On Thu, 2004-07-01 at 17:45, Jean Delvare wrote: > BTW, d15 on lm75 is sign as well. And what's the point in lm77's having > 4 bit for sign? If that means that all 4 bits will always be the same, > that's something we can use for identification. The high 4 bits can easily be the same for the LM75, too... > >Maybe we could check for the existence T_high and T_low registers? > We cannot. LM75 chips will not answer with an error if you try to read > these registers. It will return a value, just that this value isn't > relevant. I cannot remember exactly what it returns... Possible the > value of register 0x03. I see. Maybe writing could tell us more (write then read back and compare) but that can be dangerous obviously. > What we may use OTOH are: > * unused bits in config register if they differ; The config register is 0x0 at me by default. Don't know anything about the LM75 though. > * unused bits in temperature registers; they differ, and it should be > sufficent since it's VERY unlikely that ALL 4 low bits of all > temperature values are 0 on a LM77 (while the same bits will read 0 on a > LM75). That should be a good thing to start with; I'm not sure if it's really that unlikely. Some examples: 8, 16, 24, 32 (!), 40 degrees... > Apart from the fact that you are not interested, will you have the > possibility to test things like sensors-detect (perl script) or i2cdump > on the box? Yeah, I'd be glad to help. Thanks, Andras
