Resistors:
500k is a bit unusual, and the ubiquitous 470k and 560k may be too far off. With two common resistors one should get close enough (no warranty, though!): 470k + 27k = 497k, that’s only 2k off.
(Open question: Are other resistors in the same ratio OK, or are exactly the given absolute resistances necessary?)
Form types and placement:
Diode and resistors in THT type are much easier to solder to the tiny contacts on the board than SMDs. Their long thin wires are easier to handle, and also can be bent to make a larger contact area and to get less in the way of the solder iron tip.
Diode and voltage divider can be “free floating”, i.e. connected at all points through pieces of insulated wire. Fix them with a piece of tape against rattling loose and straining their soldering during transport.
Insulation
For insulation of unprotected wires and part contacts, use heat shrink tube, as an alternative take Kapton tape. Do not use electrician’s tape, its glue will become gooey and less sticky with time. Other tapes may or may and be heat resistant enough, but better safe than sorry.
I think Framework or someone else should offer kits with all the materials for a few bucks. If Framework did so, this would be an enormous boost of their ‘repairability’ reputation, and a great PR coup when contrasted with Apple’s shenanigans in that department. The longer they wait the lesser the impact.
I’m doing my homework to proceed with this change but where exactly does that Schottky Diode goes into? I don’t seem to find any reference to it in the instructions… Have I miss something?
As someone who has designed PCBs, this sounds really weird. Why was this resistor even there if it was left unpopulated by default but still ‘shorted’ by the PCB?
@amoun of course there is always components that are different between initial and mass production. What I’m getting at that the situation is weird even on its own. Because it seems to be like this:
1--|-resistor pads-|--2
|---------------|
Which means that a connection between 1 and 2 is always there, the presence of the resistor on the pads can’t have any effect on the circuit.
these are mostly for current sense resistors that we populate during pre production builds. They are replaced with a short jumper for mass production so we do not have to spend time populating 0 ohm resistors.
Can confirm this works very well. Reworked Gen11 boots with attached AC, main battery, or both; all with or without RTC cell.
Some thoughts and observations about the work
There is not sufficient room for through-hole parts above or below the mainboard. At its side is enough space. The images show the voltage divider build as (470k+27k)/150k.
The wires are litz, which is actually too flimsy and I now strongly recommend against. Even a light touch with the solder iron tip squeeze-spreads the strands enough to reach a neighboring contact and cause a short. Better use solid wire.
Step 2
That was a bit of a thrill. I had to cut blind, tilting the blade, because PU301 was placed so that its corner somewhat overlapped the cut area.
(That makes it a good idea to provide a remedy instruction for a damaged via area: Are there points on the board that could be connected to restore contact if the via area contact was fully broken, if necessary routing a wire top side to bottom side?)
Step 5
The tiny solder point under the printed “R” offers very little mechanical strength. I chose to scratch off a bit of the coating over the broad diagonal lead for a larger soldering area. If you do so, take care to not expose the neighboring leads.
Step 6
For the diode I used an SMD type like this, mounted piggy-back on DC1 with the A1/K2 contact directly on the screw-ward contact of DC1.
Strangely, the Manjaro KDE Plasma System Monitor displays negative values for charge rate both when charging and discharging. I had never before looked there, so this may well be a preexisting KDE Plasma bug.
Some Manjaroan here with an unmodified Gen11 board who could check? And ideally also after the rework. That would hopefully rule out a possible unwanted side effect of the rework (and leave me hysteric over where I have snafu’d…). Thanks!
Litz wires are terrible for stuff like this. You must be very skilled if you successfully managed to solder that.
I recommend fine magnet wire. Around 0.2mm diameter should be relatively easy to work with. (This will be small enough to target small pads and large enough to avoid accidental wire breakage and other flimsiness)
after seeing someone else attempt this, I’ll definitely try this once I get my new AMD mainboard! I’ve got a pretty nice solder station at work that I can borrow after hours, so it’ll be a fun project to try with my intermediate soldering skills!
Apparently not skilled enough to think of solid wire beforehand.
(OTOH, I once folded an eikon usb fingerprint reader board in half, to fit into a Dell Mini 9… And if @Kieran_Levin did the work with SMDs only without a binocular or microscope, and a micromanipulator, I can really appreciate that. RESPECT, SIR!)
(Maybe we should start a Brag’n Swag thread?)
If you use solid wire, bend it so that it fits the to-solder position and fix it with a bit of tape. That leaves you a free hand for the solder iron (I recommend the cold end ) and the other for small pliers to hold and guide its tip. Otherwise, even a small tremor will drain your patience very fast.
Also, contra the usual practice, prepare a tiny pearl of solder on pad and wire, so that you don’t have to feed solder while holding the hot tip to the site. All that is needed then is a brief touch of heat to merge the pearls.
I came up with a simplified rework for this which should be much easier to do, and only requires 2 parts. All rework is on the top side, and no traces need to be cut.
It only requires a small 3.3V LDO and diode. I will post more detailed instructions and BOM once I get the parts.
Wow! Love to see that we already have a simpler version of this (how convenient that it came out before I attempted the more difficult version) Now I’m even more excited to attempt the fix!
