cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
ppaysarvi
Visitor
Visitor
10,458 Views
Registered: ‎03-22-2011

MMCM problem in Virtex 6 - unexpected output phase in behavioral simulation

Hi,

 

I have a problem with the output phase of the MMCM (DCM) in simulation. I am simulating the very simple scheme of "clock deskew"using MMCM. Based on UG, the CLKIN1 and CLKFBIN must be aligned but, in my simulation, they are not aligned and CLKFBIN is some amount of time behind CLKIN1. I ran the simulation several times and looks like the time difference between CLKIN1 and CLKFBIN is exactly equivalent to the VCO clock period. BTW, I tried both the "zhold" and "internal" compensations but the results are the same. Would you please let me know why am I seeing this phase difference?

 

Thanks,

MMCM_phase.JPG
0 Kudos
10 Replies
sandrao
Community Manager
Community Manager
10,434 Views
Registered: ‎08-08-2007

There is an bug in the behavourial simulation model in 12.4 that caused the outputs to be misaligned. This was fixed in 13.1.

What version of SW are you using?

Thanks,

Sandy


------------------------------------------------------------------------------------------------

Don’t forget to reply, kudo, and accept as solution.

If starting with Versal take a look at our Versal Design Process Hub , Versal Blogs and the Versal Useful Resources .

------------------------------------------------------------------------------------------------
0 Kudos
ppaysarvi
Visitor
Visitor
10,429 Views
Registered: ‎03-22-2011

I am using the version 12.4. So the simulator has a problem not the block itself? Will I still see the phase difference in real implementation using 12.4 version?

0 Kudos
sandrao
Community Manager
Community Manager
10,426 Views
Registered: ‎08-08-2007

The issue was only with the behavourial simulation the hardware would function as expected.

Thanks,

Sandy


------------------------------------------------------------------------------------------------

Don’t forget to reply, kudo, and accept as solution.

If starting with Versal take a look at our Versal Design Process Hub , Versal Blogs and the Versal Useful Resources .

------------------------------------------------------------------------------------------------
0 Kudos
djiss
Visitor
Visitor
10,242 Views
Registered: ‎07-04-2011

Hi !

 

I've the same problem as ppaysarvi. Output clock phases of MMCM_ADV Xilinx primitive are not as expected in behavioral and post-route simulations. I've followed your advice and I've tried with ISE 13.1but this problem remains the same. I've verified in implementing on hardware the MMCM_ADV and output clocks are correct. 

 

It seems that ISE 13.1 doesn't correct this bug. Anyone would have a solution to solve this problem ?

 

Thanks for your help,

0 Kudos
rcingham
Teacher
Teacher
10,208 Views
Registered: ‎09-09-2010

Did you regenerate the IP with the new version of CoreGen?

------------------------------------------
"If it don't work in simulation, it won't work on the board."
0 Kudos
djiss
Visitor
Visitor
10,189 Views
Registered: ‎07-04-2011

Thanks for your answer rcingham,

 

I use the MMCM_ADV primitive. But I've tested the use of Clocking Wizard in the CoreGen available with ISE 13.1 but the problem is similar. In functionnal and post-route simulation, output frequencies are correct but output phases (for example, I expect a 90° phase on a CLKOUT3 but it generates a in-phase CLKOUT3). I've tested this IP generated by Clocking Wizard CoreGen on a board and all output phases are as expected !

 

This issue is known by Xilinx for ISE 12.4. Would it be a similar bug with ISE 13.1 ?

0 Kudos
rcingham
Teacher
Teacher
10,184 Views
Registered: ‎09-09-2010

ISE 13.2 now available for download...
The Release Notes might hint at whether this is a known/fixed bug.

------------------------------------------
"If it don't work in simulation, it won't work on the board."
0 Kudos
djiss
Visitor
Visitor
10,055 Views
Registered: ‎07-04-2011

Hi,

 

This problem of unexpected phase was solved in using ISim instead of ModelSim for simulations. It was not an ISE problem but a bug in ModelSim.

 

 

0 Kudos
rcingham
Teacher
Teacher
10,053 Views
Registered: ‎09-09-2010

"It was not an ISE problem but a bug in ModelSim."

Or possibly a bug in the model that ModelSim picks up but that Isim doesn't?

BTW, I don't work for Mentor, and I do know that ModelSim does have some bugs, as detailed in their Release Notes.


------------------------------------------
"If it don't work in simulation, it won't work on the board."
0 Kudos
ebaum
Observer
Observer
2,684 Views
Registered: ‎08-07-2009

Are there any news related to this issue?

I'm using ISE 14.5 and have the same problem.

 

0 Kudos