AVS Forum banner

1 - 20 of 24 Posts

·
Registered
Joined
·
6,397 Posts
Discussion Starter #1 (Edited)
I have a couple of examples of vegetation affecting DTV signal strengths. The spectrum analyzer displays in this post show the difference between my antennas with the tower fully cranked up at 71' and fully cranked down at 36'. The yellow trace is tower up and the magenta trace is tower down. One image is VHF and the other three are spaced across UHF.

Attached is a picture taken from just below the UHF antennas at 69' looking west towards my local stations. (Camera was mounted on the mast and the tower was cranked up.) There is a clear shot to the 2nd edge that the signals pass over. There is vegetation on that hill. When the tower is cranked down much of the foreground vegetation blocks the view to the hill. There is as much as 20dB attenuation from the vegetation. UHF is generally affected more than VHF. Even though the signals look strong enough to receive, only 2 out of 9 stations will decode with the antennas lowered and their SNRs are very low due to severe multipath.

Getting the antennas above the vegetation is the difference between useable OTA and no reception.
 

Attachments

·
Registered
Joined
·
3,747 Posts
Even though the signals look strong enough to receive, only 2 out of 9 stations will decode with the antennas lowered and their SNRs are very low due to severe multipath.

Your SA images are further proof that a good looking trace, with what looks like enough signal above the noise floor for a decent SNR, is no guarantee that the signal will decode.

This reminds me of what Dr. Bendov said in his paper DTV Coverage and Service Prediction, Measurement and Performance Indices.

page 4

VIII. SNR AND “FIELD STRENGTH” MEASUREMENT VIA
SPECTRUM INTEGRATION

Defining the Signal as the total received power and the
Noise as AWGN (Additive White Gaussian Noise) leads to the conclusion that the SNR at the input to the receiver increases with increased multipath.

In urban and indoor situations, there may not even be a
main signal, only reflections, some of which are of equal
magnitude.

If all multipath signals are part of the signal power, then the
SNR margin may not be an indicative figure of merit of reception robustness. In any case, even accurate measurement of the
total received power may not be trivial.

The integrated signal power is not just the Desired
Signal power. It includes, Man-made, Galactic, and
thermal noises and residual transmitter generated in-band
noise. It also includes some but not necessarily all multipath
signals. For example, pairs of identical and asymmetric
echoes, one of positive amplitude and positive delay
relative to the main signal and one of negative amplitude
and negative delay relative to the main signal, will cause
only a second-order distortion of the displayed power
spectrum. They will create group delay. Thus, in a
multipath channel, a pair of such echoes would measure
high SNR when using the spectrum integration technique
whereas in reality, the true SNR would be much lower.

There may be other combinations of echoes that would
yield essentially flat spectrum display.
http://www.researchgate.net/publica...iction_measurement_and_performance_indices304
 

·
Registered
Joined
·
965 Posts
I would think that, when strong signals won't lock, multipath is going to be a factor. A test in an area where large hills are not present would be good if you could swing that. I've got big rocks nearby. They've led to lots of crazy reception solutions.
 

·
Registered
Joined
·
3,993 Posts
I still don't understand why it is that strong single multipath (bi-path?) degradagtion forms a visually distinct "V" notch in the wave form.
 

·
Registered
Joined
·
3,525 Posts
I still don't understand why it is that strong single multipath (bi-path?) degradagtion forms a visually distinct "V" notch in the wave form.
It's essentially a math function. It can be easily simulated with the ATSC transmitter in GNU Radio. Here's how it's done.



Normally, the signal from the FFT filter (actually a root raised cosine filter) would be transmitted directly. However, in this flow graph I've split the signal into two paths. One path is direct, and the other path is delayed. It's a simulation of static multipath. The Multiply Const blocks sets the level of the multipath signal, in this case, the same level or a 0 dB echo.

Since it's a digital delay, I can only delay in increments of the sample rate. The ATSC sample rate is 10.76 Msps, so each sample is 93 nanoseconds long. If the multipath signal is delayed 3 samples, then that's a multipath delay of 279 nanoseconds. Here's what the spectrum looks like.



The spacing of the notches is equal to 1/delay. If the delay is set to 11 samples, the delay would around 1 microsecond, or about 1 MHz between each notch. In this plot, it's pretty easy to see that the notches are spaced about 1 MHz.



If you're seeing notches in your spectrum, it means you have a strong short delay echo. As it turns out, those horrible looking signals shown above can still be decoded since ATSC demodulator equalizers work well with short delays.

Ron
 

·
Registered
Joined
·
6,397 Posts
Discussion Starter #6
I would think that, when strong signals won't lock, multipath is going to be a factor. A test in an area where large hills are not present would be good if you could swing that. I've got big rocks nearby. They've led to lots of crazy reception solutions.
That's the situation in this post:

http://www.avsforum.com/forum/25-hdtv-technical/1994554-comparing-signals-over-different-paths.html

with the signals in blue. Lot's of vegetation to look through but no hills. If the vegetation was eliminated the path would be LOS. His SNRs are often low and sometimes the stations will not decode.
 

·
Registered
Joined
·
133 Posts
Do you see any day vs. night problems in getting signals under such conditions? I've got a translator site that often looses signal about 9PM on VHF ch 7 but not at UHF chs. More often in summer than winter. Path is 100 miles, 2edge is about 50 miles in--some vegetation on ridge just south of the ridge the antenna are on but tower should have us high enough to clear that. No notches in SA trace, largely flat in the ch range. You just see the SNR go down, with flutters, till not enough to decode.
I've been blaming the drop out on inversion layer ducting in the valley where the transmitters are, and that inversion develops in the evening. I would think that you might get a micro-climate inversion layer where vegetation is in line of site--might not bend or duct signal significantly but might be an issue.
 

·
Registered
Joined
·
797 Posts
Some questions to ponder:

1) Is is possible that some or much of the difference in received signal power at the lower vs. the higher antenna location is due to the height of the receiving antenna due to the change in the angle of diffraction over the diffracting edge?
2) Same for atmospheric refraction?
3) Any chance you can get your hands on a signal meter that measures EVM?
 

·
Registered
Joined
·
6,397 Posts
Discussion Starter #9
Do you see any day vs. night problems in getting signals under such conditions? I've got a translator site that often looses signal about 9PM on VHF ch 7 but not at UHF chs. More often in summer than winter. Path is 100 miles, 2edge is about 50 miles in--some vegetation on ridge just south of the ridge the antenna are on but tower should have us high enough to clear that. No notches in SA trace, largely flat in the ch range. You just see the SNR go down, with flutters, till not enough to decode.
I've been blaming the drop out on inversion layer ducting in the valley where the transmitters are, and that inversion develops in the evening. I would think that you might get a micro-climate inversion layer where vegetation is in line of site--might not bend or duct signal significantly but might be an issue.
I see all sorts of ducting related issues here but I took these readings during periods of what I call "nominal" conditions. That means typical signal strengths I see over the long term with no inversions to enhance or degrade the signals.
 

·
Registered
Joined
·
6,397 Posts
Discussion Starter #10 (Edited)
Some questions to ponder:

1) Is is possible that some or much of the difference in received signal power at the lower vs. the higher antenna location is due to the height of the receiving antenna due to the change in the angle of diffraction over the diffracting edge?
2) Same for atmospheric refraction?
3) Any chance you can get your hands on a signal meter that measures EVM?

1) I considered this and there is likely a reduction in signals due to the increased diffraction angle. At 70' the angle is 0.9 degrees and at 36' the angle is 1.7 degrees. TV Fool which doesn't take vegetation into account gives these predictions for all the stations:

70' -10.3 to +5.0 dB
36' -16.3 to +0.3 dB

Some of the up to 20 dB signal reduction is likely due to this.

2) The distance is only 1/2 mile so I kind of doubt it.

3) Probably not. What is EVM going to tell me that SNR doesn't?

Signals on this path can go down with atmospheric conditions a lot (like 20 dB at extreme times) but they rarely go up by more than a few dB.

Although it is hard for me to determine this for sure, I believe I have two different sources of multipath on my local stations. One source is from strong reflections off of the mountain to the east of me and exactly opposite from the stations. These reflections are LOS no matter what the elevation of the antennas is. This is why my antennas must have a good F/B ratio to reject these reflections.

Tests with higher F/B antennas doesn't increase the F/B past a certain point which is why I think I also have some multipath on the direct path caused by vegetation. The only thing that increases the SNR is to get the antennas up higher or wait for enhanced conditions.
 

·
Registered
Joined
·
965 Posts
That's the situation in this post:

http://www.avsforum.com/forum/25-hdtv-technical/1994554-comparing-signals-over-different-paths.html

with the signals in blue. Lot's of vegetation to look through but no hills. If the vegetation was eliminated the path would be LOS. His SNRs are often low and sometimes the stations will not decode.
Thanks for the link. I asked about testing away from hills because I was under the impression that multipath occasionally actually helps reception. Is there a way to identify useful multipath with the spectrum analyzer?
 

·
Registered
Joined
·
3,747 Posts
I still don't understand why it is that strong single multipath (bi-path?) degradagtion forms a visually distinct "V" notch in the wave form.
Like this?




The spectrum analyzer trace is amplitude VS frequency, but it doesn't take into consideration the time differences between the direct signal and the reflection/reflections. More expensive test equipment (like a Rohde and Schwarz ETL) is needed to show the time differences of the echoes, and even then it is necessary that the signal is first decoded. Catch-22.

Well, this weekend I have in my possession a Rohde and Schwarz ETL, with all the options, that I can use until Monday when I have to ship it back. I've got it hooked to the antenna on my deck at the moment, and since it has a much higher resolution than my Sencore, I do think I better understand what is going on now.

I can compare WSB-39 and WDSI-40 and the difference becomes evident. I've attached a photo.

Unfortunately, it seems that the ETL won't give me an echo plot until it decodes a signal, which won't decode in this case due to multipath which would be displayed by the echo plot. I can see, however, that my Atlanta signals have weak echoes at about 35-40µs, so it is not a stretch for me to believe the locals have strong echoes at that range as well.

I also have two Mobile DTV phones until Monday. Got WPXA at my desk with both of them and will try for the rest of the Atlanta M/H signals on my deck tomorrow and over the weekend.

- Trip
 

Attachments

·
Registered
Joined
·
6,397 Posts
Discussion Starter #13
Thanks for the link. I asked about testing away from hills because I was under the impression that multipath occasionally actually helps reception. Is there a way to identify useful multipath with the spectrum analyzer?

I'm not aware of multipath ever helping reception. If you mean can signals ever be received off of a reflection then that's sometimes true. My experience with reflections is that the SNR is low because there's still multipath.
 

·
Registered
Joined
·
3,747 Posts
I was under the impression that multipath occasionally actually helps reception. Is there a way to identify useful multipath with the spectrum analyzer?
Not that I know of. A possible exception that sometimes works, is to watch the trace for the best presentation of the pilot signal.

Calaveras has already shown us that you can have a good looking trace, with a flat scan and what looks like a good SNR, for a signal that doesn't decode. The analyzer trace implied a good SNR, but the tuner indicated a bad SNR.
Even though the signals look strong enough to receive, only 2 out of 9 stations will decode.....
So, what is needed is a way to monitor signal quality, as shown by SNR and uncorrected errors at the tuner. The FEC (Forward Error Correction) is able to correct errors up to a certain limit. Once that limit is exceeded, the signal reaches the "Digital Cliff" causing picture freeze, pixilation, and finally dropout.

Many things can cause an increase in errors (Bit Error Rate/Ratio). Some will show on the scan, and some will not.

Some factors that reduce signal quality and cause a higher BER are:

1. Improper signal level: A weak signal will cause a poor SNR. A signal that is too strong can overload a tuner or preamp, causing spurious signals that raise the noise floor and wipe out the weakest signals. A nearby FM transmitter can also cause overload, which would require an FM trap. These will usually show on the analyzer trace.

2. Multipath reflections, static and dynamic. As Calaveras has demonstrated, they don't always show on the analyzer trace.

3. Power line and impulse noise in the reception area. This will show on the analyzer trace. We find this on VHF, especially VHF-Low.

Many times I have found that the antenna aim for the strongest signal, isn't the best aim for signal quality. I first learned about this with an Apex DT502 which has two signal bars, one for signal strength and one for signal quality (the inverse of BER). I first adjusted the aim for max signal strength. Then I rotated the antenna slightly, and noticed a big increase in signal quality, with only a slight loss of signal strength. This seemed to indicate a multipath problem, and a way to deal with it.



Later, I used the Diagnostics Screen on my Sony TV that gives signal strength, Errors, and SNR. I have also used Hauppauge 850 and 950 USB tuners that give SNR and errors, but the software isn't user friendly for me. People that use Linux seem to do better with it.



http://www.hauppauge.com/site/support/support_digital_signal_monitor.html

It is possible to buy a signal level meter that will give readings of signal quality, like BER, SNR, MER, and even EVM and echo profile, but I would rather have the tuner tell me what signal quality is necessary since it makes the final judgement anyway.

http://www.solidsignal.com/pview.asp?p=digiair-pro-atsc
http://www.emitor.se/prod_digiair_pro_atsc.htm
http://www.emitor.se/index_htm_files/Digiair_Pro_ATSC_2013_2_ENG.pdf

http://www.sencore.com/product/handheld-qam8vsbdocsis-2-signal-meter#.VV4cLTjbJLM

http://www.sencoredirect.com/collections/cable-broadcast-rf/products/1ghz-qam-8vsb-asi-usb-probe
https://www.village-island.com/assets/sites/2/DTU-236-Manual-V1.0-Form7677.pdf
http://www.dektec.com/Products/Usb/DTU-236/Downloads/DTU-236A%20with%20RFXpert%20Manual.pdf

Multipath Echo Profile:
(see Attachment 3)
This is a plot of the tap energy versus time delay from the selected channel. The channel is shown as a spike at time t=0, and is represented as 100%. If reflections (echos) of the original signal are present, it will show as another spike at some positive or negative time delay and reduced percentage from the selected channel.
This graph can assist in determining whether signal problems are being caused by multipath problems, or not. The center tap (0μs) corresponds to the main signal and any other echo is interference and noise.
Digital Modulation Bar Graphs:
(See Attachment 4)
These bar graphs display the MER, Margin, and EVM for 8VSB and QAM channels. Pass (green)/ fail (red) / warn (yellow) color coded thresholds provide for quick diagnosis. The level thresholds are user definable. This section will be grayed out when measuring an NTSC channel.

MER (Modulation Error Ratio) is the digital equivalent of an analog signal to noise ratio displayed in dB. MER is a ratio of signal power to the power of the impairments within the signal. The larger the MER value, the better the signal quality.

Margin represents how far the MER value is from the threshold of visibility (TOV) / "digital cliff". Margin equals (MER – 15.2 dB) for 8VSB.

EVM (error vector magnitude) is a % RMS value that represents the amplitude ratio of the RMS error vector amplitude to the largest symbol amplitude. This is basically how closely the modulated digital "symbols" are received compared to the theoretical ideal. EVM is calculated from the MER value. The lower the value, the better the signal quality.
 

Attachments

·
Registered
Joined
·
6,397 Posts
Discussion Starter #15 (Edited)
Like this?




The spectrum analyzer trace is amplitude VS frequency, but it doesn't take into consideration the time differences between the direct signal and the reflection/reflections. More expensive test equipment (like a Rohde and Schwarz ETL) is needed to show the time differences of the echoes, and even then it is necessary that the signal is first decoded. Catch-22.

That's just plain old multipath, albeit severe. I remember this discussion about the caged UHF antenna. In order to have the signal look that bad when pointing at the station, the reflections had to be very bad. He probably didn't have LOS to the transmitter but the reflection sources were LOS. This is a common problem in cities for the receive antenna to be blocked from LOS by buildings. Other buildings are LOS to the transmitter and to the receive antenna thus causing severe multipath. In the second SA image the signal didn't get any stronger, it's just missing the reflection component. The cage worked in this situation but it's not really a viable solution for most people.

I've attached a couple images showing a similar situation here. The first image is channel 40 which is a 2 edge path. The yellow trace is the signal pointing at the station. The magenta trace is pointing 180 degrees. Six miles east of me is a range of mountains 1500' higher and LOS to channel 40. TV Fool predicts a Noise Margin of -7.5 dB at my location and a Noise Margin of +48.3 dB at the mountain location. The reflections are only about 6 dB weaker on average than my direct path signal. I'm dependent on the 91XG pattern to be able to receive the station. It looks as though there's almost no F/B ratio on the antennas because the reflections are strong compared to the direct path.

The second image is the same test for channels 38 and 39 both located at the same site. This time though there is little difference between my path to the stations and the mountains at 180 degrees .9 miles away and 200' higher. TV Fool predicts the Noise Margin at the reflected site to be just 5 dB stronger. In the image the signals off the back are fairly flat and about 28 dB weaker showing what the real F/B ratio of the antennas is.
 

Attachments

·
Registered
Joined
·
6,397 Posts
Discussion Starter #16
If you're seeing notches in your spectrum, it means you have a strong short delay echo. As it turns out, those horrible looking signals shown above can still be decoded since ATSC demodulator equalizers work well with short delays.
That's an interesting simulation but I suspect that seeing a display like that in the real world would be very rare. The delayed signal has to be identical amplitude to see it and only one signal.

Can the simulation be run with the delayed signal 10 dB weaker?
 

·
Registered
Joined
·
3,525 Posts
That's an interesting simulation but I suspect that seeing a display like that in the real world would be very rare. The delayed signal has to be identical amplitude to see it and only one signal.

Can the simulation be run with the delayed signal 10 dB weaker?
Sure, here's what it looks like with the delayed signal at -10 dB. The notches just aren't as deep.



Just to make sure practice meets theory, here's the same signal on the spectrum analyzer.



While I was at it, I took a couple more captures. Here's a 10 microsecond delay at -10 dB.





And here's a 20 microsecond delay at 0 dB.





At longer delays, it becomes very difficult to see multipath on a spectrum analyzer. You can definitely have a signal that looks perfect but doesn't decode.

Ron
 
  • Like
Reactions: rabbit73

·
Registered
Joined
·
3,747 Posts
At longer delays, it becomes very difficult to see multipath on a spectrum analyzer. You can definitely have a signal that looks perfect but doesn't decode.
Thank you.
 

·
Registered
Joined
·
6,397 Posts
Discussion Starter #19
Sure, here's what it looks like with the delayed signal at -10 dB. The notches just aren't as deep.

Thanks Ron! I've got one station that looks a little bit like that but the notches are not perfectly spaced. Not sure if it's an example in the real world or not.
 

·
Registered
Joined
·
17,346 Posts
At longer delays, it becomes very difficult to see multipath on a spectrum analyzer. You can definitely have a signal that looks perfect but doesn't decode.
That's basically the problem I had when I lived in Chattanooga. I was getting a boatload of reflections off the mountains way off in the distance while the top of the ridge I lived on blocked the main signal path.

- Trip
 
1 - 20 of 24 Posts
Top