Part Number: TLK10232
Hello,
I have a board with TLK10232 connected to CX4 connector for XAUI transfer on the LS side and to SFP+ on the HS side.
I'm using BIDI SFP+.
I connect two boards together using the optical link and perform electrical loopback on the XAUI interface.
I configured the TLK10232 as follows using the "low level configuration" option in the GUI:
- Reset device (write a 1 to 0x1E.0000 bit 15 or assert RESET_N pin)
- Make sure the reference clock selection (156.25 MHz or 312.5 MHz) is correct – this is done through register 0x1E.001D bit 12 (default is 156.25 MHz)
- Disable auto-negotiation by writing 1’b0 to 0x07.0000 bit 12
- Disable link training by writing 16’h0000 to 0x01.0096
- Write 16’h03FF to 0x1E.8020. This allows the link settings that would normally be configured through KR training to be configured manually instead.
- Depending on the link conditions, you may need to change the default configuration of 0x1E.0003 and 0x1E.0004. For optical connections, we typically recommend changing HS_ENTRACK (0x1E.0004 bit 15) to 1’b1 and HS_EQPRE (0x1E.0004 bits 14:12) to 3’b101. This can be a starting point, but you may need to do some BER testing to optimize the values.
- Issue a data path reset by writing 1’b1 to 0x1E.000E bit 3.
The link is established ok.
After few minutes to an hour I see that one board reports TX_FIFO_UNDERFLOW (0x1E.000F bit 7) while the mate board report RX_FIFO_OVERFLOW (0x1E.000F bit 4). And TX_FAULT (0x1.0008 bit 11)
In some cases, it is TX_FIFO_ OVERFLOW on one board and RX_FIFO_ UNDERFLOW on the mate.
The problem is that failure persists. The bits do not clean after reading. It clears only after I perform data path reset by writing 1’b1 to 0x1E.000E bit 3.
The problem occurs only with BIDI SFP+ units, while on duplex SFP+ it doesn’t happen.
When checking PRBS errors on the HS side I see errors on the BIDI SFPs but no errors on the duplex SFP+. BTW the XAUI interfaces are always error free.
On one hand the problem is clearly the SFP+ BIDI problem, on the other hand the SFP+ BIDI BER is ~6.7e-14 (checked over 24 hours with the PRBS test) which is better than the industry standard of 1e-12. And I’m afraid that it will happen with other SFP+ modules
Is there a way recover from the FIFO failure without data path reset?
When performing data path reset during normal operation there is lots of traffic loss.
Regards,
Yossi
