You're right, Maarten, it doesn't consume all your bandwidth if you happen to use the PCA in this way. However, isn't it silly to select an MCU for just one feature when the requirement is so poorly defined as this one? The SiLabs parts are pretty expensive for one to choose it just because he doesn't know a better way to generate a 10 MHz clock. I doubt that Purushottam Dayama is even aware of that feature. My comment asserts that it is nonsense to pursue this requirement in such a way, given that the requirement is so poorly defined. Consider the nonsense that Purushottam Dayama presented in that other thread, claiming that HC logic wouldn't work at speeds above 10 MHz or whatever, and suggesting that the load presented by an HCMOS gate is comparable with multimeter probes. I doubt that Purushottam Dayama is aware of the PCA, nor do I believe he's considered that as an option. I am curious, too, as to how the PCA, opeating from it s 100 MHz system clock, can produce a 10 MHz output on a port pin. The PCA description starts on page 326 of the datasheet, and I'm unable to find, right off, a way that's listed, that can do that. Given that it can use the system clock (100 MHz) or the system clock/4 or the system clock/12 as a timebase, I'd be interested to see how one would wind up with a 10 MHz clock on a device pin without making considerably more compromises than just programming a counter and routing its output.



I doubt you'd be willing to use that much (100%, if you can get there) of your bandwidth to produce a simple 10 MHz clock. There are other ways, and those don't consume MCU bandwidth or costly hardware resources. If it's otherwise justifiable to use the relatively costly SiLabs 'F12x, then I freely admit the PCA does offer a potential solution.