Jtag problem for Virtex II pro (XC2VP20-6FF896C).

I have a board using XC2VP20-6FF896c and two xc18v04 proms, seems Jtag
have problem as:
  1) Whenever there is a free-runnimg clock, the Jtag config fails.
when the free-runnimg clock is turned off, config successfully.
  2) Maseter seral mode never work, check cclk, it is alway there.

 Could clock interference affect Jtag configuration? but I use the
same design with virtex E, there is no problem.
 Appreciate any help on this problem.

anttonhu@gmail.com (tripledirrble) wrote in message news:<632786ca.0411051459.77f7326@posting.google.com>...
> I have a board using XC2VP20-6FF896c and two xc18v04 proms, seems Jtag
> have problem as:
>   1) Whenever there is a free-runnimg clock, the Jtag config fails.
> when the free-runnimg clock is turned off, config successfully.

I've seen EMI cause JTAG to fail configuration when using "flying
leads".  Making the leads shorter and twisting a ground with each
signal helped.

>   2) Maseter seral mode never work, check cclk, it is alway there.

Did the old Virtex E configuration have only one PROM.  IIRC, in a
multi-PROM configuration setting, the one electrically closest to the
FPGA is the mcs file with a '0' in the name.

> 
>  Could clock interference affect Jtag configuration? but I use the
> same design with virtex E, there is no problem.
>  Appreciate any help on this problem.

Yes, the old Virtex E configuration have only one PROM. what do you mean by "and twisting a ground with each signal helped"? Is that true to have a ground trace in parrallel with each signal?

tripledirrble <anttonhu@gmail.com> wrote in message news:<ee89f0f.1@webx.sUN8CHnE>...
> Yes, the old Virtex E configuration have only one PROM. what do you mean by "and twisting a ground with each signal helped"? Is that true to have a ground trace in parrallel with each signal?

I believe that magnetic coupling is a function of the loop area
between the signal and a AC ground.  If you have a wire hanging out in
free space, it is my belief that it generates a lot of garbage and is
susceptible to such.  I believe the theory behind the twist is that it
will reduce the loop area to a given magnetic field orientation.  I
believe E-field interference is reduced by a shield. Do not take my
explanation of this too seriously.

Newman