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.

TS5A21366: Power-down Isolation behavior

Part Number: TS5A21366

I am experiencing unexpected behavior in a circuit where TS5A21366 is used to isolate an I2C bus from a subcircuit that is powered down. I simplified the issue to the following stand-alone circuit:

When 3V3 is present the NO pins are pulled to 3V3 by the 10k resistors. When 3V3 is switched off, the switches are supposed to open so that the NO pins are pulled by the weak voltage divider on the right. When 3V3 drops fast, everything works fine: Green = VDD, Pink = IN1, Blue = COM1, Yellow = NO1

But when the 3V3 drops slowly (as it does on my actual board), the switches remain on for almost 10 seconds before turning off, effectively pulling down my signals on the NO side:

There's no power-down behavior documented in the datasheet. Is this expected behavior? Do you have anny suggestions for a fix?

  • The IN1/IN2 pins are high-impedance CMOS inputs and must not float. Also, the voltage at the NO pins must not exceed VCC.

    The switch should stay powered. If this is not possible, use an I²C level shifter that has overvoltage-tolerant I/Os when off.

  • But how about the explicitly mentioned feature "Isolation in Power-Down Mode, VCC = 0"?

    "The TS5A21366 signal paths are high impedance (Hi-Z) when VCC = 0. This feature ensures the signal path is isolated when not in use to avoid interfering with other signals in the system." 

  • Sorry, I only looked at the absolute maximum ratings, and they contradict this sentence.

    Anyway, as long as VCC has not yet reached 0 V, the isolation feature does not work. As far as I can see, the isolation feature is useful only when you disable the switches before powering off.

  • As Clemens mentioned, the isolation feature is spec'ed for VCC = 0V but not when still powering down. What's the purpose of this circuit exactly? You mention isolation  from a powered down circuit down stream but then you're pulling the isolated side up to some voltage.


  • Thanks. However, even if the power-down feature would not work for VCC between VCCmin (1.65V) and 0V, I still don't see why the switch would remain turned on for 7.5seconds after VCC dropped to 0V.

    Purpose of this circuit is to isolate the I2C master on the left from the I2C slave on the right. The latter isn't powered off, and therefore needs a defined high voltage level on it's I2C lines. This is done by the weak voltage divider as there's no simple pullup voltage present.

  • Hey Henk,

    I understand. That is a bit of strange behavior. In the first scope shot where it appears to be functioning properly, the Vcc doesn't seem to be actually included in the shot could you provide a shot with that included? 
    Is the 3V3 source all the same? Can you power up/down from a different source and is the ~7.5seconds consistent regardless of how slow the 3V3 is ramped down? or are you seeing the delay referenced to a particular point in the decay?


  • Hi Rami,
    Hereby a shot where you can see that VCC drops at the same rate as IN1:

    I see no consistent behavior in the delay, have seen 1 second up to 10 seconds. I powerd the circuit above with my original supply and with labsupply combined with multiple capacitances on VCC to vary slope.

  • Hi Henk,

    We are currently on Holiday in Dallas and will be returning tomorrow to help answer your inquiries. 

    We appreciate the patience.


    Parker Dodson

  • Hey Henk,

    Do you have the ability to reduce the absolute resistance of the two 680k resistors in the voltage dividers and see if this resolves the issue?
    What could be happening here is the current require to enable the isolation may be higher than what can be supplied there with the 680k resistors.