??? 07/10/04 13:04 Read: times |
#73958 - RE: Analog Devices or TI Responding to: ???'s previous message |
I can't say much about TI, but I was messing around with ADuC812 for a few months, so I can say a few words about it...
1) The app notes say the protection bits mechanism is broken. I never used it anyway (open source project), but if you plan on locking the code once mature, definitely investigate further. 2) ADuC812 Flash write works in a somewhat strange way: From CPU standpoint it's executed within 1 cycle, only the cycle gets very prolonged (just like the clock was stopping) - it may completely mess up timing issues so use with caution. Also the access is quite original - doesn't use movx or such, but 9 SFRs: command register, page address, 4 data bytes in a page and 3 timings regs ("factory settings - do not touch"). 3) It's one ADC, only multiplexed over 8 outputs (same pins as port 1, which may be used as output port when ADC is inactive). It requires quite a bit of configuration work both on startup and on each use (unless you start DMA...) - but the two DACs are genuine and very pleasant to use. Everything else I have tried worked as it was supposed to :) Note the old eval board (Rev 2) came with some MAX232 variant and quite a chunk of xram built in. The new one (Rev 4) should come with special RS232 cable with MAX chip built in, but in our case it didn't (push the distributor if you buy it!), and just room for XRAM. It's quite a bit cheaper though I heard (we got the Rev2 for free anyway so I can't grant accuracy of claim about its price). Also note the included serial programming app isn't really handy for bulk applications. There's source for a DOS version included though and with some effort may be ported to other systems - the upload format in newer ROM of ADuC812 feels much worse than the old one though - while back then you could've sent raw intel HEX file with some simple invocation, now it requires quite a bit of translation. (and the program included contained some OS-specific piece connected with casting pointers that triggered a bug on Linux, at checksum creation - it may do the same on other systems so take care while porting...) |