Timeline for How was parity used with modems?
Current License: CC BY-SA 3.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 16, 2017 at 8:12 | comment | added | tofro | Starting from around 1990, with the upcoming V42 error correction and other "high-speed modem connection protocols" on modem links, the parity setting was only valid on the piece of wire between the computer and the modem. V42 (and other protocols, like what Zyxel modems did) sets up a completely different (and much more complicated) framing structure between the two modems than on the RS232 links. So the serial link setting became kind of irrelevant. | |
| Feb 16, 2017 at 3:06 | comment | added | Leo B. | Most modems >1200 baud modems hid the fact that they worked with a "hardware" error correction from the user, but my first modem was a 1200 baud one with an option of "software MNP" in the form of an MD-DOS driver (up to MNP4, IIRC). On Soviet phone lines with pulse dialing and electro-mechanical switching it was very useful even at 1200 baud. | |
| Feb 16, 2017 at 2:44 | comment | added | RichF | @LeoB. Thank you for the info. I started at 300 baud. But I upgraded gradually as better modems were offered. I probably had 1200, 2400, 9600, 19.2k, and 56k over the years. ... And now I think 6 Mbps is slow. 🤓 | |
| Feb 16, 2017 at 2:32 | comment | added | Leo B. | It is not that the noise was so low, it's the MNP. At 2400 baud or more, working without MNP was virtually impossible, and as soon as you have MNP, you don't need any additional error detection above it. | |
| Feb 16, 2017 at 2:21 | history | edited | RichF | CC BY-SA 3.0 | correction |
| Feb 15, 2017 at 23:42 | comment | added | RichF | @supercat I don't know the history, but the major protocols improved greatly over time. The very name "Modem7" (one of the most popular BBS packages) tells a story of major enhancement. What I do know is that whatever the modems and supporting software were doing, they were really solid. The only fails I can remember is if there were a major line glitch -- little brother picks up phone in other room and starts dialing, that sort of thing. Downloads improved to the point that they could continue from a complete disconnect or power loss. | |
| Feb 15, 2017 at 23:32 | comment | added | supercat | As for parity, I'm somewhat curious what it really accomplishes beyond maybe allowing a dumb terminal to output a "glitch" character when it would otherwise have output a seemingly-valid-but-wrong one. Data errors are so often combined with framing errors that forward error correction is often of limited use within a packet. Better would be to guard each little packet with an 8-bit CRC, and each group of packets with a 32-bit CRC (if the 8-bit CRCs match but the 32-bit one fails, either negotiate the 32-bit CRCs of all the packets or else retransmit the entire group). | |
| Feb 15, 2017 at 23:26 | comment | added | supercat | TDD support would have made it possible to use the TDD equipment found on some public telephones to phone into a computer system. | |
| Feb 15, 2017 at 21:18 | history | answered | RichF | CC BY-SA 3.0 |