TCA9517-Q1: TCA9517-Q1 related issues

1. This remark indicates that the chip did not fail the tests at any current up to 100 mA. Higher currents are not tested (see section 4 of JESD78B), so it is not implied that it ever fails. In any case, these test conditions exceed the absolute maximum ratings by a large margin, so it is likely that the chip burns out before it can latch up.

2. I²C uses open-drain signals, where the idle state is high, and any device can pull the bus low. A buffer likle the TCA9517-Q1 must pull its A output low when it detects a low voltage on the B side, and it must pull its B output low when it detects a low voltage on the A side. The TCA9517-Q1 has a voltage offset of more than V_ILC on its B-side pins, so that its own output signal is not detected as a low input. A hypothetical device that behaved like the TCA9617-Q1, but without the voltage offset, would not work for I²C because it would not be able to differentiate between its own output pulling low and some external device pulling low.

Hi Jeno

jeno.zhang said:

1. As shown in the following figure, how is the latch triggered? Can it be restored? If the latch is triggered, how can the chip electrical signal be determined?

JESD 78 Class II standard defines an amount of current that can be injected in the pins of the device without latch up occurring. The tests conducted are specific to the standard posted by JEDEC. 

jeno.zhang said:

2. As shown in the following figure, how to understand this sentence with the word 'lock up'？

Clemens already described this in his answer, but I'll explain as well. There exists buffer drivers from A-side to B-side and a buffer driver from B-side to A-side. When a low is detected externally on the A-side, the A-to-B driver drives the B-side low. In order to prevent the B-side from feeding the output back into the input (therefore generating a lock-up condition because then B-side to A-side buffer drives A-side low) there exists a static voltage offset that is recognized as a low to external slave devices, but not as a low unto the device itself. The higher voltage offset of VOLB helps to tell the device which side is driving a low. In contention situations where the A to B driver is outputting a low on the B-side, and a slave device on the B-side communicates a LOW, it must output a voltage below the VILC limit listed in the datasheet to then tell the TCA9517 that a signal is being driven in the reverse direction. This idea helps to create a buffer that can re-drive an I2C signal in both directions with out locking up the I2C bus.

Regards,

Tyler