This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

PCA9306 to Isolate I2C Drivers on Different 3.3V Supplies

Other Parts Discussed in Thread: PCA9306, TCA9406

I have two I2C devices connected together that both have a 3.3V logic level. The issue I am running into though is that one of the devices is powered down via a switch at times. This is causing issues in my circuit because the device has a ESD protection diode and when off these creates a current path from the I2C pullup to GND (Don't adhere to I2C specifications of HiZ when the chip isn't powered).

I am hoping the use the PCA9306 to isolate these two devices so that when one side turns off there is no current path through the SDA/SCL lines. My plan is to place the PCA9306 chip in line and have pullups on both sides (one set to the always on supply and one set to the switched supply).

My question are:

  1. Can this chip operate correctly if both VREF voltages are 3.3 volts and there is not actually a level shift (I have seen some other similar chips that specify there must be >1V difference between the VREF values)?
  2. Will this device isolate the two sides if one of the VREF voltages is removed while the other VREF is still present?
  3. Are there any internal pullup resistors in this chip (I don't see reference to any but again have seen this is other similar chips)?
  4. Is there a preference to which VREF I connect the switched side?

The following post seems to indicate that the answer to 2 is YES as long as VREF2 and EN are tied together then connected to my supply. Is this still true if both supplies are 3.3V instead of 3.3V and 1.8V?

  

If this chip is not the best solution, please recommend an alternative.

Thanks in advance!

  • Hi Kevin,

    1. Can this chip operate correctly if both VREF voltages are 3.3 volts and there is not actually a level shift (I have seen some other similar chips that specify there must be >1V difference between the VREF values)?
      1. Yes, this chip operates correctly if both input and output SCl/SDA are 3.3V. In this case you can tie the VREF1 and VREF2, and the Enable pin to 3.3V (no 200kohm pull-up needed on VREF2)
    2. Will this device isolate the two sides if one of the VREF voltages is removed while the other VREF is still present?
      1. You can tie the EN pin to the VREF side which should control whether to disable or enable the translator
    3. Are there any internal pullup resistors in this chip (I don't see reference to any but again have seen this is other similar chips)?
      1. No. There are no internal pull-ups on this chip
    4. Is there a preference to which VREF I connect the switched side?
      1. No preference

    PCA9306 is the best solution for this type of I2C bus isolation. TCA9406 is another alternative where you can use the EN pin to connect or disconnect two sides.

    Thanks,

    Rajan

  • Thanks for the response, that is basically what I was hoping.

    Here is the schematic I am planning on using. Does this look correct to you?

    The FULL_VCC33 would always have power and the SW_VCC33 would be turned on and off. My understanding is that when SW_VCC33 power is off, the chip will act as a block between the two and no voltage would be present on the _BUF lines.

  • Kevin,
    Schematic looks good.

    Minor suggestion: For this particular application scenario you can even take the 200k resistor out and pull-up the EN/VREF2 pin directly to VCC33. Since the VREF1=VREF2=EN =3.3V there is no current flow through that switch path, so no need for the 200k resistor to minimize the current. The device will work correctly with or without the 200k resistor for this case.

    Thanks,
    Rajan
  • Rajan,

    Awesome. Thanks so much for your input on this. I removed the 200K and look to be all set.

    Kevin

  • Could this be added to the datasheet , as it states that VREF2 > VREF1 + 0.6V, which is different to above thread?
  • Hello Dan,
    We do plan to adding this to the application section of the datasheet the next time we update the datasheet. We won't put it in the specifications section because we characterized the device in this particular implementation and some of the specifications would be different if used like this.
    -Francis Houde
  • Hello,

    I also have an application similar to the above schematic. However, I want to be able to control the EN pin. In this case would I need the 200K pullup for VREF1 =VREF2? My EN pin may be low when VREF1 and VERF2 are 3.3V

    Thanks,
    Jason
  • Hello Jason,
    Can you please draw out the proposed circuit and explain what you want to do with respect to that schematic.
    -Francis Houde
  • Hi Francis,

    Here is my proposed circuit.  In the above circuit, the V_3V3 comes on well before V_RT_3V3.  First, I want the PCA9306 to isolate the buses until both power rails are up.  Second, I want to control when the PCA9306 connects and disconnects both I2C buses.  I want to make sure that when Local_I2C_EN is logic zero, that both I2C buses function as 2 independent I2C buses.

    Thanks,

    -Jason

  • Hi Francis,

    Here is my proposed circuit.  In the above circuit, the V_3V3 comes on well before V_RT_3V3.  First, I want the PCA9306 to isolate the buses until both power rails are up.  Second, I want to control when the PCA9306 connects and disconnects both I2C buses.  I want to make sure that when Local_I2C_EN is logic zero, that both I2C buses function as 2 independent I2C buses.

    Thanks,

    -Jason

  • Hi Francis,

    Here is my proposed circuit.  In the above circuit, the V_3V3 comes on well before V_RT_3V3.  First, I want the PCA9306 to isolate the buses until both power rails are up.  Second, I want to control when the PCA9306 connects and disconnects both I2C buses.  I want to make sure that when Local_I2C_EN is logic zero, that both I2C buses function as 2 independent I2C buses.

    Thanks,

    -Jason

  • Hello Jason,
    Yes, this should work. It should function as you described.
    -Francis Houde
  • Hi Rajan,

    Will this device work in the same manner if both VREF1=VREF2=1.8V (i.e., one constant while the other is switched)?

    Thanks,

    Mark

  • Hi Rajan,

    How is the Vth created for the gate drive, if all voltages (VREF1, VREF2, and EN) are at +3.3V?  The data sheet says 0.6V is needed for VREF2, as a minimum, to achieve the Vth of the gate.  This also shows in the post by Bobby Nguyen in the other thread.

    Is there an internal charge pump to create the necessary voltage of VREF1 + Vth?

    Thanks,

    Mark

    ..

  • Hey Mark,

    There is no internal charge pump to generate Vref1+Vth. In the case where Vref1=Vref2 the EN reference voltage will be whatever the EN pin is tied to which in this case is 3.3V because of Vref2. Having this higher reference voltage on EN helps other internal circuitry pull side1 to GND faster than if it were at just Vref1.

    Thanks,

    -Bobby