cabo@DB0TUI6.BITNET (Carsten Bormann) (05/30/86)
/***** tub:mod.comp.vax / bu-cs!bzs / 5:16 am May 25, 1986*/ > > I remember hearing a rumor once that there is a reason DEC's muxes > usually work past the "official" RS232 length limits, like that they > were designed to. Anyone know anymore? (of course, this may depend > on the mux.) Might help settle this periennial argument between the > spec's folks and the "but I've done it a thousand times" folks (like me.) > > -Barry Shein, Boston University /* ---------- */ DEC's newer muxes (and terminals, too) actually don't have line drivers for the RS232C standard (better known as CCITT V.28 in the rest of the world), but their drivers adhere to RS423 (really CCITT V.10), which is upwards compatible to RS232C. Cheap RS232 boards usually employ MC1488 line driver ICs, which are pretty weak (their output resistance is relatively high). DEC uses Fairchild uA9636 line drivers, which tend to get fried a little more easily than MC1488s, but are much stronger. They also have a very good control of the signal rise and fall waveform. Look at a 9636 driven line with an oscilloscope, and you will admire the Fairchild chip designers. Ah, yes: 9636s drive the line between -5.6V and +5.6V (this is according to RS423), while 1488s use the full RS232 range of -12V to +12V. Since their supply voltage still is +/- 12V, 9636s burn a lot of power (something around 600 mW for an 8-pin chip), i.e. the board gets HOT. Carsten Bormann, <cabo@tub.UUCP == cabo@db0tui6.BITNET> Communications and Operating Systems Research Group Technical University of Berlin (West, of course...) Would you believe that all this comes from a CS major who never heard a single EE class? I just wanted to write an X.25 for the IBM PC and could not find a useful card for this, so I had to help in designing one...