I was going to say much the same as Paul although if the pins of the crystal corroded off, what about the other components? Presumably they are all in the same environment so do you have dozens of components that are falling off? On Thu, Sep 25, 2014 at 7:57 AM, Paul Koning <paulkoning at comcast.net> wrote:

On Sep 25, 2014, at 5:49 PM, Alexis Kotlowy <thrashbarg at kaput.homeunix.org> wrote: It depends on what is your definition of ?near enough?. If it?s a clock crystal for a logic circuit that runs at about that speed, a 10 MHz crystal might be fine. If the frequency controls some output frequency that needs to be held to a tight tolerance, like a video signal, or if it is used for timekeeping, you will want to be close. A standard crystal tolerance is 0.01% (which would make that final 5 a non-significant digit). I don?t know what video timing tolerances are, if you need that. If timekeeping is involved, do you care if the system clock is 7% slow? :-) If not, 10 MHz would work. You might also just get a quote from International Crystal. It may be cheap enough that settling for ?close? is just not worth it. paul

In the TRS-80, the clock crystla _does_ determine the video timing. There is one master clock (this one), divided down for the CPU clock and the video RAM adressing and sync signals. There si no real time clock in an unexpanded Model 1. Thre is a heartbeat inerrupt produced by the Expansion Interface, but I think that has its own crystal. In the M3 and M4, the heartbeat comes from the video circuitry IRIC, so the one master crystal controls everything -tony

On 26 Sep 2014, at 7:49 am, Alexis Kotlowy <thrashbarg at kaput.homeunix.org> wrote: The value is fairly critical as I think all timing is derived from the one oscillator. I do believe that some of the later models (PAL only) used a 10.480MHz crystal, but they are possibly even more difficult to find. Chris