[WL-USERS] FSK Stop Bit Setup

George C. Johnson w1zt@arrl.net
Sun, 06 Jun 1999 07:15:55 -0400

Hi Michael & Wayne

Sorry I arrived home too late last night to be able to run the tests.  I
will be able to do it tonight.  Let me know when you are ready; which I
assume is after taking your Mom to the airport.

Wayne, it sounds like you and Michael have zeroed in on the timing
behavior like a laser guided bomb!!  I smiled at your comment about Bill
Gates' "practices" but also understand that we all still have to live in
the same world.  I also have WriteLog which I am just now able to run
with my station.  I needed to upgrade my ham computer to Win95 to make
the transition.  I am looking forward to trying it for RTTY.  Michael
has convinced me to try FSK vs. AFSK to take advantage of my IC 765 RTTY
filters.  At present I use RITTY with WF1B for AFSK contesting and RITTY
alone for DXing.  I want to switch over to WriteLog for all of my
contesting and Ron sent me a copy last fall but I could not switch the
computer to take advantage of it until now.  I will also need to
subscribe to the reflector and see if I need to do any update since last

Good luck with your testing.  I just opened another email and Michael, I
see you were able to track down 4L1BR and work him with the new patch. 
That's great!  That's the best evidence of success.  But I will call
when I get home later today and we can run the RITTY measurement.

73, George .. W1ZT  

"Michael C. Joens" wrote:
> Hi Wayne,
> This version or Rttyrite seems to work ok. The total character length is
> about 164 ms - close enough to the expected 165ms within the measurement
> tolerance of my Tektonics 453 scope. When I stretch the signal with the
> delay line, the stop bit actually measures (6.6 x 5 ms =) 33 ms, compared
> to the start bit of (4.4 x 5ms =) 22ms.
> The character "A" measures about 22ms High, 44 ms Low, 66ms H, 32 ms L,
> i.e. 1 start, "A", 1.5 stop. The letter "E" measures 22 H, 22ms L, 88ms H,
> 32ms L - again as expected (RS 232 has negative logic levels,as I'm sure
> you know).
> I did not overwrite Rttyrite 9.23, so I could compare the measurements to
> the "before" situation, and confirm that 9.23 still had 22ms stop bits.
>  Going back and forth between the two confirms that the difference is
> really in Rttyrite, not some other coincidental thing that may have
> happened.
> George (W1ZT) and I spoke briefly this morning and we plan to perform some
> on-air testing tonight. I'll let you know how things progress.
> On the receive side, I'm copying a couple European stations right now, and
> the copy seems no different from before. I just worked RA1WZ, with no
> problems to report. I asked him about his setup, he is using a TS-900 with
> WF1B software and a homemade modem. His signal was about S5, and he is
> using 1500 Watts (compared to my 650 W), so I probably wasn't any stronger
> for him (probably a wee bit weaker, reports of course were 599...).
> According to the 16550 data sheet, the setting of the stop bit length does
> not affect the UART's receive operation: "The receiver checks the first
> stop bit only, regardless of the number of stop bits selected." Based on
> that, there should be no problem with this. Do you actually use the UART on
> the receive side when using FSK with soundboard?
> You are absolutely correct that a UART cannot be set to 5 character bits
> with 2 stop bits, nor 6, 7 or 8 chartacter bits with 1.5 stop bits. This is
> true for any 8250 compatible UART I've ever seen, including the do-it-all
> chips. Based on that, Win95 / 98 / NT really can't mess it up too much,
> it's either 1 or 1.5 stop bits for 5 data bits. I don't know how Brian
> Beezley managed to get the alternating stop bit lengths - it's not a UART
> problem as far as I can tell.  I did a quick test of the DOS com port setup
> with the mode command. The results are as I would have expected:
> MODE COM2 110,n,5,1,n
> response: COM2: 110,n,5,1,-    <--- OK
> MODE COM2 110,n,5,1.5,n
> response: COM2: 110,n,5,1.5,-   <--- OK
> MODE COM2 110,n,5,2,n
> response: COM2: 110,n,5,1.5,-    <--- DOS caught and fixed the error
> MODE COM2 110,n,5,2.5,n
> response: Invalid parameter - 2.5   <--- DOS intercepted the error
> MODE COM2 110,n,8,1,n
> response: COM2: 110,n,8,1,-   <---- OK
> MODE COM2 110,n,8,1.5,n
> response: COM2: 110,n,8,2,-    <--- DOS caught and fixed the error
> MODE COM2 110,n,8,2,n
> response: COM2: 110,n,8,2,-    <--- OK
> MODE COM2 110,n,8,2.5,n
> response: Invalid parameter - 2.5   <--- DOS intercepted the error
> I REALLY appreciate your fast response, especially on a weekend - it is
> absolutely wonderful here in Massachusetts today. I promised my girls (5
> and 2-1/2) that I would take them to the park later _if_ they take naps,
> that's how I bought some time in the shack this afternoon.
> Unless any other problems come up due to this change, the FSK output timing
> is now perfect. Combined with the IC-745 FSK mode, the signal meets all
> standards that I know of.
> As I said, I'll let you know what George and I will find out when we test
> on the air. Maybe I can track down 4L1BR again and see whether he can copy
> me with version 9.24 now.
> On Saturday, June 05, 1999 12:55 PM, W. Wright, W5XD
> [SMTP:w5xd@alum.mit.edu] wrote:
> > Please try the following:
> >
> > 1. download this: -snip -
> >
> >
> > 2. unzip it into your \programs folder. It will replace rttyrite9.exe
> >
> > 3. now try the stop bits.
> >
> > My results are uncertain. WriteLog does not drive the COM ports directly,
> > but depends on proper operation of the Windows device drivers. I tested
> on
> > both Windows NT and on Windows 98 (just one machine of each) and
> discovered
> > that:
> >
> > a. the Windows software documentation (SetCommState call if you care)
> > clearly says:
> > "The use of 5 data bits with 2 stop bits is an invalid combination, as
> are
> > 6, 7, or 8 data bits with 1.5 stop bits. "
> >
> > b. Windows NT behaves per the documentation--a scope shows 22 msec stop
> bits
> > with a setting of 1 and 33 with a setting of 1.5. And the driver refuses
> to
> > set 2 stop bits--it returns an error.
> >
> > c. my Windows 98 machine does NOT behave per the documentation. It sends
> > stop bit for a setting of 1 or 1.5, so I tried 2 and lo and behold the
> > driver reported success (huh?) and 1.5 stop bits appeared out the serial
> > port.
> >
> > It appears that Bill Gates and company have as much trouble telling the
> > truth in their documentation as he does in depositions.
> >
> > The code I'm asking you to test, attempts to set 2 stop bits. If that
> fails,
> > it tries 1.5. I have no idea how many machine/OS combinations this might
> > work on. It works on the NT and 98 machines I have to test with. Please
> let
> > me know what it does on yours.
> >
> > The stop bit setting could also affect receive behavior. The driver could
> > very well set the UART such that fewer than 2 stop bits (44msec at 45
> baud)
> > is a framing error. If that happens, then we have an even bigger mess.
> >
> > Ron: this mod affects folks running one of the following in their "TU
> Type"
> > menu settings ONLY!!!
> >
> >       FSK on a dumb TU (Hal-8000 for example)) or
> >       FSK with a sound board setting (NOT AFSK!)
> >
> > It does NOT affect anyone running any other TNC.
> >
> > Thanks,
> >       Wayne, W5XD