Layout screwup

For a small pile of money, an eight layer board can be had in three days. For a much smaller pile, a week. Correcting the layout for a major oops like that may take a couple of weeks, depending, but it has to be done anyway. I'd do a spin just to fix that mess but the boss may have other ideas.

BTW, the problem I had a few weeks ago with the band gap reference not connected on the ARM M7 was fixed with a wire under the BGA to the pad. The CM ended up doing the work successfully on all but one board. If I have any more boards built (I have 30), I'll spin it to pick up that problem and I haven't found any other problems.

we had one big fu years ago, a ~700 pin BGA with a reversed footprint

getting a new board made is not the only things that takes time, first have to redo the layout that is no small task, they you have to find parts for the new build and get time at a place that can do the assembly

-Lasse

Den tirsdag den 26. januar 2016 kl. 00.39.32 UTC+1 skrev Tim Wescott:

many years ago some usb controller I think it was tricked us twice

it came in two different packages, but they had different pin numbers so if you just changed the footprint and used the same schematic symbol the pinout was wrong

-Lasse

I always have enough of the cheaper parts on hand to build at least twice what I need. The more expensive parts are usually samples. I can usually get them replaced in a couple of weeks. However, as you point out, layout can be a big deal, particularly if there are DDR interfaces and such. Even given the above, a new set of boards will cost around $15K, which won't make the counters of beans happy.

I did a layout and prototype build with 74HC DIPs in a row in one area. The space between their ends was more than enough to fit a bypass cap.

Until production, that is - the DIPs sourced for the prod run were ex another manufacturer, and the gap was nowhere near wide enough. Dremel saved that run.