TXS0102: TXS0102

here is the schematic

The TXS has internal pull-up resistors (10 kΩ on each side), so external ones should not be necessary.

Furthermore, the TXS is a passive switch, so when a line is pulled low, the driver must the sink the current through all pull-up resistors on both A and B sides. If the host or the display has weak I²C drivers, then you might be necessary to reduce the current to achive the required low-level voltage.

Anyway, the DDC specification recommends a pull-up current of 3 mA, so the schematic looks OK. Can you show an oscilloscope trace of the failing DDC communication?

first one is Monitor attached, CH1= B side SDA, CH2= A side SDA;

second one is only cable attached, then measuring the far side of cable(in order to make sure the level has no big difference when thru cable)
CH1= B side SDA, CH2= A side SDA, CH3= DDC GND(connector pin17), CH4= SDA on far side of cable;

We don't have actual Monitor of SQ1 series(customer cannot provide), first picture is measuring with work normal Monitor.

Thanks for the explaining...

As for the the information from Customer that modified the low-level from 1.0V to 0.8V at Monitor side then display become normally. I think that should be the low voltage level is not low enough to pass the low operation condition of TXS0102. According to the specification of TXS0102, maximum of VIL is 0.15V and VOL is 0.4V. Is this the root cause for our symptom? 

Can you suggest the value of pull-up resistor to fight this issue? And, how can I measure it correctly whether low voltage level follow IIC specification(1.5V or 0.3xVDD) or not ?
I need to know the root cause, and the fine tuning have to base on the root cause.

The TXS does not really have a V_IL; that value is specified only because V_OL depends on it. (The TXS is a passive switch, so V_OL is V_IL plus the voltage drop over the pass transistor.) If the voltage at one side is 1.0 V, then the voltage at the other side will be at least that high, and a 1.8 V system will have problems with that.

Try without any external pull-up resistors. Check how much you can increase the capacitive load (i.e., how bad a cable could be) before rising edges become too slow.

The I²C specification demands that V_OL ≤ 0.4 V with 3 mA pull-up current. Apparently, that monitor's output drivers are too weak, or they used too-large series resistors.

The most robust solution for handling unknown crap on the external bus would be an I²C buffer like the TCA9509, but it's not available in a compatible package.

Sorry about reply so late,
And many thanks for the information...

We will have two solution for this issue, and will provide to Customer to have a test again, could you please advise the solution has any side effect or you have any good ideas?

1. pull-up resistor be removed.
2. level shifter IC change to PCA9306 and pull-up resistor change to 10Kohm.

1. Removing the external pull-up resistors will solve the low-level voltage problem. But if the capacitive load of the entire bus (PC plus cable plus monitor) is too high, rising edges will become slow. (I²C is slow to begin with, so this might not become an actual problem.)

2. Both the TXS0102 and the PCA9306 are passive switches, so they would have exactly the same issues with load current.

Can you explain the item1 again?

Does it mean we cannot just remove the pull-up resistor, need to base on rising edges is follow IIC specification to fine tune the volumes of pull-up resistor ?

Thanks Clemens..

Could you check on our schematic(or you need the layout) again that out TXS0102 designing is follow IIC specification or not ? 
Or, you need more condition to check? Since the low-level voltage is low enough with many Monitor but just our Customer's Monitor is not.

So, Our Customer have question about the level-shifter IC design. 

The TXS, by itself, is compliant with the I²C specification. But it is a passive switch, so the other devices also need to have I²C compliant drivers.

The I²C specification requires that V_OL ≤ 0.4 V with 3 mA pull-up current. Does your I²C master device have enough drive strength for this?

In IIC specification, the maximum allowable of low-level input voltage is 1.5V. 
It means the level still in LOW specification even Monitor driving down to 1.5V, Right ?

but, if the low-level voltage comes from Monitor driving is 1.5V, it will not be in low-level specification for 1.8V side(0.3xVDD).
You only mentioned VO, I think VI is also the key point for level shifter both of two side.