Hi Wayne,
Yes, I can beta test a fix to the solution - I'd love to.
For one, of course, I can check the timing of the serial port with the
scope - I'm sure you can do that, too. I can also test it on the air, both
with FSK and AFSK. A friend of mine, George, W1ZT lives in the same town,
so it is easy to test while communicating on the phone. He has RITTY by
K6STI, which displays the character length in milliseconds. That makes it
really easy to verify what is being transmitted from my end. George had
originally told me that my signal sounded faster than most others, so he is
familiar with the current condition. I cannot directly measure/compare the
timing of the AFSK signal, since the scope would only show the envelope and
I wouldn't be able to distinguish individual bits. But as I said, George
can measure the character length with RITTY.
I'm planning to put some effort into the upcoming ANARTS contest (family
commitments permitting), so that may be a good testing ground.
FYI, the equipment here is a homemade Pentium 133 (nice generic
motherboard). The rig is an Icom IC-745 (I think I mentioned that before)
that produces the FSK signal directly in the 9 MHz IF. I think the actual
frequency shift is only 160 - 165 Hz instead of the specified 170 Hz, and
so far I've been unable to adjust that. Doesn't seem to cause much trouble,
though. Downstream, I'm using a Drake L-4B linear and an A3S tribander.
As I reported, the other night I tried to work 4L1BR, but he couldn't copy
me while I was transmitting in FSK mode. After I switched to AFSK, Shalva
copied me fine, I think it is safe to assume that there is in fact a
difference between the FSK and AFSK output. His signal strength did not
change, so I'm certain propagation was not the issue. I suspected that
Shalva may be using a mechanical TTY machine, so I just asked him about
this (he's on again tonite) and he's using all home-made equipment -
transceiver, computer and modem. As you know, mechanical machines are quite
finicky with the timing, and European TTY's have to be slowed down from 50
baud (I used to own a couple), but that does not seem to be a factor.
Tonight I re-measured the signals from Rttyrite and RITTY. The Rttyrite
signal is a constant 154ms / character, always one 22ms stop bit. RITTY on
the other hand alternates between 1 and 2 stop bits, so the AVERAGE is 1.5
stop bits. Actually, it sends several characters with 1 stop bit, then
several with 2 and back to 1, etc. However, when I used the RITTY demo for
qso's, I occasionally had the same problem that some stations could not
copy me, so that alternating method is not a viable solution. George also
reported that my RITTY output subjectively sounded too fast. George owns
the latest version of RITTY, but he uses AFSK, so we don't know whether
Brian ever fixed this in the latest versions he sold.
I tried to access the com port directly in DOS to see how the UART handles
1, 1.5 and 2 stop bits, but haven't been able to talk to the port, so I
could not measure any output. I'd be curious to know whether you're driving
the UART directly in Rttyrite. I imagine that it may be tricky to do since
DOS and Win95 do not support 45 baud, and you would have to test for the
PC's UART clock rate so you can set the divisor latch registers directly.
I hope I'm giving you enough detail to track down the problem. Personally,
I am convinced that this problem is limited to the direct FSK output, based
on my measurements and qso experience. Let me know if I can help in any
way.
Michael, K1JE
P.S. I hope I didn't confuse you by sending emails from two locations. The
one I sent earlier today was from work. To get the fastest responses, you
may want to send to both addresses: joens@thecia.net and
Michael_Joens@millipore.com. Feel free to call me at (978) 921-1291 if you
are working on this and need additional input.
On Friday, June 04, 1999 11:26 AM, W. Wright, W5XD [SMTP:w5xd@alum.mit.edu]
wrote:
> Michael,
> Thanks for reporting this in detail. I have had sporatic reports that WL
has
> a short stop bit, but, in retrospect, it appears that the people that
> reported it didn't know how to report the difference between running AFSK
> and FSK. That is, a few months ago I spent most of a Saturday working
> through my AFSK code only to confirm with no remaining possibility of
doubt
> that it was generating 1.5 stop bits.
>
> Now your report clearly shows its having trouble setting up the PC's COM
> port. I wish someone had reported that 12 months ago when I first
introduced
> the problem! (I'm assuming your diagnosis is correct).
>
> Do you have time to beta test a fix if I send it to you?
>
> Thanks,
> Wayne, W5XD
|