On Sun, 4 May 1997, Brian D Chase wrote:

[...]
> To add some confusion to my answer, I'm successfully talking to my 
> MicroVAX II on a serial cable with only the following on the DB9 connector 
> console port:
> 
> pin  function
> ---  --------
> 2    transmit \__ lines are crossed to the terminal 
> 3    receive  /
> 5    ground
> 
> ---
> 
> In double checking with my handy RS232 reference, it would appear that
> pin 5 is `signal ground' and pin 1 on the DB25 connector is `frame
> ground'.  Pin 1 on the DB9 connector is listed as `DCD' -- but we are
> afterall talking about DEC here.  It is quite possible that the pin 1 on
> _their_ 9 pin port is `frame ground'.
> 
> [defn: frame ground - chassis ground of equipment.]
> 
> ... Crap, according to my reference card on the DB25 it lists pin 2 as
> transmit data, and pin 3 as receive data -- BUT on the DB9 it lists pin 3
> as transmit data and pin 2 as receive data (and no it's not in reference
> to a null modem cable). Now I'm confused. I'd thought that 2/3 were TD/RD
> on both 9 and 25 pin connectors. Grrrr -- stupid world.

Well, to make you happy, this is a little more complete story...

In short: Don't blame DEC, blame IBM. :-)

There is no such beast as a DB9 RS232. An RS-232 is specified on a DB25
only. For some obscure reason, IBM in all their usual wisdom (the same
wisdom that gave you the PC ASCII) decided that they should use a DB9 on
their PC, and while using RS-232 levels, should change pin 2 & 3 from what
the RS-232 spec said, and also move the ground from pin 7 to pin 5. Their
placement of the rest of the signals seems to be just as random. And, as
usual, for some reason, people think that this is some sort of official
standard. IBM has never followed any standards but their own.

In my experience, the only two companies that *do* follow the standards
are DEC and HP. Now, the DB9 connectors are unstandard, and I wish they
had never used them. But if you look at a DB25, pin 1 is chassis gnd.
Check if you don't have a proper signal ground on pin 7 as well. (After 
all, that's where it's supposed to be... :-)

And while we're at it. A DTE should have a *male* connector, and a DCE
should have a *female* connector. Null-modem cables should have the same
gender on both ends, and a straigh cable should have different genders on
each end. If all companies had done this correctly, the world would have
been a better place to live in. :-)

Normally, you might get away without a ground at all, if you are lucky,
but I would recommend that you try to connect it anyway.

And finally, if the DEC DB9 tries to look like a DB25, pin 5 should be
CTS, which is an input, so connecting ground to it will neither hurt nor
help. It will work the same as a connection without a ground at all.

> Okay, I'll start my answer over from what is real and in front of me and
> working.  My cable consists of a DB9 female connector and a DB25 female
> connector.
> 
> DB9                   DB25
> -----                 -----
> pin 5   is wired to   pin 7 
> pin 2   is wired to   pin 3
> pin 3   is wired to   pin 2
> 
> This works really well for me... but I have no idea if it will work for
> anyone else as RS232 specs seem to be either fundamentally random or a
> function of your location in the time-space space continuum modulo the
> number of M&Ms you've eaten in your lifetime.

Believe me. The specs are neither random, wrong or difficult. It's just
that (as with many things in computerland) too many computer illiterates
are working at high places in big computer companies... :-)

(This is intended as a kick to IBM, all PC-makers, and all other companies
that make computers with RS-232 interfaces, and that can't read a
specification).

	Johnny

Johnny Billquist                  || "I'm on a bus
                                  ||  on a psychedelic trip
email: bqt@update.uu.se           ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol