Issue link: https://iconnect007.uberflip.com/i/857644
August 2017 • The PCB Magazine 25 get out of a box that's fully annealed. The crys- tal structure is very different. Holden: In one of your talks, you said the pro- cess time is 105 minutes from load to unload. Stepinski: The whole factory takes four and a half hours to go around, but it is 105 minutes to get from plate through etch. Holden: I have been comaparing 105 minutes to the typical four weeks, which is the conven- tional way with all the queues and the move- ments and things like that. Goldman: I think the conventional way, plate through etch, from the beginning of a design, taking a panel, putting it in at the very begin- ning of the core drill to the exit, completed parts, is four and a half hours. That's what you should compare to your four weeks. Although, some companies have a special prototype line or area and prototype engineers whose job it is to speed fast-turn jobs through shop. Com- paring single-sided versus the double-sided, the single-sided turnaround must be shorter. Stepinski: Not much. You just don't have to drill and you don't plate it. If you measure the cycle time starting at plate, all you're getting rid of is plating steps, and that saves an hour total. It's three and a half hours I guess, something like that. Goldman: If a designer comes to you with a pro- totype design and plops it on there, in less than half a day he's got his board to test. That's pret- ty good. Holden: You could have a lot size of one, so you could do one panel, which would be the proto- type if he wants one or two boards. Goldman: Or, if he's not here, he zaps over all the information, and you could have his panel on his desk the following morning at the latest if he's not local. Stepinski: Sure, and we've done that. Holden: One of the global strategic goals we had at HP was from bill of materials and schematic to an assembled, tested prototype in five work- ing days. In other words, we'd take two days to design the board, two days to fabricate the board, and one day to assemble and test it. You give us a schematic and a bill of materials, and five days later we'll hand you a finished, assem- bled prototype. We got down to five and a half days, versus 12 or 16 weeks for everybody else. That comes through a lot of automation and a lot of standardization. Somebody was telling me, "It takes three weeks to get a solder paste stencil." I said, "We don't use solder paste sten- cils. We apply the solder while the board is still in the panel form, clean it, and then calendar it, and the solder paste is on the board when we ship it to the assembly guy. They don't do any solder pasting, they just apply flux and put down the part." Goldman: Of course, now you can inkjet that sol- der paste, and the stencil guys ought to be worried. Holden: One of the things that we found, even 40 years ago, when HP was doing 150 GHz test equipment was that we would monitor not just the thickness but the dielectric constant of in- coming laminate. Stepinski: That is the only thing that we're not checking, but we probably will down the road. Holden: Well, Alex, our time is up. Thank you for all the time you have spent with us. Thanks for the update and for answering our questions about your most intriguing facility. Goldman: Yes, it has been a pleasure and a priv- ilege to visit Whelen. PCB Figure 7: Schmoll Maschinen drill machines. WHELEN ENGINEERING, TWO YEARS LATER