On Tue, Jun 13, 2017 at 6:59 PM, Chuck Guzis via cctalk < cctalk@classiccmp.org> wrote:
> Well, I didn't say "timing error", I did say "timing distortion", which > is not quite the same thing. My reference was the "TR1602/TR1863/TR1865 > MOS/LSI Application Notes Asynchronous Receiver Transmitter", which can > be found in the WD 1984 Data Communications Handbook (I think there's a > copy online). Page 126-127. "Thus, signals with up to 46.875% > distortion could be received." They're referring to 46.875% of a bit time as the maximum error in the sampling time of a single bit. That still corresponds to no more than 5% error of the overall bit rate, which is where use of an RC oscillator runs into trouble. > Obviously, the developer of the subject board didn't have > much of a problem either; or else he wouldn't be able to sell the thing. > They sold it, then spent a bunch of money on Field Service trips to make it work for customers. It cost them enough to justify multiple redesigns, including (finally) switching to a crystal. I'd think that if an RC oscillator could have been (inexpensively) made to work well enough at that time, DEC would have done it. Their hardware people weren't complete idiots. (Usually.) Eric
