SignalTap problems with large signal-count

--- Quote Start ---

Hi,

I am using SignalTap (9.1) to debug my design. When I have about 100 signals, everything works as expected. However, when I use much more signals, let's say 200, the behavior get's incorrect: The data is obviously incorrect (i.e. more or less complete non-sense while the design is still working like before), and/or the triggering fails, even the time-scale is incorrect (e.g. from 0 to 3072 instead of -1024 to 3072, or -2048 to 2048, althought "Pre trigger-position" is selected).

Other settings (beside signal-number) like sample depth appear to have no influence.

Has anyone observed the same behavior, or is this something specific to my design / system?

Thanks,

Thomas

--- Quote End ---

Hi,

did you get timing violation for the signaltap part of your design ?

Kind regards

GPK

Good point. I already checked the "normal" timing of the data that is recorded, but after checking carefully, I realized that there are HOLD-time-violations on altera_reserved_tck. Maybe, this should be a global signal (all of them are already in use). I will check if I can free one global signal for debugging and see if this helps...

Thanks,

Thomas

The global clock-thing did fix it. :) However, it took me some time to find out that the signal altera_internal_jtag~TCKUTAP (which cannot be found with the node-finder) has to be constrained with the global clock-assignment...

Regards,

Thomas

This option was already on. The problem with my design with the huge number of different clocks that made use of the global clocks and left nothing over for SignalTap.

The assignment "altera_internal_jtag~TCKUTAP | Global Signal | Global Clock" did the trick...

Thomas