[GreenKeys] TTY-CONNECT loop adapter thingy

gil smith [email protected]
Thu, 14 Aug 2003 17:27:49 -0700


Hi Jack:

At 03:55 pm 8/14/03, Jack wrote:
>Gil,
>
>Not sure what value the high-speed PC ASCII port would have without
>any buffering built into the TTY-CONNECT board, unless there
>is software for the PC that actually sends the high-speed stuff at
>the speed that the TTY needs (through the micro-controller).

This is easy -- no buffer needed.  Well, the micro has about a hundred byte 
internal buffer that smooths out the serial port interrupt routine, but 
when this gets near full, it sends the PC an XOFF char.  All further 
buffering is done in the PC until it gets an XON.  It is all transparent to 
all involved, and allows any PC application that can print to the serial 
port (eg: with the generic text driver), to print to the tty at 19200 (or 
whatever).  The PC serial port just needs to have XON/XOFF flow control 
enabled.



>My two cents...I'd rather see some high-capacity buffering for
>baudot speed converting. Heavymetal does agreat job of doing
>all of the internal trickery and outputting TTY Baudot. What would be
>nice would be to have a way to drive several machines at different
>speeds (which is what I do with the "Black Box" gadget).

Yes, for the baudot conversion, it needs to buffer the data for hi-to-lo 
speed conversion.  All a question of how much.  I wrestled with driving a 
big sram (with all the pins), and then realized that speed should not be an 
issue.  So I'm going to put eeproms (nice little 8-pin dips) in 
there.  Even with 5ms write cycle delays (per char) for the eeprom, there 
is more than enough time to save chars.  I wish they made the same package 
in a pure serial sram, with the same I2C interface as the non-volatile 
eeproms, but I have only seen serial srams in very small sizes.  I'll also 
provide a place on the board for a daughter card (you never know what makes 
sense later on).

Did you say that your black-box converter had 64K chars of 
buffering?   Let's see:

100-wpm - 60-wpm = 40-wpm buffer fill rate, worst case, with streaming text.

At the standard 6-chars-per-word = 40 x 6 is approx 240 chars-per-minute 
fill rate.

65536 / 240 = 273 minutes to fill the buffer?  Is that about right?

gil




>Jack WA2HWJ

;-------------------------------------------------------------------------
;  vaux electronics, inc.             480-354-5556
;  www.vauxelectronics.com    (fax: 480-354-5558)
;-------------------------------------------------------------------------