[RTTY] Flex RTTY AFSK Only

[Kok Chen]

On Feb 16, 2016, at 3:02 PM, post at lb3re.com wrote:

>  asked Flex about FSK few years ago, and it was not in their plans

"AFSK" and "FSK" can mean different things to different people.

FM AFSK has a very different meaning for example, where the "AFSK" is modulating an FM carrier.  This was the original meaning of "AFSK."

Later,  amateurs use  an audio tone pair to modulate an SSB transmitter, and this was also called AFSK  I.e., the tone pair is modulated to RF as (SSB suppressed carrier-fM) and (SSB suppressed carrier-fS) where fM is 2125 Hz and fS is 2295 Hz if the transmitter is using LSB and you are using the 2125/2295 tone pair.  This is why Mark is the higher of the two states on RF.

Even later, when I/Q modulators appeared (as in SDRs), a baseband (DC, or zero frequency centered) tone pair consists of Mark at +85 Hz and Space at -85 Hz.  I/Q modulators can tell the difference between positive and negative frequencies (depends on whether Q leads I or Q lags I in phase).  This has also been called AFSK.

Direct conversion SDRs like the SDR-1000, Flex 5000 and the KX3 still need use an offset like 2125 and 2295 Hz to avoid "images" of the Mark and Space from bleeding through.  You have to send 2125/2295 audio ("AFSK") to these transmitters.  In that sense, you need to apply the same care when you use these "SDR" rigs just as you do with superhet transmitters.

Fortunately, except for QRP folks tinkering with SoftRocks, these direct conversion "SDR" have gone the way of buffaloes.

Modern SDRs like the Flex 6xxx and HPSDR use Digital Up Convert (DUC) modulators to generate RF directly, with D/A converters running at over 100 mega samples per second.  The only thing between the D/A converter and the antenna is the PA and matching networks.

With the DUC transmitters, modulation is done in arithmetic (in an FPGA, usually) and as such, you can send baseband I/Q Mark and Space (+85 Hz and -85 Hz) to these transmitters directly, and any spurs that are caused by the output D/A converter are usually below -80 dBc.

I do not know if Flex implements FSK using I/Q into its DUC, or if they took the lazy way out and just pass an AFSK input to the SSB transmitter.  

I do know that the regular HPSDR desktop software which 99.9% of HPSDR owners use, still generates RTTY by passing AFSK (e.g., 2125/2295) through the SSB mode.  

However, those who writes their own software can use an HPSDR rig (like Hermes, Angelia, ANAN-100, etc) in the I/Q mode to generate super clean FSK signals.  The transmit PA non-linearity is the only source of distortion.

The interesting thing is that math-based software modems like 2Tone actually start out as equations which describe a baseband I/Q signal; cocoaModem did that back in 2005 or so.  Unfortunately, to modulate an SSB transmitter, that internal I/Q signal (which only exists as numbers in memory) has to be converted to a scalar audio tone pair to modulate the early-21st century transmitters. 

Interestingly, the K3 "FSK mode" is really not FSK (i.e., there is no oscillator being slewed between Mark and Space carriers) but its "FSK" signal starts out also as an abstract I/Q signal.   Now you know why it can put out cleaner RTTY signals than other superhet transmitters.

The K3 is still nowhere near as clean as a properly waveshaped baseband I/Q waveform that directly modulates a DUC transmitter (for example, write your own software for HPSDR) since the K3's I/Q signal is still going through analog mixers, etc before it reaches the antenna.  However, the K3's "FSK" mode is usually cleaner than first generating I/Q in 2Tone and cocoaModem, convert the I/Q to an audio signal, and then passing that audio signal to an SSB transmitter.

73
Chen, W7AY