[RTTY] Narrow RTTY

[Michael Keane K1MK]

At 10:08 PM 1/16/06, Bill Coleman wrote:

>In thinking of narrow shifts, you can theoretically get down to
>extremely small values. If you think of FSK as two OOK symbols spaced
>a short distance apart, there comes a point when the sidebands of the
>one symbol start to encroach in the spectrum of the opposite symbol.

For non-coherent FSK, the minimum shift that will give orthogonality 
is a shift equal to the baud rate.

Orthogonality (very loosely speaking) means that the nulls of the 
keying sidebands of the mark tone lines up with the peaks of the 
keying sidebands of the space tone so they don't interfere with each other.

For coherent FSK (which in practice this would be generated as 
coherent AFSK), the minimum  shift that will give orthogonality is 
half the baud rate.  This shift is commonly called "Minimum Shift Keying (MSK)"

Shifts that are any integral multiple of half the baud rate are also 
orthogonal; other shifts are not orthogonal.

A shift of about 0.77 times the baud rate, although it is  not 
orthogonal, gives the lowest bit error rate for a given energy per 
bit to noise spectral density  ratio (Eb/No); Eb/No is proportional 
to the signal to noise ratio. Even  though  it's a disadvantage to be 
non-orthogonal,  the  extra  shift increases the detected signal more 
than enough to compensate; the optimum shift is 0.83 dB better than 
the orthogonal minimum shift.

Wider shifts are suboptimal because the receive filter must be wider 
and therefore passes more noise.

For 45 baud RTTY, these shifts would be:

  Minimum shift keying: 23 Hz
  Optimum BER /(Eb/No): 35 Hz
  Minimum shift (non-coherent FSK):  45 Hz

If the only noise source present was Gaussian white noise, there'd be 
no reason to use more than 45 Hz shift FSK or 35 Hz shift AFSK for 45 baud.

The reasons for using a wider shift is the nature of the channel 
errors at HF; these are burst errors caused by selective fading, 
impulsive noise bursts or interference.

By transmitting a wider shift and using a receiver that employs 
separate mark and space filters with a decision circuit, it is 
possible to maintain copy even though the mark or space may be 
momentarily obliterated by a channel error.

This isn't the optimal approach for recovering maximum signal energy 
from Gaussian noise, but it's the correct choice at HF where 
non-Gaussian sources of error predominate.

73,Mike K1MK

Michael Keane K1MK
k1mk at alum.mit.edu