For example, with a real Tandy 200, one can stick an oscilloscope probe onto the CLK signal of the external system bus. I just did that and got f = 2.45841MHz.
[image: image.png] An emulator is a bit trickier to touch with an oscilloscope. :-) —b9 P.S. Speaking of "it should be easy with AI", when I googled to find the pinout for the Tandy 200's system bus connector, Google's AI gave me a shortcut before the search results. It informed me that the pins were exactly the same as the Model 100's system bus and helpfully included a complete table. Turns out that that was a fib. CLK is pin 15 on an M100 and pin 29 on a T200. GND is on different pins, too: I used pin 21 on my T200, but you'd likely want pin 18 on the M100. On Sat, Nov 14, 2026 at 2:31 PM B9 <[email protected]> wrote: > If John is measuring standard hardware, that would make sense. Heck, he > could read the expansion port directly and save a lot of trouble. > > I suspect he's doing it this way since all emulators can at least beep. > > --b9 > > > > On November 13, 2026 7:03:34 PM PST, "[email protected]" < > [email protected]> wrote: > >> Why not just use the cassette port and the SID/SOD and leave the PIO out >> of the picture? >> >> >> On November 13, 2026 10:14:04 PM UTC, "John R. Hogerhuis" < >> [email protected]> wrote: >> >>> The idea is a small program that sets up I/O registers and turns off >>> interrupts so CPU clock -> PIO -> beeper. Make a recording and do analysis. >>> With a local AI doing the grunt work so it should be easier than it >>> "sounds." >>> >>> -- John. >>> >>> On Fri, Nov 13, 2026 at 8:04 AM B9 <[email protected]> wrote: >>> >>>> I'd be curious to see how you are estimating the clock speed. Just a >>>> simple machine language loop? >>>> >>>> It would be interesting to see how emulators compare and to test them >>>> against both the M100's internal sense of time and objective "wall time". >>>> >>>> For example, I expect MAME will have the best match to the internal >>>> clock and worst match to actual time. >>>> >>>> Virtual T seems to have a reasonable clock for the M100, but it is >>>> slightly off for the T200. I'm not sure how Virtual T's clock works. It >>>> seems to magically sync with the host computer's time, but does that only >>>> happen at start-up or is it replacing the M100's time-of-day clock >>>> completely? >>>> >>>> --b9 >>>> >>>> >>>> On November 12, 2026 12:31:53 PM PST, "John R. Hogerhuis" < >>>> [email protected]> wrote: >>>> >>>>> >>>>> >>>>> On Thu, Nov 12, 2026 at 11:45 AM Daniel L <[email protected]> wrote: >>>>> >>>>>> >>>>>>> How did it turn out? >>>>>> >>>>>> daniel >>>>>> sysop | air & wave bbs >>>>>> finger | [email protected] >>>>>> >>>>> >>>>> Thanks for asking Daniel. Still working on it... went on a sidequest >>>>> of fixing several little emulator bugs and creating more software tools. I >>>>> had no idea the "undocumented instructions" were finally completely >>>>> reverse >>>>> engineered with a microscope (Shirriff)! Simplifications, little details >>>>> with undocumented flag behavior that the giants didn't figure out in 1979, >>>>> and whatnot. >>>>> >>>>> Anyway, almost back to the clock stuff. >>>>> >>>>> -- John. >>>>> >>>> -- Sent from /e/OS Mail. >> >
