TCA9802EVM: VIH/VIL different with datasheet

These are the minimum and maximum guaranteed switching thresholds:

The datasheet guarantees that all devices switch high when the voltage reaches V_IH(min). But some devices will switch even earlier.

The actual switching thresholds can be anywhere below V_IH(min) and anywhere above V_IL(max). You must use the specified values to ensure that the input switches even in the worst case (over temperature and manufacturing variations).

This is not what the datasheet says.

V_IH is a range of voltages. V_IH is any voltage that is read as high. The guaranteed minimum V_IH is 70% of V_CC; the actual minimum V_IH might be lower.

See sections 4.5.5 and 4.5.6 of Understanding and Interpreting Standard-Logic Data Sheets.

Hi Wills,

I just wanted to check how you are measuring the VIH/VIL specs and I see no problem there. I do want to say that Clemens is right in his answer. The way that VIH/VIL is defined in I2C, and in our datasheets, are the guaranteed ranges for high-level input and low-level input voltages. 

0.7 * VCCA  < VIH < VCCA is the guaranteed range. Every time you input a voltage of 0.7 * VCCA = 0.7 * 1.8 = 1.26V, you will be guaranteed a output high voltage VOH across all TCA9802 devices. This does not mean that the device cannot switch at a lower point, just like you are seeing in the scope captures at around 0.9V. 

Same thing for VIL. 0 < VCCA < 0.3 * VCCA. A value of 0.54 V will always guarantee a LOW output voltage VOL every time. This does not capture that some devices may switch to an output LOW state before that i.e. ~ 0.7V per your scope captures. 

Some devices may switch an output low at 0.7V, some devices may switch at 0.55V, but a guarantee that it will always switch and output LOW is 0.54V or 30 % of VCCA. 

It is a bit backwards to think about, but this allows for a complete standard across all I2C devices. 

Please let me know if this makes sense. 

Regards,

Tyler

Hi Tommy,

Yes, these are the scopes that I am referring to. I understand that it seems that these transitions are happening outside of specifications listed in the datasheet, but this is not what the datasheet is saying. 

Datasheet is saying that the device will always recognize a high input signal VIH at 70% of VCC. The device will always recognize a low input signal VIL at 30% of VCC. 

This does not account for devices that may switch earlier! The key word here that I am using is "always." 

If you are running a system at VCC = 3.3V, then you always expect and are guaranteed a high-level switch output at input signals > 2.31V = 3.3V * 0.70. This does not mean that your device may switch output high earlier. In the customer case, we see between 1.5 V - 1.6 V. 

Regards,

Tyler

Wills,

Our datasheet spec is the guaranteed range. This does not mean that the device will not switch before the min/max value depending on if you are measuring VIH or VIL. 

The image here showing that input to TCA9802 around 1.4V pulls the bus LOW does not violate our datasheet. If you were to input a signal < 3.3 * 0.30 = 0.99 V, say 0.5V signal, and the device did not output a LOW signal, this would be in violation of our datasheet. The fact that it is switching at 1.4V VIL is fine. 

Regards,

Tyler

I²C signals are not logic signals. The M.2 I²C pins are connected to I²C devices on both sides, so they must use I²C voltages.

(Please note that the TCA980x does not allow pull-up resistors on its B side.)

Wills,

Please let me know if you have any additional questions.

Regards,

Tyler