into 1 (1 over 1 Hz = 1 Second), a horzontal sweep across the screen takes 63.5 microseconds [6.35E5]. So if the screen were 10 inches wide (not a 10 inch screen) and the ghost was 1 inch displaced to the right, then the multipath was taking 6.35 microseconds to arrive after the main beam. If you knew the exact distance to the station, you could plot points where the reflection was most likely taking place. (same principle as radar)}
So the process is to eliminate multipath since it exists in just about every situation to a degree, is to be sure the main beam is so much stronger than any reflection, the main beam overrides the multipath in the demodulation stage so it only passes the correct bits to the decoder. In analog it would be when the main signal is so strong you either can't see the ghost or it's so weak you have to squint to pick it out.
Another important point is the longer it takes for the multipath to arrive, normally the more off axis it strikes the antenna (more obtuse angle). And the later it arrives the harder it is for the digital processor to decode the signal. This is a darn good thing, as it is very difficult to reject near axis reflected waves as even antennas with the tightest beam width still pick up signals up to 25 degrees either side of main lobe with only a 3
db penalty at the edges of the beam. So it is normally ones more from the sides and rear are the worst problem, though any strong reflection is a problem.
Solutions
So the goal to reduce multipath is to make sure the main beam is stronger than the side or rear signals arriving at the antenna. Any type of log/yagi corner reflector normally will do this quite well from a short boom Radio Shack U-75R to the long boom XG-91, HD9095P,
PR-9032 style antennas. The real key to these style antenna are two things. The biggest is the rear corner reflector is known to have one of the greatest front to back ratios for yagis, thus rejecting reflections off the rear. The other way are antennas with elements that stick straight out from the boom (orthogonal or 90 degrees) which have a much higher front to side ratio than antennas with Vee type elements, as some combo antennas. Front to side is never put in commercial home TV antennas but is important and is found rated on other communications antennas. So with the back and sides for the antenna maximized for the least reception, most of the resulting gain if off the front of such antennas.
So hence in most cases a long boom UHF corner reflector antenna will reject the most multipath in typical situations. But with all antenna installations not everything is typical.
The worst antennas I have found for rejecting multipath are 2 and 4 bay whisker bowtie styles like the DB2, DB4, 4220, 4221 to which I have experience. While they reject multipath fairly well off the back, they have little rejection off the sides. Most multipath is from the sides and not the rear leaving these antennas vulnerable to side or just forward of side reflections being received. I experienced this problem at my house by replacing a 4221A with a U-75R. Even though I lost gain, my side rejection greatly improved.
However when you look to the eight bay whisker bow tie antennas, with 4 elements side by side, the antenna narrows it's forward lobe tremendously rejecting much of the side multipath received by it's smaller cousins. Also if you look at models for the 4 bay vs the 8 bay there are more significant lobes about 45 degrees between the side and rear on the 4 bay than the 8 bay. Since Front to Back ratio is measured exactly on axis, it never tells you the story of these rear lobes, where again the 8 bay shines by having much smaller ones.
In extreme multipath situations, stacking two identical antennas horizontally will narrow the beam width side to side, rejecting even more signals off axis.
=======
Other cures and sources of multipath.
Rear Reflections:
We already know most of the antennas reject well from the rear, but what happens if the signal from the rear is actually stronger than the main signal? This can happen if the main signal is blocked by something yet something behind you is big enough to reflect a great deal of signal.
I have seen the scenario about a year ago in the Orlando Local Reception Thread at AVS. It was a mystery that spurred much discussion as it goes against most multipath logic. It turned out this person did have a LOT more signal from behind him than in front. He put up a lower gain antenna and improved though didn't solve his multipath. Someone realized what might be happening and he turned the antenna away from the station and bingo! his multipath was gone.
Ground Reflection
Another more difficult multipath to fix is one where something in front of your antenna is reflective enough (conductive ground, wet grass, metal roof of a neighbor). Then there is multipath between the direct wave from the tower and one that bounces off the ground in front of the antenna.
This multipath can often be fixed by raising or lowering the antenna, changing the angle of the reflected wave to a null in the antenna pattern.
Another way to solve this type of interference is to vertically stack two identical antennas. This greatly reduces reception above or below the plane of the antenna. Similar to side by side stacking but rejecting in the vertical plane.
Less advertising throughout
Multipath - Causes and Cures
Linear Mode
