Issue link: https://iconnect007.uberflip.com/i/1428512
NOVEMBER 2021 I PCB007 MAGAZINE 69 something like that), and see what kind of sen- sors are in the market. Talk to people and fig- ure out how to start incorporating into your equipment. Johnson: You mentioned that a lot of the Indus- try 4.0 adoption is happening in the large vol- ume facilities, but you've also proven recently that lot size one is possible using these same techniques. Stepinski: Lot size one really comes down to a different equipment set. When you're doing production scale, you have large amounts of data, and you can tune things in very tightly. With lot size one, t y p i c a l l y yo u h av e to augment with test vehicles and things like that. It depends on what lot size one means. Does lot size one mean it's the first time I've ever built something with these parameters? Or is lot size one similar, part of a product family, or a platform of products that you're historically building? It might be part number 78, but I already built 77 part numbers with similar con- ditions, and then you can correlate throughout the product family and know what your reci- pes are. Lot size one isn't necessarily lot size one. If it's the first time you've built a technology, that's really lot size one. en you can augment with test vehicles, so you must look at it a lit- tle bit differently, I think. If it's all similar prod- ucts and it's all within your DFM guidelines that you've qualified, it's not that hard. Matties: Isn't lot size one where you can cus- tomize each panel or board as it's going through, based on the sensors, so every board could be slightly different because of variation? Stepinski: Yes, but this is a little bit like quan- tum theory too, with the measurement of something compared to the ability to change, so you have to finish something to know the result, and then how do you change it at that point? is is quantum theory engineering, and that's the challenge with lot size one. With lot size one, feedback is not the right way. You must have feed- forward, and that's a different control meth- odology. M a t t i e s : T h e r e a r e some areas where you can do it, like regis- tration, or optimizing the hole through the board. Stepinski: Yes, correc- tions. Matties: at's proba- bly where we'd need to start? Stepinski: Usually the most cost-effective method isn't to change the artwork; it's to change the process variables. is is typically the most cost effective. But it requires a capa- ble process. You change the conveyor speed by 2.3% for this lot. But then you can't have the whole machine on one conveyor drive. You need to have it broken up into smaller con- veyor drives, so the next panel gets reduced by 3%. You don't have to run out the whole line to do it. You have to look at the equipment design to do this; vertical is obviously even better for these situations. Holden: Where does culture and the tipping point intersect?