[RTTY] K3 FSK TX Bandwidth

[Kok Chen]

On Jan 27, 2013, at 10:53 PM, Ed Muns wrote:

> I didn't feel empirically that other stations had more trouble
> copying me due to the somewhat lower energy in the signal.

Unless Elecraft did something terribly wrong to the first 280 Hz of the filter (and unless there is a bug, that is no reason to think they would do that), then any loss of power is (1) tiny, and (2) is discarded by the demodulator anyway (certainly true if you use 2Tone).  

What is filtered away is unused power by the receiver in the first place, just as CW keyclicks are filtered away when you tune in to the CW signal.  Many of you even use 250 Hz filter to receive RTTY.  Your modems would not even know Elecraft has filtered the keyclicks away (but the stations that operate near your frequency sure will be able to tell.)

Unless the receiver uses a Matched Filter, those keying sidebands are not used when RTTY is demodulated by a Nyquist filter.  There are only two software modems I know which uses Matched Filters to gain a few tenths of a dB of SNR when propagation conditions are quiet and when there is no QRM.  None of the common software on Windows uses Matched Filters.

>From what someone else has measured, the Elecraft FSK filter profile is at least similar to the "400 Hz filter" they use as their AFSK transmit filter.  So, as long as the tone pair is properly centered, you would expect the filter to introduce no intersymbol interference (ISI).  As I indicated in my RTTY Transmit Filter write-up, a properly designed bandpass filter can be as narrow as 280 Hz at the -6 dB points and still incur no extra decoding errors at the receiving end.

I have been modeling RTTY filters through transmit IMD (intermodulation distortion).  You will notice in Andy K0SM/2's plots that there is only so much you can do to narrow down an RTTY signal before IMD takes over.  

Stare closely at his plots and what you will see is that the filtered AFSK signals start off being nice and narrow.  But when it drops down to about 40 dB or 50 dB below the carrier peaks, you see a sudden broadening of the spectrum.  

This sudden broadening comes from 3rd order transmit IMD (second order IMD does not cause broadening).  The amount it broadens is pretty much the FSK shift (170 Hz) plus perhaps one set of RTTY keying sidebands.  I.e., that "pedestal" adds about 216 Hz to the width of the transmitted signal.

With a transmitter that has better IMD that the one Andy measured, that "pedestal" will start at a lower level.  For a transmitter with worse 3rd order IMD, that pedestal will be higher up and causes more harm.

So, even if you start off with good intentions and use a 280 Hz filter, the transmitter will broaden it to 500 Hz when you are down to the -40 dB or -50 dB level relative to the carriers.

The FSK filter in the K3 will produce a similar "pedestal" since it also goes through the transmitter's IMD.

I am currently looking into methods of producing filtered RTTY that are less affected by transmitter IMD.   

73
Chen, W7AY