Depending on the version of tinyos you're using, the 7 bit addressing was fixed to conform to standards around the 2.1 version. Besides this, there was an error in implementation of reading and writing such that you shouldn't use an odd address because it was not masked correctly. I pointed this out about 3 years ago and you should be able to find it in the archives.
Regards, Kurt > Date: Mon, 24 Aug 2009 16:27:28 -0600 > From: Michael Schippling <sc...@santafe.edu> > Subject: Re: [Tinyos-help] Fwd: I2C examples > To: Tommaso Montisci <t.monti...@gmail.com> > Cc: tinyos-help@millennium.berkeley.edu > Message-ID: <4a9313d0.5090...@santafe.edu> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > If I read that trace right you have a 0x64, not 0x32 on the bus. > Could be wrong, but is consistent with 32 << 1 in the wrong place... > MS > > > Tommaso Montisci wrote: > > Hi Michael sorry I've sent the wrong file!!! > > Yes the slave address should be shifted, infact the 2 last bits are the > > r/w bit and the ack (not received...). > > The 0x32 address should be after shifting, is a 7 bits address. I send > > the address through the I2CSA register, after sending the start > > condition. What I would expect is an ack so that then I can send my data > > (a reading command for example). > > Also there should be no conflict on the bus, because I turned off other > > chips (I use a tmote sky, so the usart bus is shared with the radio chip > > cc2420, which I disabled to avoid conflicts of any kind). > > Regarding the pullup resistances, I use the compass with an evaluation > > circuit, with capacitors and resistances following the datasheet (page 6). > > let me know if you have some ideas, also about the picture... > > > > I don't do robotics, I do telecommunication but for a project I have to > > develop a sensorboard with this compass, an accelerometer, and a gyroscope. > > > > Thanks, > > > > Tommaso > > > > 2009/8/24 Michael Schippling <sc...@santafe.edu <mailto:sc...@santafe.edu>> > > > > Actually I don't see anything but a straight line on the Oscope trace, > > but that could just be me... > > > > The slave address needs to be shifted up one bit in the first byte sent, > > because the LSB (bit 8) is the rd/bar-wr bit. It's not clear to me if > > that is what you are doing (because I can't see anything in the > > picture). > > Perhaps that is what you mean by "0x32 followed by a 0 and a 1 (no > > ack)"... > > > > It is also not entirely clear from the spec sheet if the 0x32 slave > > address is before or after shifting, have you tried both? > > > > I'm not familiar with the telos/MSP in this respect. You're sure that > > there is no conflict on the I2C bus, and also that you have the pull- > > up resistors in place? > > > > Looks like a nice compass chip. You doing robotics? > > MS > > > > > > > > Tommaso Montisci wrote: > > > > I'm using the I2C part of the MSP430 datasheet and it seems very > > clear: > > http://focus.ti.com/lit/ug/slau049f/slau049f.pdf > > From my oscilloscope (scl , sda) the synchronization between > > clock and data seems ok, has you can see in the picture attached. > > There should be an ack there on the 9th bit? > > The slave address should be ok, 0x32, the default address > > standing at the datasheet > > http://www.magneticsensors.com/datasheets/HMC6343.pdf > > > > From the picture we can recognize the bits and indeed it is > > 0x32 followed by a 0 and a 1 (no ack). > > > > I do not know what to do... > > > > thanks, > > Tommaso > > > > > > 2009/8/24 Michael Schippling <sc...@santafe.edu > > > > This is the I2C spec for the chip I was fighting: > > http://www.nxp.com/acrobat_download/usermanuals/UM10204_3.pdf > > It has a (to me) useful description of the spec and > > clock/data sequence. > > Careful comparison to your observations might turn something up. > > Pay attention to the relationship between clock and > > data...IIRC the > > data must be stable the entire time the clock is active (high > > or low, > > I don't remember for sure...) > > > > Are you sure you are sending the right slave address? If the > > target > > doesn't agree it won't respond. > > > > Is your TOS MSP I2C implemented in hardware or software? > > > > MS > > > > > > Tommaso Montisci wrote: > > > > Hi Michael, thanks for your answer. > > > > I try to explain you my problem: > > I'm able to send the START condition and the slave > > address, but > > I receive no ACK from my sensor(HMC6343 digital-compass). > > I know > > that I have sent the right-byte through the Oscilloscope. > > Without the acknowledgment it seems that I'm not able to sent > > further data to my device (is it true that I need the ack > > before > > sending data?). > > > > Have you any suggestions? > > > > best, > > Tommaso. > > > > > > <mailto:sc...@santafe.edu <mailto:sc...@santafe.edu>>>> > > > > > > I discovered a "lapse of implementation" in the T1 > > atmega I2C > > implementation, which is BTW entirely done in bit-banging > > software. > > I don't use or have access to a T2 tree and haven't > > looked at the > > MSP implementation, but I might be able to help if you > > describe > > your problem. > > > > My problem was when reading multi-byte data from an > > I2C device. > > According to the spec the slave can request a "wait-state" > > between > > bytes. The TOS code was ignoring this and getting out > > of bit > > sync. > > > > MS
_______________________________________________ Tinyos-help mailing list Tinyos-help@millennium.berkeley.edu https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
