[net.dcom] hardware flow-control on a VAXen

brian@sdcsvax.UUCP (05/04/86)

Recently, whilst installing our Ungerman-Bass NIUs on one of the VAXen
here at UCSD, I discovered that certain models of comm line multiplexors
can do hardware flow control.  In particular, the Emulex CS-11 (which
does a pretty good job of looking like a DH11) has a switch for each
line located in the multiplexor panel that when turned on causes the
UART transmitter to pay attention to Clear-To-Send.  The manual for the
CS-11 is very clear on this, but I'd never looked.  (Stupid, hah!)

This hardware flow control is completely invisible to the software.
What it does is when CTS is not asserted, it makes the transmitter stop
at the end of the current character - so you get essentially
instantaneous flow control.  The software (at least, most drivers I've
poked at) don't look at the CTS bit in the status register, so they just
seem to think that the current character is taking an extraordinarly
long time to send.  For most applications, this works extremely well.

If you have had problems with getting output stopped to avoid
overrunning buffers in some peripheral (like lan boxes, terminals,
buffered modems, etc) because the VAXen sometimes will send as many as
64 characters AFTER you ship it an X-OFF, this may provide an answer.

Moral: check your hardware manual.  It might well be that the particular
comm multiplexor you're using might do the right thing, even though the
software doesn't.

	Brian Kantor	UC San Diego

	decvax\ 	brian@sdcsvax.ucsd.edu
	ihnp4  >---  sdcsvax  --- brian
	ucbvax/		Kantor@Nosc 

   "There is more harmony in films than in life."
	- Francois Truffaut

nichols@uwmacc.UUCP (Dennis Nichols) (05/05/86)

> Recently, whilst installing our Ungerman-Bass NIUs on one of the VAXen
> here at UCSD, I discovered that certain models of comm line multiplexors
> can do hardware flow control.  In particular, the Emulex CS-11 (which
> does a pretty good job of looking like a DH11) has a switch for each
> line located in the multiplexor panel that when turned on causes the
> UART transmitter to pay attention to Clear-To-Send. ... 
>
> ...  For most applications, this works extremely well.
> 
> 	Brian Kantor	UC San Diego

We purchased Emulex muxes with the hope of using this feature.  VMS
didn't like it because writes would timeout if Clear-To-Send was held
off too long.  Worse than that, the Emulex would sometimes lose
characters from a write with the switch in the 'use CTS' position.
Because of problem one we did not push too hard to get problem two
fixed (seemed to be a firmware bug having to do with length of line,
not with the CTS signal actual going low).  Emulex may well have 
corrected the situation by now.

Dennis Nichols  University of Wisconsin-Madison Academic Computing Center

david@varian.UUCP (David Brown) (05/06/86)

> Recently, whilst installing our Ungerman-Bass NIUs on one of the VAXen
> here at UCSD, I discovered that certain models of comm line multiplexors
> can do hardware flow control.  In particular, the Emulex CS-11 (which
> does a pretty good job of looking like a DH11) has a switch for each
> line located in the multiplexor panel that when turned on causes the
> UART transmitter to pay attention to Clear-To-Send.  The manual for the
> CS-11 is very clear on this, but I'd never looked.  (Stupid, hah!)
> 

The Able DH/DM multiplexor also has some jumpers (located in the upper
right hand corner) to enable hardware flow control; unfortunately,
the manual is *not* very clear on this.  The mux will stop sending when
it senses pin 22 (RING) go low; we made up special cables at the computer
end tying together pins 22 and 8 (we chose to flow control on CARRIER
(crossed with pin 20 - DTR) on printer lines rather than CTS).

-- 
	David Brown	 (415) 945-2199
	Varian Instruments 2700 Mitchell Dr.  Walnut Creek, Ca. 94598
	{zehntel,dual,amd,fortune,rtech,lll-crg,rtgvax,ptsfa,csi}!varian!david