[GreenKeys] [External] Re: Model 32

[Jones, Douglas W]

From: E. [hanyou at xsmail.com] Friday, March 26, 2021 3:07 PM
> How can a TT properly print at motor frequency if the other side is sending it at a different speed (thus what the gears are for)?  The rangefinder can’t compensate for that — or that much, can it?

The rangefinder sets the time from the leading edge of the start bit at which successive samples are taken to get the data bits.So long as the data has nice clean square edges, if you set the rangefinder to sample in the middle of each bit, you can correctly receive data sent with a clock rate that puts the leading edge of the final bit right before the sample time, or right after the sample time.  On 8 bit data with one start bit and one stop bit, this gives you good reception over clock rates that vary from 95% to 105% of the nominal rate.  For 5 and 6 bit data (7 to 9 bits including start and stop bits, this gives you even wider tolerance.

That's why asynchronous data formats were invented in the first place.  It allows useful communication not only when the sending and receiving commutators are not in phase with each other, but also when the motors are not spinning at exactly the same speed.  Back in the day when I did lots of hardware hacking with asynch data, UARTS and computer interfaces, I took advantage of this by using cheap RC clocks for asynch data instead of expensive crystals.  Having clocks that are off by a few percent is no problem at all.

If you've got signal distortion, the flexibility you have to use the rangefinder to compensate for baud-rate variations declines because the portion of each bit-time where the data is stable gets narrower, and you've got to set the rangefinder to sample in that sweet spot.

           Doug Jones
           jones at cs.uiowa.edu