TTL to RS232 Signal Conversion

Methods:

Use a TTL HI output for the RS232 GND: Steve Walz of armory.com says "Use a TTL HI output for the RS232 *GROUND*!! Then you get +/-5VDC levels to fake out the RS232. When TTL-GND goes HI, then a TTL-XMT LO looks like -5VDC and a "mark". Even sign problems go away in hardware! When TTL-GND goes LO, and TTL-XMT goes HI, it looks like +5VDC and a "space"! You can derive the two TTL signals from one off an inverter or inverter pair! Then magically, you have direct TTL to RS232 conversion!!
M. Adam Davis points out that we should "Just *make sure* the power supply of the pic and/or the power supply of the other rs-232 device are not grounded. ie, my bench power supply has +5v and ground, but the ground is grounded to the house ground, SO is the computer's rs-232 port. I can't use this trick without isolation. This also means that any interaction with other external devices requires some thoughtful planning." [Ed: this is only true if the devices confuse signal ground and chassis ground. Remember that ground is at DB25 pin 7, and not pin 1 Pin 1 is chassis ground, not signal ground.]
Bob Ammerman [RAMMERMAN at PRODIGY.NET] of RAm Systems says:
Here is a neat trick to get a 'push-pull' output from two PORT A pins:
1: define them as analog (ADCON1)
2: TRIS them as outputs.
Now, when you BSF either of them the other will automatically be cleared.

Andrew Warren says:
Of course, if you can't configure the pins as analog inputs (because, for instance, the PIC you're using doesn't HAVE an A/D converter), you can do this:
```
        MOVLW   00000001B       ;RA0 = 1, RA1 = 0.
        MOVWF   PORTA           ;

        MOVLW   00000011B
        XORWF   PORTA           ;Toggle RA0 and RA1.
        XORWF   PORTA           ;Toggle the two pins again.
        XORWF   PORTA           ;And again...
```
Or, if those two pins are the only ones defined as outputs on that port:
```
        MOVLW   00000001B       ;RA0 = 1, RA1 = 0.
        MOVWF   PORTA           ;

        COMF    PORTA           ;Toggle RA0 and RA1.
        COMF    PORTA           ;Again.
        COMF    PORTA           ;And again...
```
Steal the -12 from the RX or other pins: Pin 2 on a peripheral connector (pin 3 on a host connector) will typically rest at -12 as driven by the external device. My Fluke 77/AN reads about -10 on pin 3 of the DB9 connected to my PC. Some people (Tom Engdahl at http://www.hut.fi/Misc/Electronics/circuits/rspower.html) have suggested using this fact to derive -12 for TX power. This makes sense as we normally are not RXing as we TX. How about adding a diode, resistor and big fat cap to provide a more or less steady source of -12v to drive pin 3 below zero when the NPN is biased off? Scenix did this on a demo modem board,

Nationals app note for a mouse actually skipped the cap

Dan Michaels of Oricom has an appnote on this at http://www.oricomtech.com/an-ecd02.zip (cached 20001016 in HTML)
and Engdahl did something similar in a line buffer circuit http://www.hut.fi/Misc/Electronics/circuits/rsbuffer.html but he was using an external power source for extra juice so the amplifier type would have to be changed from emiter follower to a voltage gain amp. They did not use RTS as a source of +12 volts, connecting the transistor to +Vcc (local +5 volts) instead. My thinking is that if RTS isn't at +12, the peripheral shouldn't be sending anyway. In fact, if I bias this correctly, setting up the TX pin as an input would allow the periperal to read the RTS status without needing another IO pin. That would be cool, but Jeff Frohwein says ``it doesn't appear to me that it would work. It appears to me that the transistor would always be on ...'' forcing TxD to always be Hi no matter what data we're trying to transmit on TX.
Here is a circuit and board in Eagle and a picture of the circuit.
Wojciech Zabolotny says
Take the higher voltages from the RS232 port. T1,T2 & T4 may be almost any npn low power transistors (eg. BC107). T3 almost any pnp low power transistor with UCEmax >=30V (eg. BC177). Using of switching transistors is however recommended. Diodes - almost any low power impulse diodes.
http://baserv.uci.kun.nl/~smientki/Lego_Knex/Lego_electronica/PicProg/RS232_basics.htm shows a simple version that he says is tested to 115kbps.