No output from PCA9507D

[kadowaki]

When port B is + 3.3V drive master and port A is + 5V drive slave, data and clock from port B may not be output from port A.
There are PCA9507Ds that do not occur depending on the individual PCA9507D.
If PCA9507D, which has a problem, is installed in a separate + 5V drive product, the problem will not occur.
Is there any possible cause?
Have similar bugs been reported?

[kadowaki]

Even if a failure part is mounted on a good board, a defect will occur.
Therefore, a defect occurs depending on the part.

[kadowaki]

Hi, G.w. Sun,

Thanks for you help.

Request IC failure analysis from the buy channel.

[guoweisun]

I forget one,do you also try use the failure part in good board to verify?

[kadowaki]

Hi, G.w. Sun,

Thanks for the reply.

Port B has a 4.7kΩ pull-up resistor on the TX side and PCA9507D side.
Port A has a 1.8kΩ pull-up resistor.

The clock speed of IIC is 76.9 kbit / s.

If the PCA9507D where the problem occurs is mounted on another product driven at 3.3V, the same problem will occur.
If the PCA9507D where the problem occurs is mounted on another product driven at 5V, the problem does not occur.
When the new PCA9507D is mounted on a product that operates at 3.3V, no malfunction occurs.
There are about 10 PCA9507Ds that fail at 3.3V.

[guoweisun]

What's your pull up resistor about both sides of IIC?the IIC speed?

Do you do a cross test between failure and good boards?