Post B4zZJcCsTsKYdmabrM by 0h00000000@ioc.exchange
(DIR) More posts by 0h00000000@ioc.exchange
(DIR) Post #B4yI7KhlpjJGVRdYQK by azonenberg@ioc.exchange
0 likes, 0 repeats
Given the reactions I'm getting to ReworkCTF stuff, I'm curious...What's the most complex PCB rework you've personally done? If you've done multilayer circuit edits, what kind of bodges and what tooling did you use?I've seen some pretty impressive work (and done some myself in the past) using handheld rotary tools, but IMO for truly surgical multilayer rework a 3-axis (or more) mill is the way forward.
(DIR) Post #B4yK632kDIcxLdxjN2 by stereo4x4@techhub.social
0 likes, 0 repeats
@pepper @damienmiller @azonenberg inner layer DDR clock lines swapped (wrong footprint of one component). Manual twisting of 0,x mm drill under microscope, using the index finger to hold the end and twisting it by thumb and middle finger. 4 holes total, and two twisted bodge wires to replace it.DDR RAM kind of works correctly with inverted clock, the shifted clock leads to interesting bit errors at the beginning/end of a read/write....and a huge big via, shorting out the input of a big FPGA board across all layers. For some reason it sneaked into the Gerber files but was not present in the design files, really weird (design was done in Mentor Expedition by someone else, I just did the fix on that prototype batch)
(DIR) Post #B4yK63b89S8J4I3By4 by azonenberg@ioc.exchange
0 likes, 0 repeats
@stereo4x4 @pepper @damienmiller via in gerbers but not design files is something I'd expect comparison against the d356 netlist to catch, unless your fab didn't do a proper cam review?
(DIR) Post #B4yKDfXDV9SMkpqjOi by azonenberg@ioc.exchange
0 likes, 0 repeats
@stereo4x4 @pepper @damienmiller and so you drilled out the traces to isolate the inner layer connections, then reconnected with a surface wire? Terminating at the vias?
(DIR) Post #B4yLj2KctQ6kOBTuzI by stereo4x4@techhub.social
0 likes, 0 repeats
@azonenberg @pepper @damienmiller yes. Drilled into the BGA only once as the depth mark was wrong. This was 15 years ago, so not sure what was available/done in terms of automated checks.
(DIR) Post #B4yXgeaNPLDZc4TkA4 by hennichodernich@radiosocial.de
0 likes, 0 repeats
@azonenberg I did one rework where I had to expose layer 2 under a QFN pad with a handheld dremel. It also involved filling the crater with 2-component glue and soldering the part back on.
(DIR) Post #B4ye3PoT686gBKJ5bk by cccpresser@chaos.social
0 likes, 0 repeats
@azonenberg I solved your reworkctf, and it was a pain without a mill.Digging out the inner layers with a knife and dremel was tedious and looked really ugly.
(DIR) Post #B4yvN9J3sQty03c1AG by tnt@chaos.social
0 likes, 0 repeats
@azonenberg I put "outer" only but technically I once cleared an internal short (plane to via) but it was just using a small drill bit right in the via and turning it just between my fingers to destroy the via and then use a bodge wire through the hole to reconnect top trace to bottom trace.I recently tried on a test board to attach a wire to a BGA ball through the board. I drilled by hand (holding drill bit between fingers) and hit the ball but couldn't solder a wire through the hole.
(DIR) Post #B4z5kKWfKR00zxfSBU by azonenberg@ioc.exchange
0 likes, 0 repeats
@tnt yeah the high aspect ratio is hard. This is why i prefer to do it like a fib cross section coming down in a stepped cavity so you have more room to work
(DIR) Post #B4z5vwSectbNq1KPtQ by azonenberg@ioc.exchange
0 likes, 0 repeats
@cccpresser i mean it's hard even with a mill, these are not easy reworks to do.
(DIR) Post #B4z5yK8DajC5pUuBaS by azonenberg@ioc.exchange
0 likes, 0 repeats
@cccpresser but with the right tools it's definitely a bit easier and definitely far more repeatable
(DIR) Post #B4z66PtM3haI0loP7A by azonenberg@ioc.exchange
0 likes, 0 repeats
@hennichodernich nice. Pad connected to wrong net?
(DIR) Post #B4z6QjbsSOH2xRKEnw by azonenberg@ioc.exchange
0 likes, 0 repeats
@tnt I've only attempted a back side bga ball connection once, on an xc7k70t-1fbg484c on a ten layer board, and was successful (the board is still in service today 2ish years later).That was 1mm pitch, i have a scaled down 0.8mm on the new reworkctf, so we'll see if i manage to pull it off.
(DIR) Post #B4zD20U7HMRljXqq8m by hennichodernich@radiosocial.de
0 likes, 0 repeats
@azonenberg IIRC we had to connect the 1V0 of an GbE PHY to an external rail because the built-in regulator wasn't working properly. I also vaguely recall bodging in a TPS74201 for the 1V0 rail. (It's been 6 years, IIRC.)
(DIR) Post #B4zFQ2u08sYk5cY6iG by AMS@infosec.exchange
0 likes, 0 repeats
@azonenberg Inner layer edit on a planar transformer. Hand drilled (25mil drill in a handle) soldered to internal layers to reverse a winding. Solved the common mode switching noise problem.Also built outer layer things of networks of 0402 and 0201 hanging off of one or two pads.Did the feeding a tinned end of a magnet wire three balls deep under a 1mm bga to hit a ball that was needed then hot aired to solder. Surprised that one worked.
(DIR) Post #B4zZJcCsTsKYdmabrM by 0h00000000@ioc.exchange
0 likes, 0 repeats
@azonenberg @tnt I would like to see how all this compares to a dentist doing a root canal - tiny drills, high aspect ratio - but target is moving, jerking, and moaning/yelling.
(DIR) Post #B4zZwMPL9fWew0tTlY by azonenberg@ioc.exchange
0 likes, 0 repeats
@0h00000000 @tnt Not if your anesthesiologist did a good job, lol.But I have seen the resemblance for sure, some of the tools I've used for this work are literally dentistry tools.Some tech youtube channel should give a dentist a PCB with some complex inner layer rework and see how well they do on it lol.