Custom "75%+1" layout with "Danger Zone" Caps & Dyed Gateron MX Tops

[richfiles]

Keyboard75_MagSafe20Compare.jpg (624.34 KiB) Viewed 1841 times

Good news and bad news... The good news, is one of my two orders of 20-pin magnetic USB-C connectors came in, and amazingly, it arrived in a mere 6 days! I'm shocked by the shipping time! As you can see above, it's barely any larger than the old connector, which quite frankly, is amazing, considering it has 4x the density of connections! The bad news... I ordered two right angle connectors for me to use on my phone and Switch, and those indeed showed up. The problem is that the connectors I ordered for the keyboard are still in transit, and are clearly being shipped by a slower method. I could tear one of these right angle ones apart, but that kinda defeats the purpose of having the two right angle connectors for my two devices.

Keyboard75_MagSafePressedOutA.jpg (501.96 KiB) Viewed 1841 times

Anyway, the connectors I plan to take apart are technically still on their way, and once they arrive, I'll tear 'em open and fit them into the keyboard trim. It felt really weird pressing the old 5-pin MagSafe connectors out of my trim. Those were the defining unique element to my keyboard... Anyway, no fretting about it, cause it's only gonna be better once it's done! I kinda wish I hadn't been so proactive now with altering my main keyboard for I2C DAC controlled lighting. It's literally the last thing I worked on, and now I have to undo that and return to the old PWM control direct from the Teensy. It's gonna be fun seeing what the inside of that connector looks like. Depending on how tightly packed things are, I wonder if I'll find myself actually needing to solder under a microscope. I don't think I'm at that point yet. For simplicity's sakes, I will very likely solder wires to a breakout board, just for the sakes of keeping things organized. The old I2C breakout will be entirely removed, as the new design change has absolutely no reliance of I2C or port expander chips of any kind. At this point, the keyboard is now a "dumb" device, in terms of the number pad. It will no longer know that the numberpad isn't attached, only that no keys are being pressed.

[richfiles]

qIXmo5v.jpg (659.18 KiB) Viewed 1691 times

Progress has been made! I am as close as I ever have been to finishing fully!
I do still have wiring to do, but the hardest part is done.
I have successfully disassembled the USB-C magnetic connectors,
and successfully installed them into the keyboard and number pad trim pieces!

3dO5fbT.jpg (346.25 KiB) Viewed 1691 times

This part went better than I expected, though it was still rather difficult. I was very concerned about damaging the connectors, but they seemed to take the heat of my hot air rework station well. I suspect there was some sort of thermo-plastic used in their construction. It took a lot more heat and effort to desolder the half of the connector with the pogo pins from the half with the circuit board and female USB-C socket.

mT69dol.jpg (323.33 KiB) Viewed 1691 times

Turns out the USB-C socket was just so new, that it took a lot of force to fully insert the male USB-C plug into it, and get a solidly registered locking insertion. Once the two were together, they were very solidly joined. As it just so happens, both the socket and plug readily accepted solder, so I simply soldered the two halves together into an absolutely permanent joint. The male plug side, as mentioned in the previous post, features the contact cups that the pogo pins mate to when the magnetic connectors are joined. With the Pogo connector separated from the PC board, I had large enough solder pads / solder posts that I can individually solder wires to. Electrically, when looking from either the top or the bottom of the PC board, in the orientation shown in the picture, Ground is always the leftmost pad, and positive always the rightmost pad This does mean that 2 pads are always wired as ground, and 2 pads always wired as positive, leaving 16 remaining pads to carry signals. This is actually exactly perfect, as we have 12 columns, 3 rows, and one LED PWM signal... a total of 16. The board had an integrated LED and resistor wired between ground and positive. I removed it, cause even though i like green, It wasn't what I wanted for this keyboard.

wRo2ZrS.jpg (921.3 KiB) Viewed 1691 times

On a side note, I had to epoxy the ring magnet back onto the contact cup connector, due to my failed attempt to trim the plastic. When I reattached it, I made the error of putting it on backwards, but this has the positive effect of making it so that none of my actual USB-C magnetic connectors or cables are attracted to the connector. They repel, and try to push them above or below. I decided to flip the magnet on the Pogo pin connector too, essentially reverse keying these, as a safety precaution, so no actual USB cable ever unintentionally attaches to the number pad port. The magnets were epoxied permanently in their reverse positions when i epoxied the connectors in place on the trim.

Bs0Ovzd (1).jpg (801.79 KiB) Viewed 1691 times

The old 5-pin MagSafe connector happened to have a key on the contact cup side of the connector, and it needed to be dealt with. My solution was actually quite simple. I filed it out just a bit, got a light pipe out of my LED kit, filed it down to the correct size and shape, and epoxied it in with the connector.

sOwceuJ.jpg (890.43 KiB) Viewed 1691 times

It nicely fills the gap the old keyway leaves, and will provide me with a port accentuating power LED.

7fDceSE.jpg (828.5 KiB) Viewed 1691 times

I used the same type of amber LED as I used throughout all the sub-key lighting across the entire keyboard, but i chose a larger resistor value, so that it would not glow very brightly. I did decide to hardwire it into power, so it will always remain lit, so long as the keyboard itself has power... Just very dimly. I painted the backside of the LED first yellow, to reflect as much light forward as possible, then painted over that with a few layers of black, so there is no bleed through in the adjacent key switch, incase all other LEDs are switched off.

4PjKo61 (1).jpg (191.18 KiB) Viewed 1691 times

The board does stick out over the key switch, but it does clear... barely... It's close enough that I will definitely wrap the board in Kaptan tape, just to make sure it stays insulated. I've removed all the I2C circuitry and the DAC based LED controller. I reconnected the LED drivers (not visible in this pic) to a PWM output on the Teensy, and strung a second wire off the PWM output to the location of the new connector. I've also taken the sense wire, I2C Data, and I2C Clock wires and now have them terminated next to the small circuit board (they are unconnected at the moment).

What remains to be done is wiring columns 1-12 to the connector, and actually soldering everything to the connector. With the old I2C Data and Clock wires, the Sense wire, and the LED PWM wire already strung, it means I already have LED control and the three row wires placed. On the numberpad side, I need to construct a pair of LED drivers (simple... it's two transistors and two resistors), and then just wire up the 18 connections from the connector for power, LED control, and to the already existing switch matrix wiring. There are only a few wires to connect from the LED drivers to the LEDs (2 drive wires, and a single positive wire).

I've also decided that I will run down to the hardware store and buy all new brass screws, as all the opening and closing has marred up the existing ones pretty badly. I'm gonna drop a little sawdust and wood glue into a few of the screw holes, to reform the holes, as a few are stripped out. Other errata include creating a suitable light pipe and shroud for the caps lock key, cutting the bottom plate for the number pad, and maybe finding some better rubber feet. Aside from running some wires and soldering the last bits... This is very nearly done!

Thanks to eliminating the need for a second controller, and no longer needing to figure out how to gracefully hot-plug said controller, the firmware becomes a very simple matter of simply doing a basic single controller firmware. There are well explained guides that go through this process, and I have no doubt I can put the firmware together in no time. So long as I don't get hit by work, or procrastinate, I think there's a real chance I have this working this month yet! I'm genuinely excited, most of all because software isn't holding me back anymore!

[richfiles]

ciuZD0O.jpg (897.19 KiB) Viewed 1620 times

I've got the connector wired up. I've tested the continuity several times, and as far as I can tell, the main keyboard's wiring is now 100% complete! I can't tell you all how happy I am to be able to say that! That said, I am having an issue, and I'm not sure why. I know my LEDs worked before, cause I have tests showing them all working. Caps Lock turns on and off properly. The sub-lighting LEDs are doing something weird. Some banks stay on, some are barely lit, others are off... If I send the signal to turn them on, they all come on properly... There seems to be a floating voltage somewhere keeping them from necessarily fully turning off, but it's strong enough that even a 10KΩ pull up resistor is not enough to overcome it. I'm worried I have a short in the new connector wiring... And that is absolutely terrifying. I have NO slack, and the LED pin is on the bottom, yet I've TESTED continuity... I found no shorts... And I'm SO confused! I need to look it over on a day I'm rested and not busy with work.

uecQCX9.jpg (911.82 KiB) Viewed 1620 times

It's a lot of wiring, packed into a small space, but it's more or less done! I fed the new wiring through the existing lacing. The Serial Data, Serial Clock, and Sense wire are no longer necessary, so those three wires are now three additional rows for the number pad matrix. I had rows 3-13 already within a space of 4 keys from the connector, thanks to the way I'd already wired the two extra columns. Since I needed 12 columns, it was only necessary to run one single new wire to column 2 on the controller.

fu26RHY.jpg (890.85 KiB) Viewed 1620 times

I've also measured and cut the wires to connect the rows and columns of the number pad to it's connector. The one curled wire in the image... That's what's left of the spool. Should be just enough to wire up the LED drivers for the two transistor banks I'll set up in the number pad. I'll have to test if I get the same floating voltage issue in the number pad, of on the controller, if I desolder the pin. I just wanna eliminate the weird floating voltage issue and close this keyboard up FOR GOOD!

After that, I just gotta configure the firmware. Matrix will be simple... Just a few blank spaces in the scan, and all on the single controller. I'll solder up the number pad another day, when I've got some time and am not busy with work.

[richfiles]

lkalkTd.jpg (523.12 KiB) Viewed 1585 times

I pulled an all nighter and finished wiring the Number Pad! I can't even begin to tell you all how happy I am to have it finished! The LED drivers are split into two banks, each driving about half the LEDs. Did I need to... Again, like the main keyboard, I'm greatly limiting the total current each transistor drives, so they should never even consider running hard.

ZRS6qgm.jpg (778.75 KiB) Viewed 1585 times

Once again, I laced all the loose matrix and connector wires. The physical arrangement of the wire paths made this one exceptionally clean... I'm almost sad it's all hidden. Had I never found those absolutely perfect 1960's era test equipment panel covers with that texture, I might have otherwise chosen an polycarbonate back, to show off the internals. I can't say no to retro test equipment parts, (that were used for manufacturing aviation parts, no less!)...

sVwS1tb.jpg (730.61 KiB) Viewed 1585 times

I cut another square of material for the bottom plate. I was tired, and for whatever reason, felt I needed screws in the middle of each corner... SO I guess I added those??? I blame my sleep deprived brain for that decision. At the very least I know that the numberpad will be very rigid. Still need to get a proper set of matching rubber pads to match across both the number pad and the main keyboard. Might need to carefully slice the pads to the right thickness to "absorb" any intolerance that drifted into my build, thanks to any potential warping of the wood or plates. Just like, fractions of an inch I mean... Just wanna set it on the granite surface stone at work and make sure it's absolutely perfectly level. I'll definitely either use a thinner front rubber foot/pad, or carefully trim one to be physically thinner. I'm serious about wanting the front to sit low. Only reason I don't rely exclusively on rear feet, is cause the screws aren't perfectly flush, and I don't want them to scratch anything... I suppose I could try to fix that, but no joke, the front is thin as can be. Remember, even the front row of switches are shaved on the bottom, to remove the plunger cap, and part of the terminals!

W0vEh9Q.jpg (551.58 KiB) Viewed 1585 times

With the exception of troubleshooting my LED driver's floating voltage issue in the main keyboard (and for that matter, seeing if it carries over to the number pad), the keyboard, both main and number pad are completely wired! All that remains is the Caps Lock light pipe, feet, and firmware!
(plus the aforementioned troubleshooting)

PiYmHZM.jpg (892.78 KiB) Viewed 1585 times

[richfiles]

sEQr20A.jpg (859.16 KiB) Viewed 1580 times

Oh yeah, quick addition! The magnets have the perfect amount of pull!
I can easily detach the numberpad from the keyboard, yet can also suspend the number pad by it's magnets!

From a compact 88 key 76% layout, to a 122 key battleship of a behemoth, and back again, all on demand!