# human based digital modes

Discussion in 'General Technical Questions and Answers' started by AF7TS, Nov 21, 2018.

1. ### AF7TSHam MemberQRZ Page

In https://forums.qrz.com/index.php?th...ll-i-need-the-math.635381/page-6#post-4873154 the topic of other 'human based' digital modes came up. I figured it was worth a separate thread.

Well I presume you still want radio involved, so it won't be entirely 'all human' HIHI.

Coherent CW has already been mentioned. I think that 'hand sent ear decoded' PSK is plausible and a variation on the coherent CW theme. The _essential_ theme of 'digital' modes is quantization, in amplitude, frequency, phase, and time, so that you have a limited set of discrete symbols. The each transmitted symbol is reliably distinct from any other transmitted symbol. Actual received signals (which are _analog_) are assigned to one of the discrete symbols, ideally with very little chance of error.

For the below it is helpful to think of CW as being encoded as 1 and 0 (it is the original digital mode, after all). '1' means full amplitude for a unit of time, '0' means zero amplitude for a unit of time. A 'dit' is thus '10' (on for a dit time and off for a dit time). A 'dah' is '1110'. Add '00' after each complete letter and '0000' after each complete word.

One could transmit a constant carrier, using a straight key to selectively invert the phase. On the 'other side' you would decode the phase and on-off key an audio output. But this wouldn't work very well unless you synchronize the two sides _very_ accurately.

A better method would be similar to what PSK-31 actually does, where your 1s and 0s are sent at a constant rate tied to your 'carrier', where 1 is sent as constant phase and 0 is sent as phase inversion. The condition of 'key up' would be a regular series of phase inversions, shaped to minimize side bands to the level required to actually transmit a message. (If you look at a waterfall display, a series of 1s would look like a single line, and a series of 0s would look like perfect parallel tracks, and actually sending a message is a mix of the two.)

One can clearly hear the difference in sound between a PSK-31 string of 1's and a string of 0's (single frequency carrier versus 'pure' pair of sidebands).

One could easily send the 1's and 0's pattern of Morse code using the PSK-31 modulation scheme, rather than the normal Varicode used. Call this PSK-31M. It would work just fine for 'keyboard to keyboard' communication.

Now could a human _send_ this Morse based PSK-31M. The issue here is that now you are forced by the mode to send your dits and dahs at a constant rate tied to the mode. I think that someone good with a 'keyer' who can tie their sending accurately the the 'dit' rate generated by the keyer could do this.

Then there is the question of decoding this 'by ear', meaning be able to understand this with a simple analog receiver connected to a speaker.

We always 'shape' the envelope of the on-off keying used for ordinary CW. If you were to instantly transition from off to on (true square modulation) then the bandwidth of the CW signal would be _infinite_ because of the keying sidebands needed to get the square wave. By having finite rise and fall time for the signal we get the narrow bandwidth that we expect for CW. When CW is decoded 'by ear' we the transitions to be somewhat 'sharper' than strictly necessary, and thus a wider bandwidth than strictly necessary to carry information.

My gut says that we would need the same 'extra sidebands' to decode PSK-31M. This could be, for example, by increasing the sharpness of the modulation envelope, or perhaps by using the standard raised cosine modulation but at higher rate (eg. doubling the phase inversion rate for '0' and then doubling each bit, so you get 2 transitions for each 0 transmitted).

The other possibility would be to have a circuit that decodes the PSK and turns it into on-off keying. However this would depend on the filtering capability of the circuit, rather than the filtering capability of the human brain for the decode. You could possibly have a circuit that looks for the entire 'train track' of frequencies associated with modulated PSK, but then narrowly filters only the center frequency and sends it to the audio output.

My guess is that someone good with CW would be able to instantly pick up some version of the above, if the modulation was wide enough.

73
Jon
AF7TS

2. ### K2CAJXML SubscriberQRZ Page

I was thinking of QPSK, which sends a carrier at either 0, 90, 180 or 270 phase shift. If you tuned in to that with direct conversion, a 0/180 signal would be 0v in the quadrature output and +/- DC in the in-phase output. The converse would be true for 90/207 degrees. That would be the case, at least, if the signal were coherent with your receiver's oscillator.

Maybe there's some way to wing such a thing into an audio signal with distinguishable tones, each in a left and right headphone channel, sending two bits per symbol.

3. ### ROBERTHARRISQRZ Member

Thanks for information. Really usefull.

4. ### KL1THam MemberQRZ Page

The limits for a human ear would be distinction between tones and speed that a person can comprehend the message, 50wpm morse proves speed isn't the main issue. This would take a huge bandwidth, but what about a piano? 88 keys, even using just the white keys would give letters, numbers, and several symbols. Just "play" the message. I'm sure anyone with musical training would be able to figure out the note at a slow data rate. Interweave the different characters so that letters don't sound similar with the most common letters 4 keys or more apart and the lesser used symbols as filller. Or use 2-key symbols and only use one or two notes from each octave. Whatever the limits of the human ear allow. (Isn't this basically how JT65 works, just with a 10x bandwidth for humans?)

KI4AX likes this.

Maybe.....the only problem I see is relative speed. A string of 4 1s would sound the same as a string of 8 1s if sent at half speed. That's not impossible to overcome, but still a hurdle for the receiving end.

You are basing your entire idea of a human digital mode on a comparison with CW. I've been flamed on the Zed several times for calling CW a digital mode. You may get the same reaction. I hope not.

6. ### AF7TSHam MemberQRZ Page

True. This only works if you have a minimum number of zeros in the signal to create transitions which give a symbol rate. No zeros, no symbol rate anf no synchronization.

Thanks. I figure that the 'morse is _the_ original digital mode' crowd is pretty large.

-Jon

N2SUB likes this.

The human ear should be able to decode RTTY or a similar BFSK mode. The Morse-equivalent keying speed is roughly 54WPM (1200 / 22ms).

So people who can do QRQ in their head should be able to learn Baudot by ear using some appropriate speed-building method. I'm not sure if the 170Hz shift would be too close, but the solution to the shift is just to wipe out one of the tones using a filter, and the remaining town will just sound like QRQ CW using a different alphabet.

The alphabet need not be Baudot (which has shift characters that give each symbol two different meanings)... varicode might be a much better choice since its format is much closer to Morse (variable-length characters, fixed "spaces" between characters, common characters shorter than uncommon ones, etc.).

KI4AX likes this.
8. ### W5DXPHam MemberQRZ Page

11101110101010001010101110111000

Yep, not only digital but binary digital.

9. ### KA5IPFHam MemberQRZ Page

We had several instructors when I went thru "High Speed Radio Operators" course in '70 that could decode RTTY in their head and type it out. The one that amazed most of us was the CW guru. 60WPM CW and typing it on a Teletype machine. AND could carry on a conversation with you at the same time. He was limited to 60WPM because that's as fast as the TTY machine operated. I don't remember him ever saying just how fast he could go. It was almost like his fingers had a direct link to his ears and the brain was for talking. Remember this was all 5 letter code groups, not plain text.

KI4AX likes this.
10. ### AI3VHam MemberQRZ Page

I got a dozen crisp new us hundred dollar bills for anybody who can copy rtty (baudot) by ear.

That's like saying you know someone who can run 100 mph.

Rege

KI4AX likes this.