(Answering to myself.) > I have somewhat good results, since: > > 1* The power led lights up. > > 2* The bit_test option of our i2c-algo-bit module (which I had to > tweak a bit because the eval board doesn't let me read SCL back) > suggests that SDA can be set and read back correctly. > > 3* When running some commands on the bus (using i2cdetect for example) > I detect electrical activity on both SDA and SCL. I'm using a low-cost > multimeter, unfortunatelly I don't have anything better to watch the > signals. I'd love a good oscilloscope these days :/ > > Now the problem is: i2cdetect doesn't detect anything. Not a single > chip. Running i2cdump on address 0x4c (where I know the adm1032 has to > be) fails reporting a single valid byte too. It looks like the adm1032 > doesn't want to answer, and I can't understand why. > (...) > Sean, is there something I'm supposed to do that does not appear on > the schematic? Some pin of the parallel port that must be forced to > either high or low for proper operation? I'm kind of lost now. I was > expecting everything to work OK once I'd have worked the bit-bashing > driver out, and am diappointed that it still doesn't work. Your > Windows tool being working on the same machine, I know that there is > no hardware incompatibility, so that must be something I am doing > wrong somewhere. I've even been trying various speeds, but it did not > help. OK, I finally got it working! :) After trying various modules and parameters, and even trying to link the eval board to one of my computer motherboard's SMBus, with absolutely no results for 3 days, I was about to give up. Then I read the schematic once again, to make sure I had not missed something, although after reading it ten times I was pretty sure I wouldn't discover anything new this time. And finally I understood what I had made wrong. You'll probably find it ridiculous since it was so obvious. It was all plain written on the schematic. All inputs and outputs are inversed. I had to write *1* to SCL or SDA to pull it low, *not* 0, and all the rest works the same way. I can't blame anyone but me there. Would I have been able to read the nice schematic correctly at first, I would have it all working within an hour or so... I'm not a hardware guy, so it did not even imagine that I would have to write a 0 to set the line high, and so on. So, sorry Sean for annoying you with my silly questions ;) I'll now move on to the ADM1032 driver itself, and that should be rather fast since we already support a very similar and mostly compatible chip (LM90). I'll read the datasheet once again to make sure there's no hidden trap, and then will let you know when I got it working :) Thanks again. -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/
