TCA9517-Q1: The method to control the VOL of B-side

The B-side outputs of the TCA9517-Q1 have a voltage offset of about 0.5 V; changing the pull-up resistors will not make the V_OL lower than that. If the MCU can accept this voltage, then you can swap A and B sides.

(I would recommend the TCA9803, but it is not available with automotive qualification.)

The offset is not affected by the pull-up resistors.

The SCL and SDA lines can be swapped (they are identical on the electrical level), but this would not help with your problem.

The only way to get a low V_OL for the image sensor is to connect it to the A side of the TCA9517-Q1.

Hi Yoshikazu,

I agree with Clemens, the offset VOLB appearing on the SDAB & SCLB is set to a typical offset voltage of 0.52V, either when IOL is draining 100uA or 6mA. This is due to the static voltage offset on the B-side, which is unchanged by the selection of pull-up resistance. 

Yoshikazu Kawasaki said:

1. They changed the pullup resistor only on B-side from 1.5kOhm to 6.2kOhm, but the VOL was still 0.56V.  Would you please tell me the reason why it is?

Static voltage offset of 0.52V is static and does not change even with adapting the pull-up resistor from 1.5k to 6.2k. My suggestion is the same as Clemens, can you flip A-side and B-side? 

I would suggest the TCA9509 which actually flips the static voltage offset, and has lower voltage offset, but tighter noise margin. It is not offered in Q1 auto qualification. Is this is hard requirement? 

Yoshikazu Kawasaki said:

2. They changed the PU resistors on both A-side and B-side to 6.2kOhm and it showed VOL on B-side 0.8V.  Would you please tell me the reason why it is?

I am not quite sure why the VOL on B-side tended to drop. Does the customer have a scope capture showing this? 

Yoshikazu Kawasaki said:

3. They'd like to swap SCL and SDA like SCLA/SCLB to I2C SDA and SDAA/SDAB to I2C SCL.  Is it possible?

I don't think this would change anything. SDAA + SCLA and SDAB + SCLB have the same driving capabilities. Only difference between the two outputs might be external such as PCB trace lengths, tolerance with pull-up resistance, difference in parasitic bus cap on the line, etc. 

Its possible, but it wouldn't change that the static voltage offset is still present on the B-side of the device. 

Yoshikazu Kawasaki said:

4. What would you suggest to control the VOL to meet VIL≦0.54V required by the image sensor?

In previous answers, controlling VOLB (either on SDAB or SCLB) is not possible. This is the reason that it is called a static voltage offset. The output voltage is only subject to change within the datasheet limits (0.45V < 0.52V < 0.6V) (min, typical, max). You might get a lower static VOLB if the VCCB voltage is lowered. 

Regards,

Tyler

Hi Yoshikazu,

Yoshikazu Kawasaki said:

Thank you very much for your comments.  The description shown on page 14 is a little bit confusing us.  It would be better to have additional comments like showing the pull-up resistors don't change the VOL that much or something.

The description is confusing because I do think it is not explicit to the customer telling them which side is being driven. The VOL on page 14 can be adjusted for when a slave/target device is connected to the B-side, and the downstream slave/target device is driving the B-side externally low hopefully to at least a VILC which is = 0.4V. The VOL from the slave device is dependent on the pull-up resistor. Increasing the value of the pull-up yields a weaker resistance, and therefore, the slave device can pull VOL further to ground 0.3V... 0.2V... etc. A stronger pull-up i.e. small value, would yield a higher VOL due to the fact that the open-drain driver within the slave device become to weak to overcome such a strong pull-up. 

VOL of a device on the B-side can be adjusted through the pull-up resistor. This is not talking about the VOL of the TCA9517-Q1, which cannot be adjusted. Hence why we call it a static voltage offset or SVO for short. 

Yoshikazu Kawasaki said:

VOL of a device can be adjusted by changing the IOL through the device which is set by the pull-up resistance value. The pull-up resistance on the B-side must be carefully selected to ensure that logic levels will be transferred correctly to the A-side.

"Of a device" not that of the TCA9517-Q1. 

Yoshikazu Kawasaki said:

I proposed swapping the A-side and B-side as you mentioned if they still use TCA9517-Q1.  I also proposed replacing it to another level shifter(TXS0102-Q1) or analog switch(SN74LVC2G66-Q1) instead.

The devices proposed here are level shifters and not true buffers, meaning they do not actively re-drive the I2C signal. Level translation is applicable here, but as for driving a heavily loaded I2C bus, the TXS or SN74 will not accomplish the same purpose as the TCA9517Q-1. 

It might be beneficial for the customer to look into the PCA9306-Q1 also as an optional level shifter Q1 graded. 

Regards,

Tyler