[SOLVED] Low coverage from Tetramax

[Jennis]

I am getting the low coverage from TetraMax. There is no issue related to uncontrolleble clock and set/reset but the ATPG untestable faults are more, is this the issue?

Also from top-level coverage report , three sub-blocks having very low coverage (around 70%). These blocks are very small block and consists mostly the combinational logic, do I need to insert test point to those blocks? I tried to insert some test-points but it did not improve test coverage.

Could anybody suggest how can I fix the low covearge or what are the reasons for low coverage.

Thanks

 

[skal81]

I've experienced some low coverage problem with block having a lots of fixed values. In my case I had a programmable filter with a hard wired coefficient selection block. Because there was a lot of tie cells, the coverage was quite low (untestable). Fixed value could be a hint.
If that's the case it might be difficult to improve scan coverage (as for my case I implemented a direct test of the filter response to compensate the low coverage).

 

[Jennis]

Thanks skal81!

My design have some tie cells but I am not sure if that is the reason. My design having around 4k untestable faults which are coming due to primary constraints. Is it expected results. I am setting some ports to constant to make the design in scan-mode , but due to this various internal nodes getting fixed value 1/0.

Could anybody experienced this kind of issue (due to primary constraint setting) then please let me know any fixes.

Thanks

 

[skal81]

I forgot one point. Try to run the following command
report_fault -all > fault_report.flt
It will dump all the fault Tetramax detected and there class like this:
sa1 UT net1
sa1 NO net2
....
The second column is the class code, and it will indicated you the reason the net cannot be test. For example here UT is undetectable-tied and so on. You will find the meaning of the code in the manual or in the report given by the following command:
report_summaries

Using those data you can try to check what fault are not tested and the nets creating the problem.

 

[Jennis]

Thanks , I was using that command and found that most of the faults are due to primary constraints. I did some modification in RTL and now coverage is okay.