You've failed to read the datasheets correctly.
here you go again, I have NOT referred to any chips and you keep coming back to that.

you keep going on about receiver noise margins which in no way are associated with the negative supply. While I admit a negative supply COULD, possibly, perhaps, improve the noise margin
here you go again, I have, again and again stated "it is NOT the supply, it is the Vil"

there's no indication anywhere that any RS232 receiver uses it, so I challenge you to support the notion that the receiver section of a transceiver that doesn't use the negative supply benefits from it.
here you go again, I have NOT referred to any chips and you keep coming back to that.

Now, if you can show that the MAX-series uses the negative supply to enhance noise margin, then I'll agree.
here you go again, I have NOT referred to any chips and you keep coming back to that.

Until then, and I submit it will be a long time, I'll continue to believe and assert that the negative supply, be it -3 or -15, won't contribute anything in these circuits.
here you go again, how often do I have to repeat that it is not the supply, it is the use of it that makes the difference

Secondly, you need to practice spelling. I've pointed out before what the correct spelling of a few words you frequently misspell is. These aren't typo's, as you constantly mistype them the same way.
Richard, how am i to correct the spelling of words I do not know that I misspell?

Start by showing where any datasheet for any RS232 receiver offers greater noise margin by using a negative supply.
here you go again, I have NOT referred to any chips and you keep coming back to that.

You keep babbling about noise margins but don't indicate what or where they are, let alone what role the negative supply has in the process. Any idiot will agree that six volts are greater than three, but that fact has little to do with noise margins on RS232 communications.
You state "twice the noise margin has no effect", you keep asking me for "proof" DO prove that twice the noise margin is not better than half!

The post from Tony that started it all:
A simple communications setup is to use 2 npn transistors as switches and invert the signal from the micro to the PC and the signal from the PC to the micro.

I do not know how we got to discuss input only which, I have repeatedly stated, will work with 0-3. While a 6 V noise margin definitely is better, the real problem with the post that started it all is what I highlighted above.

I suspect that the "input only" originated from Richard not knowing that the specific programming (AT89C51ED2) discussed involves feedback to the PC which can never even approach reliability with "a npn transistor".

Erik

Again,
OK a proposal
Since you discuss chips, I will agree that many chips use 0 and +3 as input thresholds.
Since I discuss noise margins, you will agree that +3 and -3 will give a better noise margin.

That should settle it, you get your chips (apples), I get my science (oranges). and we can stop this "discussion" which isn't because the first requirement for a discussion is that both parties talk about the same thing.