Tetramax: using patterns from external BIST module

bist module scan test

Hi

I'm developing a circuit for evaluation of fault models and bist logic (master thesis). So far I have made a circuit consisting of one CUT (circuit under test), BIST logic and clock controller.

The CUT is run through DFT compiler and scan is inserted into it. FC is 99,8% when I use Tetramax as ATPG.

Now I want to use the vectors generated from my bist logic instead (LBIST/STUMPS), and get tetramax to evaluate the FC of these vectors.

My question is how can I do this?

I tried to just make a EVCD dump from the CUT when running a BIST simulation but can't get tetramax to understand what I want it to do.

My best guess now is that I have to run a BIST simulation and write the scan inputs (vectors) and outputs (responses) to a file and then write a program that makes a verilog pattern file similar to those tetramax makes...

Suggestions?

simulation evcd

Hi,

As far as i know the patterns of the BIST will be loaded from the output of the PRPG and captured on the MISR. There is no role of the PI/PO in the BIST. And when doing BIST it is a recommendation that all the PIs and POs are registered to get more coverage.

Now when you simulate your BIST patterns and generate the evcd there will be nothing changing on these PIs and POs and hence nothing is captured in the EVCD.
Then how do you think it can be used in TetraMAX.



-cheers
vlsi_eda_guy

well, I read something about tetramax and evcd...

but, never mind

here is what I did:

wrote a testbench and instantiated the CUT (with scan chain) and BIST logic in it. Added dummy scan chains, which was connected to the PRPG, which was used to store the test vectors. the actual scan chains in the CUT was also connected to the PRPG. Dummy scan chains was also connected to the scan outputs of the CUT, and was used to store the test responses. PIs and POs was also stored.

tetramax was used for atpg, and the vectors saved. then I studied the verilog testbench generated by tetramax and saved my vectors in the same format, then used tetramax to read them and run fault_sim...

This approach works, but simulation of the entire CUT take some time, and since I change a lot to study different scenarios, errors (that i made) did occur quite often... well well... probably a much better way to do this, but I dont know anyone that can tell me how...

anyway, just 3 weeks left now, and I have finished all my simulations (I hope)...