Actions
Forum Thread Details
  • State Not Answered
  • Replies 1 reply
  • Subscribers 339 subscribers
  • Views 1015 views
  • Users 0 members are here
Related

ADV7511 I2C slave address confusion

henry10210

As documented in my blog entries (https://docs.google.com/document/d/1bTdE2tcweKvikt1bn5L0AepkJdAa-o75LmfEZ-oMBMM/edit and http://henryomd.blogspot.com/2014/10/qt-gui-on-zedboard.html), I have been debugging the ADI's DRM driver for ADV7511 (and ADAU1761).  It seems that every new discovery is followed immediately by another mystery, such as:

* Why not put ADV7511 and ADAU1761 on the same SCL/SDA, since I2C supports multi-slave?  This one already posted to the ADI wiki forum.

* The ADV7511 HW and programming guide clearly says the I2C address is either 0x72 or 0x7A.  I checked the schematic, and since the pin controlling the address is tied to the ground, it should be 0x72 according to the ADI documentations.  BUT the Zedboard schematic shows I2C address 0111001 next to ADV7511 (the address next to the ADAU1761 is garbled, but could--if I squint--be interpreted as 0x3B, consistent with the ADAU1761 documentation.  The DTS shows the ADV7511 address as 0x39.  Even though I haven't yet gotten the I2C bus find to work on ADV7511, other people have clearly gotten the reference design to work (at least with Vivado 2013.4), so the schematic can't be wrong.  So I don't understand WHY 0x39 as the slave address is correct.

Thanks in advance for any insight!

Parents
  • 0

    Hi Henry,

      I ran into this same confusion when I first started working with the PC9548 mux on the Mini-ITX board.  It turns out that the LSB does double-duty for R/W selection, so the address used by the S/W is only the top 6 bits.  To obtain the actual 7-bit hw address of the device on the I2C bus, left shift the address by 1 bit, which turns 0x39 (0011 1001b) into 0x72 (0111 0010b).  So you will see 0x39 used in the software, even though the real address is 0x72.   This works because the LSB of the two valid address selections (0x72 or 0x7A) is always 0.

    Ron

Reply
  • 0

    Hi Henry,

      I ran into this same confusion when I first started working with the PC9548 mux on the Mini-ITX board.  It turns out that the LSB does double-duty for R/W selection, so the address used by the S/W is only the top 6 bits.  To obtain the actual 7-bit hw address of the device on the I2C bus, left shift the address by 1 bit, which turns 0x39 (0011 1001b) into 0x72 (0111 0010b).  So you will see 0x39 used in the software, even though the real address is 0x72.   This works because the LSB of the two valid address selections (0x72 or 0x7A) is always 0.

    Ron

Children
No Data