Bad Vibes ... Illustrations of a RTTY Problem
                          By Dick Stevens, N1RCT 2/18/99 / revised 2/19/99

      Your TNC or sound card AFSK/LSB audio source can create problems if it contains distortions products. I have used RITTY and a SB16 in these examples. They are not any more prone to problems than any other equipment; I use them as the levels can be set precisely rather than the screwdriver adjustment typically found on a TNC. Some equipment may not have the physical range necessary to produce over-drive, a nice thing, but a computer sound card is capable of very high gain due to it's DSP nature. All TU suppliers and program authors emphasize the importance of setting the drive level correctly. The general principles shown apply to all "audio input / afsk" modes such as Clover and PSK31.

      Here is what the soundcard looks like when RITTY is off .. just some very weak electronic background noise that is 40 dB down or 1/10,000 the power of the zero dB level.

    This is the output of a "CL Blaster 16 Basic" inputted to a " CL Blaster Ensoniq" on another computer, without going through a radio. It was monitored with Spectrogram , a fine freeware program for analyzing audio signals, which is what your RTTY AFSK tones are. There is more noise at the lower audio frequencies. The numbers on the side are the loudness decrease in dB. A decrease of 30 dB is equivalent to a 1000 time decrease in actual power. Along the bottom is the frequency from 0 to 5400 Hertz. All these plots have a frequency resolution of 21.5 Hz and are the average of 128 scans, all intended to show the "big picture". The program can be used to focus in a narrow part of the pass band with high resolution, once you know what you want to look at closer.

    Next, here is what it looks like when RITTY is turned on, using a steady mark tone (no diddle or letters) so the display is less confusing. The center frequency has been set very low, 850 Hz (Mark=754 actual)) as this is sometimes used by stations who want to use CW filters on lower side band (LSB). This aggravates a problem as will be seen.

I need to digress here and explain a peculiarity of RITTY .. even though you can apparently select 170 or 200 Hz shift, it actually puts out 183 Hz in each case as K6STI had determined that was actually best from DSP theory. Also, K6STI measured a 1% discrepancy in his personal soundcard output frequency versus a frequency standard in the early days. He then applied this correction to his output equation for all sound cards. I think his frequency standard was actually off. The net result is that when the RITTY Menu is set to 850 CF and 200 shift, which should give 750 Mark and 950 Space, I measure as 754 Mark and 937 Space with three different sound cards. Luckily, I don't own a frequency standard.

   Here is what the soundcard is putting out when it is over-driven ... 35 dB on the RITTY menu (that's 100 times more power than 15 dB). It actually still sounds "OK" to the ear, just at a higher pitch than it should be (which you would not know on the air) and perhaps a faint echo and a "ratty" sound. My ear seems to pick up the higher frequency sounds .. I don't seem to hear the main low pitched sound now .. perhaps this makes me a little tone-deaf or at least musically challenged.

This picture is the "Smoking Gun" .. Clearly, the original clean 750 Hz signal has several new big signals added .. a strong but simple one at each ODD numbered harmonic, which decreases about 12 dB per occurance, and in between are the weak Even numbered harmonics, which are multi-peaked and more complex as harmonic number increases. Each letter and letter combo will give a somewhat different appearing trace here. The EVEN also do not decrease in strength as rapidly as the odd harmonics but are quite low to start with. The EVEN harmonics will not usually be heard unless they are broadcast on a quiet band with a powerful transmitter (such as a contest station on 10 meters). The frequency of the basic harmonics goes like this: 1st Harmonic is the Main signal; 2nd Harmonic is 2X the Main Freq, etc (and that's the end of the mathematics.)

     We now take a look at what happens when these audio signals are sent to your rig for transmission. Running AFSK tones means you also have the radio in the LSB mode, usually used for transmitting the human voice, which has tones from 200 to perhaps 3000 Hz. The radio (and amp) attempts to faithfully transmit the "voice" it is hearing. What it does can be monitored with a second rig and the audio is now taken from the 2nd radio (IC-756) that is actually listening on-the-air to the first rig instead of at the sound card output of the 1st rig (IC 706/2). Here is what the empty 17M band sounds like to the receiving radio (with the bandpass controls wide-open) before there is any transmitting:

This is a quiet band with a faint buzz (17M). The noise level is down 40 dB. The noise is really constant over the entire range; the changes are caused by the LSB bandpass filter in your radio which limits the sound to app 200-2800 Hz in this case. The bandpass can be narrowed with the front panel Bandpass controls or a special internal SSB filter.

Next, this is how the good clean signal of 15 dB power input to the transmitting rig looks; the letter "M" is being transmitted over and over. Each letter and combination of letters have different mark and space components so that when the harmonics are added, they have different peaks and valleys. "M" (3 Spaces, 4.42 Marks) was one of the worse in my checks. Can you calculate what letter combo will give equal power to Mark and Space? Hint: it's not RYRYRYRY and it depends on the length of the stop bit.

Most listeners would perceive this as a very clean signal from my IC 706/2.

        The main frequency is fine and there will be no problem copying the transmission. The problem is that another station which is a 1000 - 2000 Hz above and below the main signal will be severely QRMed by the harmonics and intermodulation products. Note that a station that moves in close and uses CW filters or a narrow passband LSB may be able to exist alongside the main signal. People that are a little further away are in big trouble as the signal is in the range that they must not filter as it would take out any caller's signal also. The only choice is to QSY or move in very close.

       An intriguing part of this display is that the original signal can actually be decoded at several other places .. the major two are an apparent "wide shift" signal and the other as a narrow shift but reversed tones. Here is the same data as above but shown in a scrolling format with Spectrogram for ease in seeing the relationships between the harmonics. You should be able to match this picture one-to-one with the above picture in the "scope" format. Actually, what we have done is switch from the "Frequency Domain" to the "Time Domain". In this picture, the bright yellow is fairly strong (red would be super strong) and darker blues are weaker. There are several faint inter-mod products visible that result from all the mark and space harmonic combinations. The Rings of Saturn also have a fine-structure like this (although from a different cause).



    Coupled with the spectrogram several above (that shows all the harmonics at the original sound card level), you can deduce many combinations of signals that could be decoded by selecting the proper shift and polarity. The most unusual to me is the combination of the 4th Harmonic Space and the 5th harmonic Mark, giving narrow shift and inverted tones. I actually observed people vainly trying to work this pair (although I'm not certain it was this exact pair) during the WPX99 contest and then sending "UR UPSIDE DOWN" to the CQer when he didn't reply. He didn't hear them as he was really far off that frequency (I wasn't). Some equipment can decode on just the mark signal (it assumes quiet is a space) .. giving even more opportunities.

---

      When the center frequency is set high (and that is why the USA standard is so high at 2210 Hz), the badly modulated signal does not create such a mess on the air. Here's what an overdriven signal looks like at "High Tones", i.e. 2210 Hz CF, as it is heard at the transmitting sound card:



The sound card seems better able to generate a cleaner signal at the higher frequency but the same output power. So lets increase the power to the max of 80 dB output of the sound card!

   We have a very harmonic/intermod rich audio signal from the sound card which has gone thru the bandpass filtering
of the transmitter AND the receiver .. essentially, no harmonic by-products can be heard by the receiving station. Although this letter "M" signal may seem to a bit "wide" with NBA elbows, real world character sequences will cause destructive interference which will reduce the signal out from the main peaks.

Lessons? FSK avoids all these potential problems but your radio may not have this capability. Some other modes (such as Clover and PSK31) require AFSK and this applies to them also. If you must stay on LSB/USB, use Spectrogram or perhaps a buddy nearby on an empty band to monitor your setup (for each mode). If you must use low tones (or even the European mid-tones), better check it out. Few things are more annoying than splatter/harmonics on the band. If you use an amp, there are more opportunities for strange peaks. Very few hams exhibit bad signals on the air; if you want to help them, try this Spectrogram method so you can be precise in saying what the problem is. It also documents problems as some cannot believe they have a problem. 73 de Dick and your comments are always welcome.