Issue link: https://iconnect007.uberflip.com/i/1305670
NOVEMBER 2020 I SMT007 MAGAZINE 43 Webb: Yes. It can be set up as flags to opera- tors and send an email to production people to say this is what's now going on in real-time. It could be either one or both. I also want to cover defect parts per million opportunities (DPMO). We were talking about the coverage calcula- tions earlier. We use the PPVS metric, but we also balance that by the DPMO. The value of doing that is you really want to have a good test coverage where you have the opportunity for defects, and if you have great test coverage on something that never fails, it doesn't really help you. Having an understanding on where your defects are is very important to the whole pro- cess of making sure your coverage is in the right place. Part of the database we talked about is collecting DPMO information. And because for a lot of people these days, even a lot of our sophisticated customers, you start introducing this concept to them and they don't have a good way to get their own DPMO. Now, in our tool, there are some built-in models you can use for people who don't have their own DPMO. We suspect as people shift more to Indus- try 4.0 concepts and collect big data, then they should be able to collect and make better use of those DPMO figures. You can feed that back into the tool now, and you get more accurate coverage results because now you know where your defects actually lie. Johnson: That gives you better results to pass upstream for better process optimization to begin with. Fewer errors escape down-process, and everything starts to cycle upward with increased yields. Webb: Yes, in an upward direction. Johnson: Will, in this environment, what are some of your customer's biggest challenges? What are some of the key issues pushing them to decide that this is a solution they need? Webb: There can be a couple of ways they decide that they need to have some sort of a tool like ours. One is, as I mentioned before, identifying that they're going through board spins. They're not identifying the key items in a timely manner. That can be a flag that, "We need some sort of a tool to automate this pro- cess so that we can do it every time the same way. If somebody is gone or somebody's left the company, we still have our built-in rule set that we can run in an automated fashion that's going to look at the things that we should be looking at for testability." Equally as important, they need to have a way to produce test coverage numbers. A lot of times, we find that some customers of ours— and I'm sure Bert can identify with this—oper- ate a little bit in silos. You might have the ICT department, but they don't talk to the AOI engineers or X-ray people at all. Everybody is focused very much on what they're doing, so trying to pull together two different teams to look at the combined coverage or looking at reducing overlap, etc., becomes very challeng- ing from a management perspective. The tool can really help that happen and start pulling in data from all the various teams. Somebody then becomes a coverage engineer or DFT engineer who has good visibility to the whole manufacturing test process and can start to understand what we've been saying. They can look at gaps that are missing and see what must be covered at functional coverage because we're not able to cover it with struc- tural tests up until that point. They can also look at the redundancies that could be elimi- nated as a cost-saving. It really ties together the manufacturing floor, and it can tie in design with test very nicely, which sometimes doesn't take this either, so the tool can really help bring those two groups to a place where they're talking more. Johnson: Bert, as you're building your busi- ness and serving your customers, providing tests and services, you're not on the man- ufacturing floor. What are some of the chal- lenges that this software solution has helped you with? I'm thinking mostly about feedback rather than test optimization. Horner: We do not know what is upstream as far as inspection, but there is a lot of redun-