Signal to Noise Ratios - Rainbows in the Dark - Part 3

Not all S-Meters are calibrated, mine sure is not. Below you'll find a nice chart for the ideal S-Meter, a standard originally proposed by the Collins Radio Company 50+ years ago and it is now accepted as the de facto industry standard. (The original standard stated S-9 = 50uV, the rest are linear extrapolations)

S-Meter    Signal    dBm  
S-9 +60db  50,000uV  -13
S-9 +50db  15,811uV  -23
S-9 +40db   5,000uV  -33
S-9 +30db   1,581uV  -43
S-9 +20db     500uV  -53
S-9 +10db     158uV  -63
S-9            50uV  -73
S-8            25uV  -79               
S-7          12.5uV  -85     
S-6          6.25uV  -91      
S-5          3.13uV  -97      
S-4          1.56uV  -103      
S-3          0.78uV  -109      
S-2          0.39uV  -115      
S-1          0.20uV  -121      
S-0          0.10uV  -127

The source of this chart is ARRL's QST Magazine DEC2009, Page 45

There many ways to calibrate your S-Meter, but that lies outside of the scope of this blog. For our purposes here relative measurements will more than suffice.  So let say you are on 40 meters, typically the "noise" will be S3, or about -109dBm (your location and radio will vary, but work with me here). If your receiver has an MDS (minimum detectable signal) of -131dBm, it means you’re loosing 22dB of your dynamic range to the noise! (131-109=22dB). In this case, the S-meter is more-or-less giving you an absolute power DIFFERENCE between its MDS and the noise floor, in dB.

You then hear a CW (or any other kind) station calling CQ, and he has a S7 signal. Therefore S7-S3=S4 or 109-85=24db, so a +24db, that is a fairly clean and easy signal to work. Are you with me so far?

In my previous blog Signal to Noise Ratio - Rainbows in the Dark I created files of where I thought certain digital modes could be work under the noise level (SNR).  With some digital modes listening and decoding signals that are much much harder than that. In my previous example file of about a -7db signal, I'm talking about a -109db+7db=116db or one S-Unit under the noise! By now, we have all seen that WSPR can do -27db (maybe less).  In my last blog in this series, I posted that "CW@20WPM could only be done at -7db or better" (this was sure to draw fire from all camps)

I had a few folks tell me that they could do much better, one fellow whom I trust Dick /k2UFT, basically agreed with my -7db level estimates, but wisely suggested that I should re-do the files and embed a secret word with each file. In that way the testers would be blind and could then report what they heard and decoded.

So here is my new challenge, listen to the new files, preferably just once, just like a real QSO copy. Then  please email me (or blog post) the correct secret word embedded in each. Never mind about -3db and above, those are easy, right? -- just focus on -5db and below. Since so many folk were so confidant that could do better than -7db, I have taken the time to create four additional files with even more noisier levels. (HINT: all of the words are 8 char long)

SNRtest_ki4SGU_091116_1_pos9.wav
SNRtest_ki4SGU_091116_2_pos7.wav
SNRtest_ki4SGU_091116_3_pos5.wav
SNRtest_ki4SGU_091116_4_pos3.wav
SNRtest_ki4SGU_091116_5_evn0.wav
SNRtest_ki4SGU_091116_6_neg3.wav
SNRtest_ki4SGU_091116_7_neg5.wav
SNRtest_ki4SGU_091116_8_neg7.wav
SNRtest_ki4SGU_091116_9_neg9.wav
SNRtest_ki4SGU_091116_10_neg11.wav
SNRtest_ki4SGU_091116_11_neg13.wav
SNRtest_ki4SGU_091116_12_neg15.wav

I know, I know... I can hear you thinking out there, "What if you use a computer?", knock yourselves out, FLdigi away and/or any DSP filter you'd like to use, remember just keep it like a real QSO. Also there is no prize for being right or first, so why cheat?

As an update, I also worked with Eric /ab3DI on PSK31 vs WSPR last week, we were not able to validate that a two-way PSK31 QSO can take place @ -10db (so for now this remains a guess-ti-mate)

For those of you that are only on SSB -- What value does this provide you?? WSPR can provide you a baseline, granted for voice SSB the SNRs have better be positive. If you like to check what your link budget is for a given path, simply have your station and the remote station fire up WSPR and let them run for a day or two on the given band and given power. This will reveal the observed SNR between those points (hourly). Please understand, time of day and time of year, WX, and solar activity will all be variables, but at least you'll know that on that day you had that SNR, albeit it will vary over time but if your net is always at the same time of day, chances are the numbers you observed will provide you with a "reasonable" estimate to base further decisions concerning moving your net to a different band or time. Also no need to do just 2 stations at a time, for a regional net have a bunch of the guys try it all at once, the WSPR database will sort out all of the link budgets at once.

Once last idea, if you have a program like VOAprop by g4ILO (an excellent program!) you can use your WSPR data (observed data) to calibrate your other VOA predictions (predicted data), and in the end be able to make better predictions.

73 ki4SGU

PS. This blog series has been mentioned on SOLDERSMOKE PODCAST 118 (QRP & Homebrewers Inspirational Incense)

Signal to Noise Ratio - Rainbows in the Dark

In analog and digital communications, signal-to-noise ratio, often written S/N or SNR, is a measure of signal strength relative to background noise. The ratio is usually measured in decibels (dB). If the incoming signal strength in microvolts is Vs, and the noise level, also in microvolts, is Vn, then the signal-to-noise ratio, S/N, in decibels is given by the formula;

S/N = 20 log10(Vs/Vn)

If Vs = Vn, then S/N = 0. In this situation, the signal normally borders on unreadable, because the noise level severely competes with it. In digital communications, this will probably cause a reduction in data speed because of frequent errors that require the source (transmitting) computer or terminal to resend some packets of data. If Vs is less than Vn, then S/N is negative. In this type of situation, reliable communication is generally not possible unless steps are taken to increase the signal level and/or decrease the noise level at the destination (receiving) computer or terminal.

And this was the world we lived in for most of time, but recently (last 40 years) even HAMs can communicate under the noise. As you may know if you have been following my blogs here, I have been smitten by WSPR and chalk up -30db contacts. PRIMER

Here are my findings to date (with the equipment at hand)

Mode Min SNR
CW@20WPM +3db (machine decoded by MFJ-461)
CW@20WPM +1db (machine decoded by fldigi or CWget)
HFpacket (300baud) +1db
RTTY45 -5db
CW@20WPM -7db (other claim like less, like -13db*)
PSK63 -7db
FELDHell -7db
PSK31 -10db
Olivia 64/2000 -13db
Olivia 16/500 -14db
WSPR -30db (maybe less)

NOTE: * some talk of minimum detectable signal (MDS), but I mean where one can actually decode it, not just hear it.

I prepared some files with simulated noise so I could judge for myself, now you can click on the links and judge for yourselves.

(I left them as .wav files because they sounded better, and are SMALLER than the .mp3 files)

Signal to Noise Ratio Files 091115 // CW string of 'ki4SGU em73ox' // CLICK TO LISTEN

1. Postive 9
2. Postive 7
3. Postive 5
4. Postive 3
5. SNR of 0 (Zero)
6. Negative 3
7. Negative 5
8. Negative 7
9. Negative 9

So here is the deal, if you are using WSPR (and you should be by now), and you make contact with me, drop an email and lets sched up a real QSO using some digital mode CW, PSK31 or Olivia and see if my estimates are right, or just full of hot air. 73 ki4SGU