08-20-2019 02:49 AM
I use ultrascale FPGA(xcku060) and generate GTH （GTHE3， UG576) for SATA PHY. All is ok in simulation , But in FPGA test , It seems something is wrong in RX channel.. Since the same error occured in SATA3, sata2 ,and sata1, now I want to resolve this issue in GEN1 firstly.
From the attched figure 1 , we can see the OOB stage is finished correctly and the ltssm enter ALIGN state( host and device send align primitive to the other side) , there is 8b10b decode error and disparity error occasionally( rxctrl1[3:0] / rxctrl3[3:0] output from GTH is not zero) , And when link layer ltssm goes into link_ready state,(some correct Align primitive is rxed and device begin to send SYNC primitive to host), then host send SYNC primitive to device too. But in RX channel of device , the 8b10b decode error and disparity error occured very frequently .....
Attachment screenshot 1 were captured from chipscope， and screenshot 2-6 is for the configuration of IP.
Why there is 8b10b and disparity error in rx channel ? is there configuration error when generating the IP or need I modify the parameter for the GTH core ?
Thanks a lot..
08-21-2019 07:16 AM
it might also depend from a poor signal quality.
Could it be possible for you to measure an IBERT eye, eventually tune the RX equalizer (LPM is used typically up to about 11 dB of insertion losses - maybe you could try a different cable, or a DFE equalizer)
Can you try without SSC initially?
08-21-2019 07:16 PM
Thanks a lot for your reply.
I have test with ibert , and the eye is ok and no bit error with far-end loopback test . And the parameter( LPM , SSC ,insertion losses = 14DB) for ibert is same with the sata phy project.
I have try to use DFE and without SSC , the issue is same .
Can you help to confirm if other parameter need to modify such as CDRCFG ?
And I have searched GTH used for sata PHY on the Internet, and there are two cases ( GTH for SATA PHY) and have same issue . They switched to use GTX at last . So I am not confirm is there any success case for GTH used as SATA PHY?