PCA9306: VIH/VIL level on SCL1/SDA1 and SCL1/SDA1 on PCA9306

Hello Maruyama-san,

VIH/VIL specifications are typically defined for a digital input buffer, i.e., a circuit that would take some analog input voltage and resolve it into either a "logic high" or "logic low" state based on these threshold voltages.  The PCA9306 device's SDA/SCL IOs are not digital in nature, though, and instead act like an analog switch.  This device accomplishes level translation by conducting between the signals on sides "1" and "2" whenever the voltage on either side decreases below the VREF1 level.  For voltage levels of VREF1 or higher, the conduction is blocked - this allows for a larger high-level voltage to be pulled up on side 2.

Because of this design, the low-level voltage that is driven by the device on one side of the PCA9306 will also appear on the other side of the PCA9306.  (Note that the voltages on either side will be very close but may not be exactly the same - there will be an offset voltage introduced by the current flowing through the switch multiplied by the series resistance of the switch.)  The high-level voltages on each side are just set by the pull-up voltages used on each side (minus any voltage drop across the pull-up resistances due to pull-down leakage currents that may be present on the PCB).

Please let us know if this explanation is not clear to you.

Regards,
Max Robertson

> For voltage levels of VREF 1 or higher, the conduction is blocked-this allows for a large high-level voltage to be pulled up on side 2.

There was a description.
When a voltage higher than VREF1 is applied, the connections (1 side and 2 side) are blocked, so the output logic becomes HiZ and the output voltage becomes Hi under the influence of pull-up,
I understood that.

On the other hand, if the pull-up voltage on the SCL1 and SDA1 side is the same as Vref1, I think that it is impossible to make these input voltages literally higher than Vref1.
It is presumed that there is probably a lower limit value such that the voltage of SCL1 and SDA1 is “more than ○○% of VREF1” or “more than ○○ V”.
Could you tell me about the conditions?
(For example, if the voltage of SCL1 and SDA1 is 99% of Vref1, does it correspond to "the conduction is blocked"? Also at 98%? 90%? )

Hey Yoshihiko,

"It is presumed that there is probably a lower limit value such that the voltage of SCL1 and SDA1 is “more than ○○% of VREF1” or “more than ○○ V”.
Could you tell me about the conditions?
(For example, if the voltage of SCL1 and SDA1 is 99% of Vref1, does it correspond to "the conduction is blocked"? Also at 98%? 90%? )"

The further below you go under the Vref1 voltage, the more conduction occurs. I would say, if side 1 (SDA1/SCL1) goes below Vref1 by ~0.4V then we will see conduction occur (higher temps make this conduction occur sooner and lower Vref1s make conduction occur sooner). The lower you go, the stronger the conduction becomes. The conduction is a linear function of the difference between side 1 and Vref1 voltages. (Being below by 1V will be twice as strong as being below by 0.5V).

-Bobby

Hi Bobby!

● Confirmation 1
I understood this, is it correct?
---
Assuming that the voltage drop at 200 kohm is negligibly small, let Vref2 be 3.3V.
In order to make side 2 2.31 V, it is necessary for conduction (conduction) of 3.3 V-2.13 V = 0.99 V.

Conduction [V] = Vref1 [V] / side1 [V]
Conduction = 0.99
Vref1 = 1.8
Solve for side 1
side1 = 0.55

Thus, if side1 is greater than or equal to 0.55 V (at a certain temperature), it is guaranteed that side2 is greater than or equal to 2.31 V.
ー

● Confirmation 2
> higher temps make this conduction occur sooner
For example, how much does it change at 0 ° C, 25 ° C, and 50 ° C?
Five %? Ten%? Is it smaller?

"Conduction [V] = Vref1 [V] / side1 [V]"

The conduction voltage is not a percentage, it is a value that varies across temperature and doping levels. Typical I would say is around ~0.6V at room temperature though I would use 1V as a maximum value (most likely you won't see it this high).

"● Confirmation 1"

If you pulled side 1 to 0.55V then side 2 would be slightly larger than 0.55V (Maybe 0.6V, depends on the load current). 

"● Confirmation 2

> higher temps make this conduction occur sooner For example, how much does it change at 0 ° C, 25 ° C, and 50 ° C? Five %? Ten%? Is it smaller?"

This kind of data is not catalogued, If you need an estimate you could likely get this estimate by looking at a few diode Vforward versus Temperature. Google search tells me that it could be around 30% lower at very high temperatures (125C) but its starting point depends on the load current. Lower load currents provide a lower Vforward. graphs seem very linear.  At ~75C the percent change is likely around 15%. At 50C this change is likely around 7.5%. Again with a 25C change (referenced to 25C) then 0C should show a 7.5% increase from 25C. These would be my estimates based on general diode behavior.

-Bobby