2009-03-06
Hellschreiber QRSS
Remember how I said the QRSS beacon was finished? Yeah well, is any project really finished?
When I originally designed the FSK modulator I envisaged adding sequential Hellschreiber to the beacon. The FSK shift line was connected to a PWM-capable pin and low-pass filtered just for this purpose. It is only now that I finally got around to implementing it.
The software is quite simple. A text file specifies the pattern. 8 shift levels are supported and an arbitrary total length, giving 8xN font support. The 8 shift levels are mapped into the bits of a byte, LSB-MSB min-max shift. A pre-distortion table of 8 bytes linearises the frequency shift which would otherwise be highly compressed near the maximum shift because of the varactor Vr/C character.
The previously collected varactor characterisation data was used to construct the pre-distortion table. To allow resolution of all 8 shift levels as unique frequency bins at remote grabbers the maximum shift level is about 8-9 Hz. A hard-coded level corresponding to a maximum shift of 5 Hz is used for the FSKCW modulation. The calibration of the shift gain pot still represents the maximum shift for Hell, but FSKCW is a limited to a little more than half the "width" of the Hell modulation to conserve QRSS spectrum.
The beacon still supports CW identification and cycles through all three modulations (CW, Hell and FSKCW) continously.
At the receiving end much more signal power is required to resolve Hellschreiber modulations. The average power is spread over more FFT bins unless the pixel time is significantly longer than the bin length. I've picked a 500 ms pixel time, which is somewhat of a compromise between spectral smearing (due to FM sidebands) and excessive "sloping" of the image produced.
The hardware can not support concurrent Hellschreiber as its power chain is non-linear and there is no method to suppress the other sideband even if it was, a true SSB transmitter would be required. Sequential Hellschreiber has some advantages in required power anyway; all TX power is devoted to a single frequency at a time. Still the modulation sidebands waste some power and provide a "visual SNR" that is poorer than simple FSKCW. There is also the possibility of the FFT bin length beating with the pixel rate, this was observed in early experiments with only 5 shift levels (and no pre-distortion linearity correction).
The 8-level compensated end-result is fairly pleasing, especially for its simplicity. K6HX and ON5EX captured my initial Hellschreiber experiments and despite having a non-optimised pixel length and no pre-distortion the DX reception was quite good. David VK6DI's more "local" grabber (about 3300 km away) receives a somewhat brighter picture.
Leave a comment on this article.
Parent article: 30 Metre QRSS Beacon.