FPGA Intellectual Property
PCI Express*, Networking and Connectivity, Memory Interfaces, DSP IP, and Video IP
Announcements
Intel Support hours are Monday-Fridays, 8am-5pm PST, except Holidays. Thanks to our community members who provide support during our down time or before we get to your questions. We appreciate you!

Need Forum Guidance? Click here
Search our FPGA Knowledge Articles here.
5881 Discussions

Stratix 10 L tile Native PHY unable to receive incremental data sent from TX and tested using FMC loopback card

HBhat2
New Contributor II
478 Views

Hi,

I am using Stratix 10 SoC dev kit and testing the transceiver with FMC loopback card.

I tested with the Board test System and all are looking fine with BER = 0.

Even I opened the FMCA project available in the Dev kit example design and checked with transceiver toolkit for BER and it is showing Zero with some of the channels (GXT) running @ 25Gbps and GX channels running @ 17Gbps.

After that I created a simple project with Native PHY, ATX PLL and reset controller for one channel of Bank 1D (TX channel 0 and RX channel 0) . I am using the 644.53125MHz clock @ 1D bank reference clock pin from FMC loopback card. With this project Transceiver toolkit is locking and BER = 0. In the same project (PCS-PMA width is 64 bit with enhanced PCS or PCS direct) , I send 64-bit Incremental pattern and I expect the same to be looped back as I have inserted the FMC loopback card. But, I see only few incremental data I am able to see in RX parallel data bus. All other status looks fine. rx/tx_cal_busy = 0, rxlockedtodata = 1, tx_ready=1, rx_ready=1. rxlockedtoref is toggling randomly.

Even I checked with seriallpbken =1, with this none of the data in rx parallel data bus are matching with that of tx parallel data.

Can anybody suggest what settings migh be wrong in the PHY config.?

Attached the related .ip files for the review.

 

With Regards,

HPB

0 Kudos
1 Solution
Nathan_R_Intel
Employee
120 Views
Hie, Its good news that you can receive the signal correctly. Yes, since you are using PCS direct, there is no word aligner hardware available. Hence, you will need to implement word aligner in the fabric. As for meeting timing in Stratix 10, we have a document to guide you. https://www.intel.com/content/dam/www/programmable/us/en/pdfs/literature/hb/stratix-10/s10_hp_hb.pdf Please refer to section 2.3 and 2.4 mainly. Please let me know if you need any other information. Regards, Nathan

View solution in original post

9 Replies
Nathan_R_Intel
Employee
120 Views
Hie, Could you please attach your design achieve , .qar instead of just the zip version of the ip files. It will be difficult to analyze the fault with your design without the complete design achieve. Regards, Nathan I have received your design achieve
HBhat2
New Contributor II
120 Views

Hi @NathanR_Intel​ ,

 

Now I have uploaded the QAR file as well.

 

With regards,

HPB

Nathan_R_Intel
Employee
120 Views

Hie,

 

I checked the design achieve, I could not find any connection issues. But do let me test in hardware. Will update you in a day or two.

 

Regards,

Nathan

HBhat2
New Contributor II
120 Views

Hi @NathanR_Intel​ ,

 

Thank you for the update.

 

With Regards,

HPB

HBhat2
New Contributor II
120 Views

Hi @NathanR_Intel​ ,

 

Did you get a chance to test the design in the hardware?

 

With Regards,

HPB

Nathan_R_Intel
Employee
120 Views
Hie Hariprasad, My apologies for the delayed response as I had to take time to get the hardware and test it. I understand your failure. Currently your design cannot close timing, hence suspect this is causing the failure that tx_parallel_data is different from rx_parallel_data. I am trying to address the timing issue when change the tx_coreclkin and rx_coreclkin clocks. I will keep you posted. Regards, Nathan
HBhat2
New Contributor II
120 Views

Hi @NathanR_Intel​ ,

 

Thank you for the update.

 

We came to know that Enhanced PCS and PCS direct doesn't support word aligner by default. With this info. I changed the Native PHY config to Enhanced PCS and enabled RX bitslip feature and port.

Now, I am sending known pattern first and checking the reciever whether it is able to lock for the first time. My observation is it is not locking without bit slip. I am pulsating the bitslip port until rx aligns for the known pattern. Once bitslip is done, then I start transmitting the incremental data.

Also, one more change I have done in native phy is change the "RX adaptation mode" to adaptive from Manual.

 

With this, I am able to receive the data continuously @20.625Gbps without missing any data.

 

Please let me know if you have any suggestion on meeting the timing. It will help us with our final implementation using startix10 FPGA.

 

With Regards,

HPB

Nathan_R_Intel
Employee
121 Views
Hie, Its good news that you can receive the signal correctly. Yes, since you are using PCS direct, there is no word aligner hardware available. Hence, you will need to implement word aligner in the fabric. As for meeting timing in Stratix 10, we have a document to guide you. https://www.intel.com/content/dam/www/programmable/us/en/pdfs/literature/hb/stratix-10/s10_hp_hb.pdf Please refer to section 2.3 and 2.4 mainly. Please let me know if you need any other information. Regards, Nathan
HBhat2
New Contributor II
120 Views

Hi,

 

Thank you for your confirmation.

 

With regards,

HPB

Reply