> Wow, some programs are really slow enough that they cannot keep up?
> (Remember, all you need to do is to keep up with the RTTY character
> rate, not the bit rate.)
The problem is that some programs *DON'T* bother to time characters!
They simply send them to the serial port and expect the internal
buffer management in the 8250 to indicate when the port is ready to
accept another character. Such port buffer management is quite
brain dead and harkens back to systems that were completely character
based I/O.
Although MMTTY and MixW do a good job of pacing their data there are
still programs that simply dump an entire buffer on the operating
system and expect the "no more data" indication to reliably tell
them when to drop PTT.
73,
... Joe, W4TV
On 3/21/2011 7:16 PM, Kok Chen wrote:
>
> On Mar 21, 2011, at 1:21 PM, Gary AL9A wrote:
>
>> "Some programs rely on the the UART "buffer full" signal for proper PTT
>> timing and drop PTT (unkey) when the UART buffer is empty..."
>
> Wow, some programs are really slow enough that they cannot keep up?
> (Remember, all you need to do is to keep up with the RTTY character rate, not
> the bit rate.)
>
> I remember when I encountered the FSK interface of the microKeyer
> for
the first time (on Mac OS X, you have to write everything yourself, so
you end up knowing all the nitty gritty of the stuff you use :-) . One
of the obvious things is that the only way to interface to a MicroHAM
keyer properly is to use its FIFO as a single character buffer and use a
high priority thread to feed characters to it whenever the single
character FIFO is empty. As long as the thread wakes up and sends the
next character within 150 ms there is no extra delay. 150 ms is an
eternity as far as modern computers are concerned. The microKeyer has a
multi-character FIFO, but you cannot really use more than one character
since there is no way to "flush" characters that are sent ahead to the FIFO.
>
>> I checked MMTTY and on the Decode tab the StopLength
>> item selected is "Rx=1bit, Tx=1.5bit". I changed this to "1bit" and tried
>> the same test exchange - 8:68 seconds! My exchange was set up with spaces
>> between the elements as "RST NR TIME NR TIME. I changed that to have a "-"
>> in between the elements so that now it is RST-NR-TIME-NR-TIME. The test
>> exchange time is now down to 8:05 seconds! That's 2:39 seconds off the
>> orginal exhange time!
>
> Uh oh, I might have turned Gary into a microsecond counting monster :-)
>
> BTW, you have to weigh the time saved by using the dashes against
> the
probability that the receiver gets an error hit within the first group
of NR-TIME that causes one of the characters to turn into the LTRS shift
Baudot code. If that happens, you *may* be asked to repeat the entire
exchange (experienced RTTY ops won't be fazed by it). So you need to
weigh the repeats against the savings that you have accrued.
Probabilistically, I myself would pick using dashes over using spaces
between numbers.
>
> While you are shaving microseconds, here is another hint... the
microKeyers and digiKeyers also suffer from the fact that its FSK
interface cannot achieve 45.45 baud (if you use the Mac you'd already
know this since it is in the documentation for µH Router :-). The
software has to choose either 45 baud or 46 baud.
>
> So, if you are a crazed microsecond counter, you might want to
> choose
46 baud instead of 45 baud when using FSK on a MicroHAM keyer. As long
as you are using inaccurate baud rate, you might as well be wrong on the
higher speed side :-). It will give pretty much the same error rate (46
baud only degrades from 45 baud by a practically immeasurable smidgen),
but it is 2% faster than if you let the software round 45.45 baud down
to 45 baud.
>
>> Kok, would you also like a bowl of Fish Head Soup?
>
> You know that the cheeks are the best part of the fish head, right?
:-) :-) At least for Asian fish head dishes.
>
> 73
> Chen, W7AY
>
> _______________________________________________
> RTTY mailing list
> RTTY@contesting.com
> http://lists.contesting.com/mailman/listinfo/rtty
>
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|