In a recently published Design Idea, Jim Hagerman proposes a two-transistor circuit that claims to replace the LTC4300 hot-swappable, two-wire LTC4300 bus buffer ("Two-transistor circuit replaces IC," EDN, Feb 7, 2002, pg 104). I feel compelled to point out that the LTC4300 offers numerous advantages over this circuit for both buffering and hot-swapping the I²C bus.

The NPNs in Hagerman's circuit provide no capacitive buffering between the card and the backplane, which allows card capacitance to add directly to backplane capacitance. The LTC4300 isolates card capacitance from the backplane capacitance, making it possible to meet the 400-pF specification.

The LTC4300 further aids in meeting rise-time requirements by providing boost pull-up current circuits on all four SDA and SCL pins. This feature allows users to choose weaker pull-up resistors on the bus that reduce power consumption and increase noise margin while still meeting system rise-time requirements.

When hot-swapping the card, the Hagerman solution immediately connects the card bus to the backplane bus with no regard for the condition of either node. The LTC4300 monitors the backplane side for a Stop Bit or Bus Idle and the card side for logic-high states before connecting the buffers. This approach ensures that no data transaction is occurring during the instant of connection.

Although I do not doubt the attractiveness of Hagerman's idea, I felt it was important for your readers to appreciate the additional benefits of the LTC4300.