The Official AVS Antenna and Related Hardware Topic! - Page 498

I would like to try an experiment like that, but I didn't do those tests to reduce urban multipath by putting an antenna in a shielded enclosure. Credit goes to Bill Naivar at Georgia Tech in Atlanta for his novel solution.


Yes, exactly

The shielding that he used probably did improve the SNR, but more importantly it reduced the errors from multipath so that the FEC capacity was not exceeded. That's why I think that monitoring BER is better than signal strength when aiming an antenna.

http://www.avsforum.com/avs-vb/showt...3#post21754233

I read through your post. I think you'd be better off aiming for highest SNR when the strongest signal and highest SNR don't correspond. The reason I say this is because (from what I've read) that the corrected errors go to zero above an SNR of 20 dB but the SNR can keep going up. To put it another way, zero errors doesn't necessarily equal the highest signal quality.

I've run BER tests in the lab on BPSK through 256QAM signals and it only takes a signal increase of 1 dB to go from a non-zero BER to a BER of zero. Of course that non-zero BER might be in the range of 10E-6.

Sounds to me that when you moved your 4221 and saw a large increase in signal quality, that was because you were able to place a major reflection in an antenna null. I don't have any information on how common that situation is. All I can say is that is never the situation here since I have hundreds of reflections coming from a 60 degree swath of terrain east of me. It wouldn't surprise me to find out if most people have to deal with multiple reflections.

Chuck

Yes, I agree that SNR is an important figure of merit for an antenna. What I had in mind with my previous post is when there is a multipath problem that challenges the FEC, then BER becomes equally important.

I had a discussion on the TV Fool Discussion Thread about antenna gain, saying:

http://www.avsforum.com/avs-vb/showt...1#post15698511 post #393

Nitewatchman answered me saying:

http://www.avsforum.com/avs-vb/showt...6#post15698876 post #394

An antenna with more gain does indeed increase the signal strength, but so does a preamp.

The difference between the two is that the preamp increases the signal strength and the noise plus adding some noise of its own, but the antenna with more gain increases the signal strength without adding noise resulting in an improved SNR.

The post by Andy Lee about Visual explanation of Noise Margin is interesting:
http://www.avsforum.com/avs-vb/showt...9#post15700679 #397


You do have an unusual situation that requires a very high F to B for your antenna.

Dr Bendov and I believe in Reciprocity...but I understand your
reluctance to blindly accept anything without PROOF....

If you pursue the "wrong" impedance coax, it probably should be the SAME LENGTH
as the "correct" impedance coax to replicate the multipath bouncing back and forth.
I'm not so sure you're going to find coax with the requisite impedance....

You might compare to an antenna with a KNOWN BAD SWR, such as the way too
small U-TUBE 4-Bay Bowtie and see if there is "Excessive Noise Figure", which
would be fairly difficult to PROVE in an Apples vs Pommes de Terre (Potatoes)
antenna comparison: http://imageevent.com/holl_ands/multibay/4bay/utube

Alternatively, you could use a 1:1 Coax Balun instead of the usual 4:1 Balun
[could be Coax] with a known GOOD SWR antenna and see if the Noise Figure
degrades when a 300-ohm antenna is terminated in 75-ohms (4.0 SWR).
This would be a better Apples vs Candy Apples comparison....
But you'll need to use very weak stations to see NF degradation,
and you should conduct the test for multiple channels and perhaps also
perhaps multiple antennas to verify repeatability....

holl_ands:

Thanks for the tips about what would be suitable test.

Yes, I plan to use weak signals and also an attenuator just before a splitter that feeds two tuners and a signal level meter to bring the signal to the cliff and be able to monitor signal strength, SNR, and errors.

This will enable me to see where dropout happens (in dBmV and dBm) with the two antenna systems.

When the NF is greater, then dropout should happen sooner at a higher signal level because the noise floor is now higher. Or, to put it another way, the system that has the largest margin-to-dropout is the better system.

Four pieces of coax in parallel might do it.


Thanks for understanding; I've been burned too many times.

When trying to align a high gain antenna, it's difficult to find the narrow "sweet spot",
esp. if you rely on a friend to shout out directions....and/or you're poking around the attic
looking for that elusive spot that has barely a sniff of a signal to begin with.....

So I find Laptop PC with an ATSC USB Stick Tuner to be an essential test tool....
Or a CECB Box with a small luggable TV.....at the end of a very long extension cord....

I've found that (FREE) TSReader Lite will display Bit Error Rate statistics when SNR
is well below 15 dB ATSC Threshhold for reception, widening the "search area":
http://www.tsreader.com/tsreader/index.html

NOTE: TSReader only works with a small number of ATSC PCI Cards & USB Sticks:
http://www.tsreader.com/tsreader/hardware.html
Be sure to skip past the DVB-S (Satellite) and DBV-T (Euro TV) sections until
you reach the Terrestrial (ATSC) section....Unfortunately, it appears that the
Hauppauge WinTV HVR-950 and HVR-950Q are the ONLY compatible USB Sticks.

(FREE) VLC Player is used to actually display the video derived from the decoded
MPEG2 data steam. Search to find one of the many download websites.

Not so bad, as those two models are one of the better USB tuner sticks with a 6th gen tuner.
Before I gave mine away as a gift, (I had to test it to make sure it worked OK, ya know) it was very comparable in sensitivity to my Zenith CECB box.

This is very good.

I put together a spreadsheet that I use to calculate System Noise Figure. It's just a fancy way to do a 2nd stage noise figure calculation but it's easy to change the numbers around to optimize your system.

http://images.aa6g.org/System_NF.xls

The graph is helpful because you can determine at-a-glance if your preamp has too much gain or not enough. If the preamp gain is too low then the line will be straight with little or no flattening at the bottom. If the line is flat at the bottom then the preamp has too much gain and you'll increase the possibility of overload. The default numbers entered are the parameters for my existing system at channel 51.

The table let's you experimentally determine noise margin using fixed attenuators or a step attenuator at the TV when you have a preamp installed at the antenna. The accuracy of this is dependent on the accuracy of the numbers you entered.

Chuck

Chuck:

Thanks for the information and link about your System Noise Figure spreadsheet. That looks interesting and useful.

Have you seen the first post in the DHC Signal Amplifiers (Amps, Preamps, Distro Amps) thread? It lists:

1. tczernec's Loss Calculator Spreadsheet
2. majortom's Cascaded Noise Figure Spreadsheet
3. Preamplifier Comparison Chart by holl_ands

http://www.digitalhome.ca/forum/showthread.php?t=42426

When I described my discussion with Nitewatchman in post #14915, you could see that I was aware of the importance of SNR when evaluating an antenna system back in Jan '09. That reminded me of something that I had read on the DTVForum.info, which is the Australian version of AVS:

The consenus in Australia seems to be that a "certified antenna installer" needs to have a signal meter that reads signal strength, BER, and MER (similar to SNR) to do a proper installaton.

My step attenuator is one of my favorite tools for making measurements, as you can tell from the link in my signature.

Looking at your animations you posted, I noticed there are ATSC pilot signals at the beginning of the 6 MHz spectrum. The rest of the spectrum is suppose to be noise-like due to the pseudorandom modulation. The pilot is the unmodulated carrier at a reduced value.

Is it possible to just measure the RF power of the pilot and try to maximize it by moving the antenna to different locations? Max power would indicate minimum multipath interference. What do you think?

It's easy to measure the pilot with the spectrum analyzer and certainly you can point the antenna by measuring the pilot. In most cases I think you'll find the highest SNR occurs in the peak direction. Occasionally there are exceptions to this.

Chuck

holl_ands & 300ohm:

Thanks for the tips on some useful tools to make my measurements more accurate and consistent.

Got that. I have an Apex DT502 which isn't a very good CECB in general (it locks up on ION stations, you can't add a channel after scan, and ota.dt.man's box died on him), but it does have dual signal bars that give consistent signal quality and strength readings. I can see the reduction in quality as I approach the cliff and dropout happens about the same time as with my SONY KDL22L5000.

I have resisted the urge to add a computer to my measurement system because I don't like them very much and only tolerate them for what they allow me to learn about DTV.

You got my attention with that one, because I don't have any equipment that will do that. All of my tuners stop giving me information when they can't decode. My SLM, however, will still continue give me signal strength readings below dropout if I add a preamp to it.

They look similar to mine except I added the graph and table which I think are helpful.


IMO measuring BER is overkill. Uncorrected BER is inversely proportional to SNR and almost every TV will give some version of SNR even if it's only a 0-100% Signal Quality bar.


I used to hunt for these at the electronic flea market in the south Bay Area. I've got one for 75 ohms and two for 50 ohms and a bunch of fixed attenuators and dummy loads.

Chuck

I'm finding this discussion very interesting, but I'm lost on two abbreviations. What are BER and MER? I've never heard of them before.

Also, Holl_lands wrote:
"NOTE: TSReader only works with a small number of ATSC PCI Cards & USB Sticks:
http://www.tsreader.com/tsreader/hardware.html
... Unfortunately, it appears that the Hauppauge WinTV HVR-950 and HVR-950Q are the ONLY compatible USB Sticks."

It also works well with the Fusion HDTV USB stick.

Larry
SF

In ATSC, MPEG2 data packets or "messages" are transmitted in 186-byte
data groups, each with a 20-byte Reed-Solomon Parity Check Code. Hence:

DETECTED MER (Message Error Rate) refers to how often these 1488-bit
packets failed the R-S Parity Check.

CORRECTED MER refers to how often packets failed R-S Parity Check but
the R-S Correction Algorithm THINKS it fixed all of the detected bit errors.
[Reed-Solomon typically corrects up to 20-bytes out of each 186-byte "message".]

UNCORRECTED MER can NOT be determined except when a known test data
pattern is transmitted. These are the errors that a viewer actually "sees" (or hears).

ATSC Pass/Fail Criteria has been established as 3 x 10^-6 BER (Bit Error Rate),
which is an average of one bit error in 0.33 Million Bits....or ABOUT one
sound and/or video "glitch" in 222 messages....which is about the point that
ATSC reception become "minimally acceptable" (based on early engineering tests).

BER = Bit Error Rate

When a digitally signal is demodulated you expect to recover the same number that was transmitted. Noise added to or interference to the signal can cause the recovered number to be different from what was transmitted. That's an error and ATSC is designed to be able to fix a certain number of those. In my experience BER is the number of errors per second.

MER = Modulation Error Ratio

This is very similar to SNR but is tailored to a digitally modulated signal. SNR and MER are often used interchangeably even though there are subtle differences. There's somewhat more of an explanation here:

http://broadcastengineering.com/RF/b..._transmission/

Chuck

I've never heard of MER meaning that. A Google search on "message error ratio" 8VSB returns no results.

Chuck

That's because the manufacturers don't disclose the algorithm used to determine
their displayed "Signal Quality".....Detected MER is the primary parameter,
because it is readily available from the R-S Decoder...and does the job....

Reed-Solomon Codes (like most Error Detection and Correction Codes) will
provide a Decode Status for each Message, whether a) NO DETECTED ERRORS,
b) DETECTED xx-Symbols in Error and TRIED to FIX them or c) CAN'T FIX ERRORS.
Corrected MER is derived from b). Detected MER is derived from c).
http://en.wikipedia.org/wiki/Reed&#8...ror_correction
http://hscc.cs.nthu.edu.tw/~sheujp/lecture_note/rs.pdf

We seem to be talking across each other. My comment refers to the meaning of MER that everyone uses, not how manufacturers determine Signal Quality. Message Error Ratio for MER is used by nobody while Modulation Error Ratio is used by everybody. Larry was asking for the meaning of MER in the context of the current discussion.

Chuck

It would be very difficult to construct a meaningful "Signal Quality" from R-S performance. The difference between 0 errors and totally unwatchable is only about 2 dB in signal to noise ratio. Here's a post I made a few years ago with some measurements made with a precision step attenuator and a demodulator evaluation board.

**********************************************************

I've found a brand new evaluation board for the ATI/AMD T316 8VSB and QAM demodulator chip. It come with a Windows User Interface that interacts with the board over USB. Here's what it looks like on my PC:



The UI controls the tuner and show if there is MPEG-2 frame lock. Separate windows show the demodulator status and AGC status. I'm also running a DVB-ASI capture program (Enensys DiviCatch) and VLC to decode the MPEG-2 Transport Stream from the eval card.

I took some measurements on my strongest signal, KNTV-DT on RF channel 12. Here's the results:
KNTV-DT delivers a rock crushing signal to my lab in Milpitas, CA. It's at least 43 dB more than necessary for perfect reception. Also note that the difference between a watchable signal (at 49 dB of attenuation) and no signal (at 51 dB of attenuation) is only 2 dB.

Here's the KNTV-DT 8VSB constellation diagram with 10 dB of attenuation:



The eval board works better than my old Samsung SIR-T165 receiver. I had actually thought KCNS-DT was off the air, but in reality, it's just too weak for the Samsung. Here's the constellation for KCNS-DT on RF channel 39 when it was pretty impaired on the eval card (and non-existent on the Samsung):



Ron

My understanding of BER is the same as yours, but BER can mean Bit Error Rate or Bit Error Ratio. The terms are often used interchangeably, but that is not proper. Bit error rate is, as you said, the number of errors in a given period of time. Bit error ratio is the number of bit errors in a given total number of bits, and is usually expressed in scientific notation as in ATSC A74:

As you can see, they are using a number that expresses bit error ratio, but they are calling it bit error rate.

My understanding of MER is also the same as yours and agrees with wiki:
http://en.wikipedia.org/wiki/Modulation_error_ratio

Thanks for the interesting link from Broadcast Engineering. There is a section in that paper that expands on your answer to retiredengineer about measuring the pilot:

dr1394:

Thanks for your interesting post with the VLC Media Player screen shot and about using an attenuator at the cliff.
With my equipment I'm not able to monitor pre RS errors, only post, but with my 1 dB per step attenuator I also notice that it takes about 2 dB to go from nice picture to terrible. If we didn't have the FEC, we would notice the gradual loss of signal quality, but with the FEC we see a sudden loss of quality when it reaches its error limit.

My OTA signals vary in strength because there are trees in the path, so it is a bit difficult to manage the attenuation at the cliff in a smooth manner, making it necessary to repeat the test a few times for consistent results. I am, however, able to set up in my car at a remote area for LOS signals that come across the water with less variation in strength (except when a boat goes by!).

I have dreams about making an in-line attenuator before the step attenuator that would keep the signal a little more constant perhaps using an AGC circuit that drives a lamp in front of a light sensitive resistor that is in the shunt resistor of a 75 ohm T attenuator.

An 8VSB signal generator is not in my price class.

Here's the original thread from 2009, with pictures of the evaluation board.

http://www.avsforum.com/avs-vb/showthread.php?t=1122271

The setup was antenna (and USB control) -> T316 evaluation board -> DVB-ASI -> Enensys DVB-ASI capture device -> USB to VLC player. I was playing the KNTV stream with VLC at the time, but when you take a screenshot of VLC, the video gets blacked out.

Ron

???

From the Video Lan documentation:

Then use a picture resizer to reduce the picture so it will fit on these bulletin boards.

Actually that's not clear to me at all. When I was performing these measurements in the lab we'd just give the number as a kind of shorthand. The "/sec" part was always left out so I can't tell from above what they really mean. I just assumed what was meant was 3x10E-6/sec.

I also see a problem using R-S errors to determine SNR. If at 20 dB SNR (or whatever the number is) the errors go to zero, how do you calculate higher SNRs? It must be done another way.

Chuck

Just so I have this correct in my head once and for all.......

From your table above, the non-corrected R-S errors are essentially zero at an SNR of 20 dB, meaning the error correction system isn't doing anything. Between 20 dB and 15.2 dB the errors rise but the error correction system can handle those errors and produce what appears to be an error-free picture and sound. At 15.2 dB and lower SNRs there are too many errors to be corrected and the system collapses, or falls off the digital cliff as we like to call it. So the whole purpose of the error correction system is to extend the useable SNR from 20 dB to 15.2 dB.

Please correct me if I have anything wrong in my summary.

Chuck

I use the VLC "snapshot" function all the time. When I said "screenshot", I meant just hitting the PrtSc key to capture the entire desktop. I was trying to show all the demodulator windows in one picture. The other open windows (like VLC and the Enensys capture device) were just incidental. Also, that desktop is Windows XP in "Classic" theme.

Ron

You're close, but there's one more thing to know about. There are two levels of error correction. The first level is the Viterbi convolutional code, and the second level is the Reed-Solomon code. The demodulator application is not showing the amount of errors being corrected by the Viterbi code, only the RS code. Bit errors are starting to creep in an some higher S/N, but it's not being shown.

So your synopsis is correct, except that it's happening at two levels. First the Viterbi code extends the system S/N and then the RS code extends it even further.

In general, the Viterbi code does well with single bit errors, while the RS code is better for error bursts. At the same time, the bits going into the RS decoder are interleaved. That is, the bits that make up one 207 byte packet were not sent sequentially, but interleaved in time. This helps when there are noise bursts on the channel. Instead of the noise burst ending up in one or two packets, it's spread out into many packets.

Ron

http://www.linuxtv.org/wiki/index.ph...our_DVB_device for anyone interested in Linux related DVB and hardware

FYI: EVM (Error Vector Magnitude, a percentage) and MER (Magnitude Error Ratio, in dB)
definitions can be found on pages A-5 and A-6:
http://128.238.9.201/~kurt/manuals/m...00A%20User.pdf

I've only seen EVM used in some (early) technical white papers, such as Tektronix's
"Signal to Noise Relationships for 8-VSB" and Dr. Bendov's re VSWR degradation:
http://e-sites2.tek.com/cmswpt/tidow...&ci=2238&lc=EN

Sorry if I jumped to confusing MER....should have used R-S SER (Segment Error Rate),
although I try to reserve the use of SER for Symbol Error Rate. I also avoid using
Block Error Rate to avoid confusion with BER....