Erik is correct when he says that experience is a hindrance when dealing with the SiLabs "deviates."
thanks

However, if their feature sets match your application (which they do, in my case) then don't be afraid of them.

Just carefully RTFDS, which is advised for any chip.

the problem there is that SILabs datasheets are 300+ pages long. I tried a f3xx and got caught time after time by little things hidden in multipage paragraphs. A couple of examples: using my experience with the SILabs f12x series I designed a f3xx thingy and copied the IIC driver - OOPS, that was nonstandard, had to go through the pain of adapting another faulty appnote. Then I were to use T2 for the UART, but there was no RCAP registers, moved what T1 was used for to something else in order to use T1 for the UART and THEN found that the RCAP registers were there, renamed just for fun (I can see no other reason). Then I got the prototypes and they did not work, I looked and looked and THEN found that just for fun (I can see no other reason) that XTAL1 and XTAL2 had been 'mislabelled' i.e. external clock input was on XTAL2. OK, cut and paste and the chip came to life, but the serial did not work THEN I found out they had 'redesigned' the UART, THEN I THREW THE STUPID THING OUT, I redesigned with a real '51 (SILabs derivative) because I did not have the time to work through finding the rest of the 'improvements'.

If SILabs would start the deviate datasheets with a few pages here is what we have improved (mutilated would be better, but they would never go for that) pay attention to this, for the rest you can go by experience, then I might reduce my opposition towards these deviates but hiding the differences in little words spread over 300+ pages is just plain indecent. Of course, it would be even better if SILabs would take the 'standard' route and make datasheets like everybody else that ONLY list additions and differences from "the bible" (for all their chips, derivatives as well as deviates)

Erik