Finney@ssc-vax.UUCP (Ken Finney) (07/20/85)
The official voltages levels are +12 and -12, giving an awful lot of noise margin. To save money, microcomputers years ago started using +5 and 0. I have seen a lot of peripherals that clip the incoming voltages to +4.7 and 0.7 using a zener and a resistor. These peripherals work whether you are using the official voltages or not. -Ken Finney- @Boeing Aerospace "I know the world is hollow, for I have touched the sky."
hes@ecsvax.UUCP (Henry Schaffer) (07/22/85)
As I remember it, the official voltages (i.e. according to the RS-232C standard) are something like -3 to -12 for one state and 3 to 12 for the other for RECEIVE. For TRANSMIT they change the 3's to 5's. Therefore if you transmit according to the standard there is alot of leeway in receiving - and alot of hacks will receive reliably, although with some (often negligeable) degradation - particularly just slicing at or near the TTL 1.4V. However, if you take shortcuts in transmitting, (which would work with a receiver which works according to the standard) and then you meet a receiver which also takes shortcuts -- you can end up with something between high error rates, intermittent operation, and complete lack of info exchange. (I have the standard at my office - so if you really want to know, ask me for details.) --henry schaffer
henry@utzoo.UUCP (Henry Spencer) (07/23/85)
Note that running your RS232 drivers off a +-5 supply will quite possibly violate the RS232 standard, because +5 supply voltages (from which the -5 is probably derived) are not guaranteed to be >= 5.000V, and 5V is the absolute minimum for the transmitter end of the RS232 spec. Using +5 and 0 as "RS232" is a gross violation of the standard, and works sometimes only because most RS232 receivers are very tolerant. Put that together with a sloppily-done receiver and you're up the creek. Note also that the major reason you can run RS232 for far longer distances than the spec says, is that the transmitters and receivers are overbuilt. If you start cutting corners, you have to be very careful. To quote (very approximately) McNamara's book on data communications: "In the old days, every designer took a bash at meeting RS232 in his own way. The result was a lot of headaches. Nowadays most everybody uses the standard chips [at standard supply voltages -- hs] and the results are much better." -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,linus,decvax}!utzoo!henry
rsellens@watdcsu.UUCP (Rick Sellens - Mech. Eng.) (07/24/85)
In article <55@ssc-vax.UUCP> Finney@ssc-vax.UUCP (Ken Finney) writes: >The official voltages levels are +12 and -12, giving >an awful lot of noise margin. To save money, microcomputers >years ago started using +5 and 0. > > -Ken Finney- > @Boeing Aerospace The specs I've read say that the on and off states are voltages anywhere between -3 and -15, and +3 and +15, with anything between -3 and +3 as "no man's land". The result is that any machine that follows the specs to the letter will not talk to something using the ttl levels of 0 and +5 volts. This gave me trouble interfacing a Novation J-Cat modem (ttl) to an IBM PC serial port (RS232, to the letter). To solve the problem I had to add an interface circuit consisting of an inverter and a line driver to bring the J-Cat's signal levels to spec. Shortcuts on specifications can be a real pain! Rick Sellens UUCP: watmath!watdcsu!rsellens CSNET: rsellens%watdcsu@waterloo.csnet ARPA: rsellens%watdcsu%waterloo.csnet@csnet-relay.arpa
hull@hao.UUCP (Howard Hull) (07/26/85)
Ken Finney writes: > The official voltages levels are +12 and -12, giving an awful lot of noise > margin. To save money, microcomputers years ago started using +5 and 0. > I have seen a lot of peripherals that clip the incoming voltages to +4.7 > and 0.7 using a zener and a resistor. These peripherals work whether you > are using the official voltages or not. > Sometimes, however, they don't work. One way to get one that doesn't work is to use a long, shielded twisted pair cable. The high capacity of the cable causes the normally sharp bit transitions to be spread out as exponentials. Then, depending on the bit pattern, it is possible to get a shift in the bit transition detection point if the threshold point is not exactly between the + and - voltage of the sending source. This results in some bits coming out narrower than others and, in addition to confusing the logic that locates the bit center, this can result in either a missed bit or a misread bit. We have actually had this happen here with PE Fox terminals that have a threshold separated by a clamp diode and a base-emitter junction from ground. Howard Hull {ucbvax!hplabs | allegra!nbires | harpo!seismo } !hao!hull
dale@wucs.UUCP (Dale Frye) (07/29/85)
In article <55@ssc-vax.UUCP>, Finney@ssc-vax.UUCP (Ken Finney) writes: > The official voltages levels are +12 and -12, giving > an awful lot of noise margin. To save money, microcomputers > years ago started using +5 and 0. I have seen a lot of > peripherals that clip the incoming voltages to +4.7 and > 0.7 using a zener and a resistor. These peripherals work > whether you are using the official voltages or not. > The 1489 line receiver (or is it 1488, I can never remember, the other one is the driver) uses -1 and -3 as the threshold voltages however this can be adjusted by a resistor to a bias pin (adjusted up). Military specs are +- 6 volts. Question: Why does DEC have their receivers load the line. I've seen DEC equipment pull a 12v signal down to 3 or 4 volts. Can't they design something that needs less than 10ma. WARNING!!!: Don't use a passive (i.e. with no power source) breakout box on DEC equipment. Between the the drain of the box and DEC, things have a habit of not working. Dale Frye @ Washington University in St. Louis