[RTTY] Information please

[Kok Chen]

> On May 26, 2017, at 1:20 AM, John Barber <john at bordertech.co.uk> wrote:
> 
>  With low power and poor
> propagation, PSK can work a lot better than RTTY. 

Perhaps you mean PSK31 is better than RTTY when conditions are _good_?  PSK31 fails pretty badly when there is multipath, for example. 

PSK31 also cannot take advantage that RTTY can under selective fading -- when there is selective fading, the two tones of an RTTY signal do not fade simultaneously.  You can imagine that when Mark has faded completely but Space has not faded, a good decoder can behave like a Space-only decoder and still decode the character, albeit with a greater error rate than having both tones available.  With PSK31, there is no complementary signal to decode from; selective fading impacts PSK31 as much as flat fading.  This is the "frequency diversity" that is associated with RTTY.

However, when conditions are good, a perfect PSK decoder will beat out a non-coherent FSK decoder by about 3 dB in SNR.

A 50 watt PSK transmitter will produce similar error rates as a 100 watt non-coherent FSK transmitter when all else is equal.  However, not everything is equal, and PSK31 has other advantages.  

PSK31 further beats out steam RTTY because under optimal filtering for both cases, PSK31 gives another 1.6 dB advantage over RTTY due to the slower baud rate.  The efficiency of Varicode vs Baudot helps some more.  Start and stop bits are very wasteful when it comes to synchronizing characters.

(A _coherent_ FSK signal can make up for much of the loss when compared to PSK; many will remember how nice the error rate of Amtor (SITOR-B) was when decoded by a coherent FSK demodulator.  But unless we stop [sic] using stop bits of nondeterministic length ("stop bit jitter"), we are stuck with non-coherent FSK decoders -- although a couple of us are playing with "pseudo-coherent" decoder ideas for RTTY. )

An extreme case to illustrate PSK31 under poor propagation is polar- and transequatorial- flutter.  PSK31 will never get through however much power you use, but RTTY will eventually get through with enough power.

How PSK31 fares under the CCIR Flutter condition is shown in the blue line here:

http://www.w7ay.net/site/Technical/DominoEX/Measurements/ITU/flutter.html

Notice that no matter how much SNR improves (towards the right hand side of the chart), the PSK31 error rate stays pretty much at about 50%.  I.e., you might as well toss a coin and guess what was being sent :-).  Basically, the flutter has caused so much phase error that no matter how clean the signal is compared to background noise, the error rate does not improve when you are depending on phase information to decode your data.

On the other hand, when conditions are good and all you have is Gaussian noise, PSK31 does quite well, see the blue line here

http://www.w7ay.net/site/Technical/DominoEX/Measurements/AWGN/dominoex11.html

You can compare the AWGN error rate of this blue curve to the RTTY AWGN error rate at Alex VE3NEA's web site.

Other ITU conditions than flutter is linked from this page:

http://www.w7ay.net/site/Technical/DominoEX/Measurements/itu.html

I had made up those charts some years back for DominoEX.  But each plot has a PSK31 curve too.

There is one hidden advantage RTTY has over PSK31: the power from RTTY is constant (well, almost constant in the case when you waveshape an AFSK signal).  

To keep the bandwidth tight, the PSK31 signal is envelope modulated.  The result is a 3 dB peak-to-average power envelope ratio ("crest factor").  Since many PA are peak power limited, you are forced to run lower average (RMS) power.  If the PA clips at 100 watts, you simply cannot run PSK31 at over 50 watts without severely harming others who are sharing the spectrum.  With RTTY, if the PA can handle power without melting, you can run pretty much the power where the amplifier begins to  clips. 

What mode is better is really a pretty complex question.  As you can see, there is no black-and-white answer.

As an aside, some of the worst flutter I have experienced is the path from Tahiti to California, and at one time, from Phoenix to Fernando de Noranha.  Neither are polar paths.  In the Noronha case many years ago, I heard a JA who was visiting PY0F issue a CQ on 20m PSK31.  Lots of errors, but I decided to try anyway.  However, all I got back are "AGN?" and I gave up.  A few minutes later, the op switched to RTTY and I got through in one call.

73
Chen, W7AY