CS8900 Link Light/LED

Mar 05, 2005 1 Replies

Hello,



I've designed a little board with a Cirrus Logic CS8900A-CQ3 ethernet controller on it, in 16-bit mode. When I connect my board to a hub, I only get a link light on my board, but not on the hub. However, I am able to read 0x630E from the PacketPage 0x0C register as Cirrus says I should be able to do in the AN 205 application note.



Furthermore, using an external loopback connection (TX+ tied to RX+, TX- tied to RX-), I do not get a link light on my board and the LinkOK bit in the LineST register is clear (it was set when I was connecting to the hub). Note: I set the FDX bit in the TestCTL register to 1 for the loopback test.



I have checked that the 20MHz crystal I am using is producing an oscillation.



It seems that my TX+/- connection is bad, but I've double checked my schematic against those in the AN 83-3 application note and I have checked for bad solder joints. Cirrus says in their datasheet that I should get a link light on both ends with only power applied, so I don't think I need to set any registers to transmit data to get a link light on the hub. Has anyone discovered otherwise?



Any ideas what could be wrong? Any ideas for tests I should run to debug this problem? I have posted the schematic on

formatting link
The ethernet section is on the last page. Please note that I have made one revision since this version: TEST (pin 76 is now tied high. I had it in boundary scan mode :P).



On a related note: how do I probe with an analog two channel scope a twisted pair ethernet line? I connected one channel to Tx- the other to Tx+, set channel 2 to invert, set the channels to add, set both to



1V/div, and ran at 10ms/div. I also connected the common clips on each lead together. With only one test lead connected I see a 2.8Vpp 16ms period link pulse, but I see this on all 8 wires. With both test leads connected I get nothing. What am I doing wrong?

I would appreciate any help.



Cheers, Chris



Fyi, I fixed the problem. My SMT transformer had a bad solder joint on the Tx+ pad that was not visible with the eye, but was with the continuity checker.

Cheers, Chris

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required