Guys, stop the bickering. From the outside I can see the point being made. As for standards, my manager always pulls me up if i mention 'standard' without a concise reference to the standrd AND revision. Only then can we both be talking about the same thing. So can one of you pull out the relevant latest EIA standard and settle on the spec in question? At least we'll all be informed.

Erik, ease off on the coffee. I can't believe you don't know of the ubiquitous 1488/1489 RS232 driver thats been around since the 60's I think. The 1489 receiver does not have a negative voltage rail. The 1488 driver does have a negative rail.

As for noise margins, the receiver has its noise margin but its up to the transmission side to ensure the signal as received has sufficient margin to ensure reliable communication. Sure you can transmit RS232 from a TTL device and over short distances (~3m) it will work quite happily, over a greater distance maybe not so. So using a bipolar transmit signal is preferable. Anyway, for long distance RS232, the 1488 has more drive capability than the now ubiquitous MAX232. Back in the days when there were mainframes & minis with a multitude of terminals hooked up via RS232, had they used MAX232's most installations probably wouldn't have worked.

Richard, I've got locked in arguments with Erik on many occasions - one ripper was '4 wire RS485'. My point that regardless of whether is was allowed in the standard, it does exist in industry. I chuckled to myself the other day when helping a collegue interface to a chart recorder that used '4 wire RS485'. I believe it is actually called up in the EIA spec, but I'm not motivated to actually purchase the spec just to prove a point.

If it was a boxing match, Richard is winning on points having landed more blows. Erik, he's got you on the ropes.

Erik, try following Kai's method. Kai presents a good body of evidence when there is a dispute, you present your own biased pseudo fact. No-one reading these arguments is able to gain any useful insight into the problem and it is just a waste of space on a forum. If I want my mind filled with trash I watch Big Brother(c). The recent discussion on SMT tantalums brought some understanding about some problems that we'd been having with a product. Evidence was presented and I, at least, learned something new. Whilst I'm here, your stance on RTOS on a 8051 describing them as 'wrong' is just plain wrong. So what if the RTOS copies the cpu context into XRAM - as long as you're aware of the 'cost' in doing this, its up to you to decide whether you can tolerate the 'cost' or not. Is it ok to run a RTOS on a 68HC11 at 1mhz where it might take 20uS for a context switch vs a souped up 8051 that might do it in 18uS by copying the context to xram? You could liken it to whether you design in a Maxim part - the specs are great but the cost is high vs another chip with lesser specs and cost. If the Maxim part clearly outperforms the other for your application then you consider the cost. Sometimes the design can tolerate the cost sometimes not. Thats engineering - chosing the best compromise.