I'm using FSK to transmit RTTY in Writelog. Works beautifully with my old
Icom IC-745, and I enjoy being able to use the RTTY setting for the proper
bandpass filtering for the high tones, including a 250Hz wide filter (I
could copy 4L1BR while someone in my town was working OM3NY only 1 kHz
away, with an S9+50dB signal).
I also like the fact that the FSK generated signal is very clean. In the
IC-745 it is generated by shifting the oscillator on the 9MHz IF, and the
rig allows continuous duty at 100W. None of the splatter problems when
using AFSK. I understand that not all new rigs generate true FSK even when
so labeled.
There is only one problem: the signal is too fast, resulting in some
stations not being able to copy me. They can hear me, but can't decode.
Most stations have no problem, but 4L1BR for example could not copy me.
When I switched to AFSK, he called me!
People have reported that the diddle sounds too fast, and measurement with
an oscilloscope confirms that the total character length is only 154 ms. At
45 baud (really 45.45 baud, i.e. bit length of 22ms), that amounts to 1
start bit, 5 information bits and 1 stop bit. Each bit is 22ms long, as it
should be. However, according to the teletype standard, the stop bit should
be at least 1.4 bits long. I've set the serial port to 1.5 stop bits (1.4
is not available in PC's) and to 2 stop bits in both the Win95 setup and
Writelog.ini, but the character length remained 154ms no matter what. If I
could slow the signal down to 1.5 or 2 stop bits, the output would be
"perfect". Also, distant stations would probably get a more stable
reception if the stop bit length meets the minimum requirement (it can be
infinitely long - there was no diddle in the old mechanical machines).
I have tried several serial ports. COM 1 and 2 are the built-in ports (on
the COM/LPT/Floppy do-it-all controller chip) and COM 4 is a true 16550
chip. I also tried a true 16450 chip, same problem on all chips. I seem to
have the same problem with the demo version of the RITTY progam by K6STI,
too. RITTY apparently tries to "fake" the 1.5 stop bits by alternating
between 1 and 2 stop bits, although that may be caused by the com port
hardware, not the software (I cannot distinguish that).
Of course, I could use AFSK, but I MUCH prefer the FSK method, especially
if I could get the stop bit problem solved. I would be very happy to settle
for 2 stop bits, i.e. 176 ms character length. True 1.5 stop bits (not
alternating 1/2) would generate near-perfect 165ms character length.
Has anyone experienced the same problem? How can I get WriteLog / Rttyrite
to give me a longer stop bit?
Any help would be appreciated.
Michael, K1JE
|