Article View: comp.arch.fpga
Article #43R: DFT , Design For Test HELPPPPP
From: Alessandro Capobianco
Date: Sun, 03 Nov 2002 19:06
Date: Sun, 03 Nov 2002 19:06
50 lines
4213 bytes
4213 bytes
Hi Djohn, the DFT strategies are used to allow testing components after production. Sometimes the process is somehow faulty, so the silicon vendor must be able to discard faulty pieces and investigate the reason. One of the most used DFT strategies used for big IC (1 Mgates) is the scan chain. The design is synthesyzed using scanable flip-flops. They have one more input called Test Input TI and a selector called Test Enable. When TE is active (test mode), the FF uses TI as input, else the conventional input. TI is connected to the ouput of another FF and all TE are connected toghether. Thus all the FFs are connected to make a long chain (the Scan Chain) and selecting TE you can scan-in a special pattern and check the result (pattern+result=Test Vectors). For example: put the circuit in test mode (the scan chain is like a long shift register); scan in a pattern; put it in normal mode and run 1 clock cicle (the FF will now load new values according to the logic between them); put the circuit in test mode; scan out the result (the result will depend on the design itself). You can simulate this test vectors and see the behaviour. If the behaviour is OK you will release these Test Vectors to the silicon vendor who will run them on the chips soon after the production. If the outputs are as expected, the pieces will be delivered to the customer, otherwise they will be trashed. The faulty results can be analyzed to find where the circuit is faulty. ATPG (Automatic Test Pattern Generator) is the tool which takes as an input the circuit netlist, containing the scan chain, and generates the input patterns (and I guess the result). The vendor has a test machine where the components are placed (it's a sort of pattern generator and logical anayzer togheter). The input patterns are scaned-in the chip and the results are compared to the expected results. BIST is another story. Built In Self Test is a test usually designed for memories. Memories are likely to be faulty, so they need special test algorithms to be run. The BIST block implements this specail algorithms (March-...) and is usually provived togheter with the memory block. During the normal mode BIST is quiet. Usually it is connected to the TAP. The TAP should be designed so that you can issue an instruction to start the BIST and check the result. SO, if you have memories in your UART controller, you can implement it, otherwise there's no point. But I suppose if you had had a memory, you would have known about the BIST. Thats all. I am a HW designer from Italy and work for Ericsson, I'm 32 and my experience is not so big. What do you do ? Student ? Well, I hope I clarified you doubths. Regards Alessandro
Technical Details
Message-ID:
<etex9.111498$%M1.2879622@twister2.libero.it>Path:
arc1-ca.pugleaf.net!archiver1.google.com!news1.google.com!newsfeed.stanford.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!nntp.infostrada.it!twister2.libero.it.POSTED!not-for-mailReferences:
<anbej7$cg0fi$1@ID-159866.news.dfncis.de>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 112a73,28297834f289370c X-Google-Attributes: gid112a73,public X-Google-Thread: fec08,28297834f289370c X-Google-Attributes: gidfec08,public X-Google-Thread: 112e4c,28297834f289370c X-Google-Attributes: gid112e4c,public X-Google-Thread: f736a,28297834f289370c X-Google-Attributes: gidf736a,public X-Google-Thread: 1089ad,28297834f289370c X-Google-Attributes: gid1089ad,public X-Google-ArrivalTime: 2002-11-03 11:06:51 PST Path: arc1-ca.pugleaf.net!archiver1.google.com!news1.google.com!newsfeed.stanford.edu!news-spur1.maxwell.syr.edu!news.maxwell.syr.edu!nntp.infostrada.it!twister2.libero.it.POSTED!not-for-mail Reply-To: "Alessandro Capobianco" <alexacapob@libero.it> From: "Alessandro Capobianco" <alexacapob@libero.it> Newsgroups: comp.arch.fpga,comp.cad.synthesis,comp.lang.verilog,comp.lang.vhdl,fj.comp.lang.vhdl References: <anbej7$cg0fi$1@ID-159866.news.dfncis.de> Subject: R: DFT , Design For Test HELPPPPP Lines: 53 X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 5.00.2615.200 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2615.200 Message-ID: <etex9.111498$%M1.2879622@twister2.libero.it> Date: Sun, 03 Nov 2002 19:06:50 GMT NNTP-Posting-Host: 62.98.210.95 X-Complaints-To: abuse@inwind.it X-Trace: twister2.libero.it 1036350410 62.98.210.95 (Sun, 03 Nov 2002 20:06:50 MET) NNTP-Posting-Date: Sun, 03 Nov 2002 20:06:50 MET Organization: [Infostrada]