SUBADRx bits should not be used by default

[pvint]

When I "fixed" my lockup problem in #2 I added the sub address bits (_PCA9685_SUB[1-3]BIT) but I should not have.

After running init with these set the PCA9685 sets more I²C addresses.

My device's base address is 0x70, here is output from i2cdetect -y 1 before init with those bits set:

     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
00:          -- -- -- -- -- -- -- -- -- -- -- -- -- 
10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
40: UU -- -- -- -- -- -- -- 48 -- -- -- -- -- -- -- 
50: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
70: 70 -- -- -- -- -- -- --                         

And here is what I see after running init:

     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
00:          -- -- -- -- -- -- -- -- -- -- -- -- -- 
10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
30: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
40: UU -- -- -- -- -- -- -- 48 -- -- -- -- -- -- -- 
50: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 
70: 70 71 72 -- 74 -- -- -- 

(The device at 0x48 is a temp sensor and is not relevant to this issue)

Removing the SUB bits prevents this, but I still need the ALLCALL bit set for my device to work.

(TODO: I would like to add the use of the SUB bits, as it would be nice to have groups so I can control certain zones of lights together!)

[edlins]

So I'm looking deeper into this and this is what I've found.

• When I use 0x20 for MODE1 (Auto-Increment ONLY) and run the demo, it works and the two MODE registers show up as 20 and 04, respectively. I still don't know why this works for me but not for you.

  pi@raspberrypi:~/src/libPCA9685/examples $ ./PCA9685demo 
  PCA9685_openI2C(): opened /dev/i2c-1 as fd 3
  _PCA9685_readI2CReg(): 40:00:46 20 04 e2 e4 e8 e0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  20 04 e2 e4 e8 e0 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  _PCA9685_readI2CReg(): 40:fa:05 00 00 00 00 1e
  00 00 00 00 1e 
  _PCA9685_writeI2CRaw: 00: 06
  PCA9685_setPWMVal(): reg fa, on 00, off 00
  _PCA9685_writeI2CReg(): 40:fa:01 00
  _PCA9685_writeI2CRaw: 40: fa 00
  _PCA9685_writeI2CReg(): 40:fb:01 00
  _PCA9685_writeI2CRaw: 40: fb 00
  _PCA9685_writeI2CReg(): 40:fc:01 00
  _PCA9685_writeI2CRaw: 40: fc 00
  _PCA9685_writeI2CReg(): 40:fd:01 00
  _PCA9685_writeI2CRaw: 40: fd 00
  _PCA9685_readI2CReg(): 40:00:01 11
  _PCA9685_writeI2CReg(): 40:00:01 11
  _PCA9685_writeI2CRaw: 40: 00 11
  _PCA9685_writeI2CReg(): 40:fe:01 1e
  _PCA9685_writeI2CRaw: 40: fe 1e
  _PCA9685_writeI2CReg(): 40:00:01 01
  _PCA9685_writeI2CRaw: 40: 00 01
  _PCA9685_writeI2CReg(): 40:00:01 81
  _PCA9685_writeI2CRaw: 40: 00 81
  _PCA9685_writeI2CReg(): 40:00:01 20
  _PCA9685_writeI2CRaw: 40: 00 20
  _PCA9685_readI2CReg(): 40:00:46 20 04 e2 e4 e8 e0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  20 04 e2 e4 e8 e0 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  _PCA9685_readI2CReg(): 40:fa:05 00 00 00 00 1e
  00 00 00 00 1e

• When I use 0x21 (Auto-Increment and All Call) I get the same result. Also note that MODE1 is initially still 20 from the previous test but subsequently is set to 21 and read as 21 and the demo also works.

  PCA9685_openI2C(): opened /dev/i2c-1 as fd 3
  _PCA9685_readI2CReg(): 40:00:46 20 04 e2 e4 e8 e0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  20 04 e2 e4 e8 e0 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  _PCA9685_readI2CReg(): 40:fa:05 00 00 00 00 1e
  00 00 00 00 1e 
  _PCA9685_writeI2CRaw: 00: 06
  PCA9685_setPWMVal(): reg fa, on 00, off 00
  _PCA9685_writeI2CReg(): 40:fa:01 00
  _PCA9685_writeI2CRaw: 40: fa 00
  _PCA9685_writeI2CReg(): 40:fb:01 00
  _PCA9685_writeI2CRaw: 40: fb 00
  _PCA9685_writeI2CReg(): 40:fc:01 00
  _PCA9685_writeI2CRaw: 40: fc 00
  _PCA9685_writeI2CReg(): 40:fd:01 00
  _PCA9685_writeI2CRaw: 40: fd 00
  _PCA9685_readI2CReg(): 40:00:01 11
  _PCA9685_writeI2CReg(): 40:00:01 11
  _PCA9685_writeI2CRaw: 40: 00 11
  _PCA9685_writeI2CReg(): 40:fe:01 1e
  _PCA9685_writeI2CRaw: 40: fe 1e
  _PCA9685_writeI2CReg(): 40:00:01 01
  _PCA9685_writeI2CRaw: 40: 00 01
  _PCA9685_writeI2CReg(): 40:00:01 81
  _PCA9685_writeI2CRaw: 40: 00 81
  _PCA9685_writeI2CReg(): 40:00:01 21
  _PCA9685_writeI2CRaw: 40: 00 21
  _PCA9685_readI2CReg(): 40:00:46 21 04 e2 e4 e8 e0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
  21 04 e2 e4 e8 e0 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 
  _PCA9685_readI2CReg(): 40:fa:05 00 00 00 00 1e
  00 00 00 00 1e

• Finally, I see that setting the SUBADRn bits simply directs the PCA9685 to respond on addrs 0x71, 0x72, and 0x74. Removing those simply removes those addrs from i2cdetect. I think removing them is correct until SUBADR functionality is specifically added. So I do approve this PR, but I'd like to get to the bottom of the MODE1 discrepancy so I'll re-open an issue for that. Furthermore, I'll merge this into develop but not into master until there is enough modification to justify a new minor version for the lib.

[pvint]

Good stuff. I'd like to understand the SUBx addresses as well (especially because I think I would like to use them!).

As luck would have it I received new PCBs today and plan to assemble a couple tonight. On the off chance that it's something off with my particular PCA9685 device I was testing with I will make sure to use parts from the new batch.