Ok, so...

"Check
http://www.atmel.com/dyn/resou...oc4231.pdf
page 9 and 10. No 18MHz Xtal is in the table.
So, the auto baudrate routine does not lock and no ISP is possible.
Daniel"<-Good point but my training board has an 18.432MHz crystal and works flawlessly(I have its schematic too, and it doesn't use any frequency divider circuit etc after the xtal so I guees 18.432Mhz is suitable).

Hey theres another chap called Kai too :). Ah, I see 470uF is not suitable. It's a shame I didnt include 0.1uF decoupling capacitors (didn't include one even on the max232). Bad design! I'll make sure to include them. To be honest there was some noise on the micro pins (bout 500mVpp).

But what's the reason that 0.1uF is better than 470uF when it comes to high frequency noise? Hmmm...wait I have the art of electronics (Horrowits and Hill) I'll check that out. I'm pretty sure theres a formula somewhere that includes the dv/dt (rate of change of voltage) somewhere. Unfortunatley i don't have any other 89C51 microcontrollers (had 5 of em and uh.....all of em where destroyed during the debugging process....shame).