I've been working with a reasonably sized Chinese PCB fabricator, and we've had random problems with boards for the last few months.
I'm not an expert PCB layout guy, but we aren't talking about EMI issues here.
Here is my most recent issue:
On the trace circled in red from NetJ1_2 is a data signal that goes from an MCU loading tool to the MCU. I've had this set of plated holes for 6 months in a few revisions of a board.
Last month, I conducted a minor re-routing of this signal. All the dimensions / topology of the routing are well within fabrication limits / DFM checks in Altium and within any stated limits of the PCB fabricrator. This is a toy board with 0.2mm traces -- nothing earth shattering here.
After this minor re-routing, pictured below -- I get the boards and 3 of the 4 boards I am unable to "see" the MCU when I connect the loading tool to it. When I say "see", I mean the loading tool is not able to boot up and communicate with the MCU.
I desolder the boards with not-working MCUs, and solder some known good MCUs with software already loaded and the MCUs function with the existing preloaded software. But my loading tool still doesn't "see" the proven good MCU.
I know think the data input trace (circled in red) is defective. I do probe the trace with a multimeter and there is continuity between the plated hole and the pad.
Has anyone had a circumstance where a PCB trace has continuity but marginal functionality? Seems unlikely, but I'm not coming up with any other explanations...
Has anyone going down this rabbit hole and not get burned, going forward using a board house having issues like this? I've made some minor changes each time to a board to fix some flakey fabrication problem areas, but I'm worried I make 5000 of these boards and they have an issue.
[I've had another crazy fabrication issue with this board shop, a signal trace to an LED somehow picked up power on the board, and permanently kept the LED on! Don't know why -- cause there ain't anything touching in Altium...]