The TLK2501 data sheet has this to say about the LCKREFN pin:
Lock to reference. When LCKREFN is low, the receiver clock is frequency locked to GTX_CLK. This
places the device in a transmit only mode since the receiver is not tracking the data. When LCKREFN is
asserted low, the receive data bus terminals, RXD[0:15], RX_CLK and RX_ER, RX_DV/LOS are in a
high-impedance state.
When LCKREFN is deasserted high, the receiver is locked to the received data stream and must receive
valid codes from the synchronization state machine before the transmitter is enabled.
And later on, there is a simpler explanation of the use of this pin:
The TLK2501 allows users to implement redundant ports by connecting receive data bus terminals from two
TLK2501 devices together. Asserting the LCKREFN to go to a low state causes the receive data bus terminals,
RXD[0:15], RX_CLK and RX_ER, RX_DV/LOS to go to a high-impedance state. This places the device in a
transmit-only mode since the receiver is not tracking the data.
This latter explanation seems benign. Basically it says that pulling the pin low disables the receiver. But the pin description above says that the transmitter is disabled until the receiver indicates that it is synchronized.
How can this possibly work? It would seem as if the 2501 could never work in loopback mode, as the receiver is waiting for valid IDLE or carrier-extend characters, which never come because the transmitter is off. Likewise, in non-loopback mode, it is dependent on an external transmitter to be active sending characters or IDLEs before the receiver syncs and then enables the transmitter, but if there are 2501s on both ends of the medium, then you're back to the chicken and egg.
Thoughts? What am I missing?
