~/komar

Keep the git conflict "side" under cursor in (neo)vim

2026-06-26T00:00:00+00:00
tl;dr mc?^\([=><\|]\)\1\{6\}<CR>d?^<<<<<<<?0<CR>/^\([=><\|]\)\1\{6\}<CR>d/^>>>>>>>/0<CR>`c</code></p>


60% keyboard from a Cherry G80-8100
2015-09-05T00:00:00+00:00
Here is an example of what you can do with your cheap POS Cherry keyboard if you have a few days and
some basic plastic cutting and soldering tools. This is not an instruction, but a description of my
mod written a couple of months after I made it, so it may be missing a few details. Feel free to ask
questions;)</p>
This article was written back in 2012 as a post on geekhack.org</a>.
Unfortunately, the forum was having problems as a result of which the post lost its attachments. The
article below is a copy of the old post with attachments fixed.</em></p>
TL;DR?</a></h3>
I built a 60% keyboard from a Cherry MX 8100 and an ikea cutlery container. It looks like this:</p>
</p>
Thanks for reading.</p>
Not TL;DR?</a></h3>
Here it goes...</p>
Firstly, a link: G80-8113LRCUS-0 on eBay for
$20</a>. I bought 3 of
these keyboards from rawko on ebay, because they were cheap, and I'd never tried Cherry Clears
before. One of the ideas was to build my first custom keyboard, so I thought this was a good source
of clears.</p>
Anyway, I'll present how to modify the big cherry board with magnetic card swiper and turn it into a
60% compact keyboard. Or rather how to destroy the board completely, and then build a keyboard from
the parts and pieces. This is not an actual instructable, but it presents the basic steps of
building it and documents the process.</p>
Step one</a></h4>
shrink the keyboard to desired size, don't care for broken connections;)</em></p>
</a>
</a></p>
the 60% block detached (left) and keyboard in two parts (right)</em></p>
The photo on the left shows the pcb with all the switches but those I needed removed, after cutting
it into two parts — the 60% block for the new build and the unnecessary rest. You can see that I
removed the keycaps close to the cutting line, because my milling bit was touching them while
milling. I even destroyed one cap before I realized that. These diamond mills are real beasts, at
20,000 rpm you won't even feel you're milling through plastic or your finger!</p>
The top part is pretty rough, because I didn't have any support, and I was basically holding the
board with my hands; the target board is already smoothed with sandpaper. There's one more thing I
didn't think about when milling, I should really have removed all the switches and resoldered them
after cutting, because the dust got inside and it took a long time to clean it later, even though I
had protected the switches with a solid layer of duct tape. In the background of the right photo is
the original keyboard (well, not exactly that same device obviously...)</p>
Step two</a></h4>
trace the matrix and check which connections you've destroyed</em></p>
After cutting out the main block, some of the matrix connections were obviously broken. I traced the
connections and it appeared that after connecting some sections together and fixing those
connections I had broken I could end up with an 8×8 matrix. Pretty good for a 61-key board.</p>
Unfortunately I don't have a photo of the pcb with just the matrix fixed. In the next step you'll
see the added connections in the background. This Cherry board used switches with diodes, so you
have to be careful what you're connecting together to make sure you have all the cathodes/anodes on
either columns or rows.</p>
Step three</a></h4>
make the controller</em></p>
I wanted to have the keyboard working as soon as possible so I decided to prototype a controller in
as little time as possible. The result of this is an improvised device with just what is necessary
to get the ATmega32u4 up and running. I don't have any schematics, because I never made one, but
what you see below is basically a teensy 2.0 with the unnecessary stuff removed mounted on a tqfp
prototype board with leftover resistor/led leads used as connectors.</p>
</a>
</a></p>
improvised controller (left) and controller and the main board (right)</em></p>
You can see some "innovative" solutions like wires going around the board to the other side or
additional drills, to make the USB B socket as stable as possible to prevent it from damage while
the cable is moving. The controller in these pictures is already connected to the pcb with a ribbon
cable. The other ends of the wires are soldered to switch pins chosen nearly randomly within nets in
a way which doesn't require too much mess in the cables. In the right picture is the controller laid
out as it should be with the ribbons routed and connected.</p>
</a></p>
bottom of the board connected</em></p>
</a>
</a></p>
ribbon cable wiring (left) and controller — other side (right)</em></p>
Above is the bottom of the board after connecting everything together. You can also see my desk and
my professional atx-fan-in-multimeter-box fume extractor;)</p>
Step four</a></h4>
if you don't know how to connect two parts together, use the glue gun... and regret it</em></p>
</a></p>
glue!</em></p>
So I thought, this is a really dirty mod and I don't even know if I'm going to use it, so let's just
hot-glue the controller to the board. Well, even if I did have the photos of the thing glued, I
wouldn't show them here, but basically I didn't put anything between the two surfaces before
applying glue and I had shorts appearing randomly as I pressed the board with my hands. Believe me,
it's really hard to cut through hot-melt adhesive, even with a sharp exacto knife.</p>
So after a few more hours I had the controller reattached and protected from shorts with the bottom
pcb. The gluing job looked even uglier, especially after I realized that it wouldn't fit in the case
(see step six — building the case) and had to remove some of the glue to lay the capacitors flat and
then reglue them to the board.</p>
Step five</a></h4>
clean your desk</em></p>
</a>
</a>
</a>
</a></p>
improvised base — shots 1, 2, 3 (top row), and 4 (bottom)</em></p>
I downloaded the example USB HID code from the teensy project's page, replaced the dummy parts with
some matrix scanning and it basically started to work;)</p>
Then I cleaned my desk and cut some cardboard and packing foam to form a sensible base for the
keyboard. Did I mention, that the thing couldn't even lay flat on the desk, because of the USB
socket on the left? After a few hours of use I decided to do something about the current "design",
because the front of the keyboard (space, modifiers, etc) was really too high. Time for a case!</p>
Step five and a half</a></h4>
clean the switches</em></p>
After cutting, a lot of dust got into the switches and some of them had some minor problems, like
not registering presses or registering them too many times. So I decided to clean everything
thoroughly to make sure the keyboard would work for some more time.</p>
</a>
</a>
</a>
</a></p>
I removed the plastic tops (a bit tricky at first, but after a couple, it's really easy) and sprayed
the housings with compressed air. Then I sprayed the stems with "universal silicone oil" (that's
what the label said) and put some additional green grease of unknown origin on the bottom parts of
the stems. The whole process took about an hour and the switches started working much better, so I
think it was worth it.</p>
Step six</a></h4>
build the case</em></p>
</a>
</a></p>
case parts (left) and back side idea (right)</em></p>
I really got convinced that the 61 keys are more than enough for all my needs. Such layout has some
advantages, for example the mouse is as close to the hands while typing as possible. And you can
press enter with your right hand while it's resting on the mouse. So I was already using this board
as a daily-driver, but well, even considering my aesthetic requirements, this thing didn't look too
good. And it was sliding all over the desk. So I decided to make a case. But I'd never designed a
case in my life. Well, wait. I've never designed a case in my life.</p>
</a></p>
keyboard back</em></p>
The following description can't be treated as an instruction or anything like it, because it's
basically a few photos which I took while building the case. The process was not planned, so I don't
even know how I did it, but somehow the improvisation turned into a case. Maybe not a nice-looking
one, but still a case.</p>
For the plastic, I chose an ikea cutlery container. The link is here: IKEA cutlery
container</a>. For some reason they don't have
them on the US webpage, I don't know if they have them in the stores... The thing is really cheap
and I already had one at home so I thought, a case for a dollar? Why not;)</p>
In the pictures you can see a part of the base of the container cut out to form the base of the
keyboard case. The long brim will be attached to the base to form the back finish, as shown in the
third picture. I had to make the back of the keyboard of two parts of plastic, because the brim of
the plastic thing was facing outwards which I really didn't like. All in all, I think that line
where the two parts join you'll see in the next pictures looks pretty good.</p>
It appears that this kind of plastic is absolutely impossible to glue with nearly anything. I've
tried over 10 different plastic glues and none of them was able to form a connection that would
survive a light twist or pull. But what do you do if you can't glue something? Use the glue gun! I
can't really say that my glue gun glued this plastic well, but at least the case is in one piece and
if you don't throw it against a wall, it will survive a regular wear and tear. I have to say that
the plastic itself is very thin and flexible, so it's really not a good material for a case, but at
least it's very easy to cut.</p>
</a>
</a></p>
case corner (left) and right side of the case (right)</em></p>
I glued the two parts together and used another brim of the container to close the case on the side.
I also cut a hole for the USB connector. As you can see, the glue looks terrible, but that's the
interior. Nobody's going to look inside. After gluing I carefully cut the remaining plastic and
finished the surface with sandpaper. Then I used the brim again to make the front finish and the
other side and glued everything together with even more hot glue.</p>
After some more cutting and sanding, this is the effect just after cleaning everything up:</p>
</a></p>
As far as I remember, the case took me about 2 evenings to finish, mainly because of the lack of a
project and having just one plastic container, which I was afraid of destroying. And I also learned
one thing. Use protective glasses when cutting plastic with your drill!</p>
Step seven</a></h4>
put the keyboard into the case and finish the work!</em></p>
</a>
</a></p>
mounting bolt (left) and mounting bolts (right)</em></p>
The pcb and controller are mounted with 2 bolts and 2 nuts on the lower side. The upper part of the
keyboard is not connected to the case, but since I don't use it in really hard conditions, it seems
unnecessary, especially with the USB connector holding it in place.</p>
</a>
</a></p>
PCB spacer (left) and rubber foot (right)</em></p>
Before I put the keyboard into the case, I made 2 legs from plastic spacers (and hot-melt adhesive
of course) to keep the pcb in an angled position compatible with the case profile. In the end I
attached two rubber feet from the donor board, so that the keyboard didn't move all over the desk
while typing.</p>
</a>
</a></p>
finished board</em></p>
The firmware currently supports two layers and full NKRO over USB. I'm still working on it, because
it's now used in my Universal Keyboard Controller. If I ever manage to finish it, you can expect it
on geekhack;)</p>
This mod would never have happened without the geekhack community. Thanks for your support and help
choosing keyboards and for making me aware of the existence of mechanical keyboards at all!</p>


LED strip dimmer and potting attempt
2014-05-14T00:00:00+00:00
This is just a quick build log of my super-simple dimmer for a LED strip while I'm working on the
second part of How to make a keyboard</a>.</p>
I needed a very simple LED dimmer for a short piece of LED strip I'm going to mount in the cupboard.
A quick google images search for "555 pwm" led me to this idea by Giorgos Lazaridis: An LED Array
PWM Dimmer with the
555</a>.</p>
Here's the schematic of my version:</p>
</p>
555 based PWM LED dimmer schematic</em></p>
The circuit is rather simple. It's powered from a 12V supply. The 555 timer works in its regular
astable mode, where pins 6 and 2 (Threshold and Trigger) are shorted and are used to sense the
voltage across the capacitor C1. The charging/discharging circuit is, however, different than
usually. In order to achieve a reasonably stable frequency, the sum of charge and discharge times
must be independent from the duty cycle setting. For this reason, the capacitor is charged and
discharged using the two complementary halves of the pot RV1. Because these two virtual resistors
have to be connected to the same capacitor pin, diodes must be used to ensure the current flows only
through the right part while charging and only through the left part while discharging.</p>
The capacitor is both charged and discharged using pin 3 (Output), because it is capable of both
sinking and sourcing current. When Output is high, the current flows through the right half of the
pot and D2 and charges the capacitor. When the capacitor is charged to approximately 2/3 of supply
voltage, Output transitions to low and the capacitor is discharged through D1 and the left half of
the pot. When the capacitor voltage drops below approximately 1/3 of supply voltage, Output
transitions back to high and the capacitor is charged again.</p>
Pin 7 (Discharge) is used as this circuit's output, and its state follows that of pin 3 (Output).
Because this pin is an Open Collector</a> output, it
needs a pull-up resistor R1. When it's low, the resistor wastes about 12mA of current. When it's
high, the resistor limits the base current of Q1, driving it to saturation, which in consequence
turns the LED load on. This means that the LEDs are on while C1 charges and off while it discharges.</p>
Building</a></h3>
After building a breadboard prototype and verifying that it works, I decided to keep the size as
small as possible and built the circuit without any PCB by directly soldering wires together.</p>
</a>
</a></p>
Dimmer front (left) and back (right)</em></p>
I managed to put R1 and C2 under the 555's dip package and also make some extra connections using
the remaining component wires. Then I squeezed D1, D2 and C1 under a pretty large pot I happened to
have. It was necessary to use some heat-shrink tubing to protect some wires from shorting. I also
protected the pot's stem with a bit of adhesive tape (sticky face up), so that the hot-melt glue I
was going to use wouldn't prevent it from turning. Next, I connected the two bundles together
(Output to the remaining pin of the pot and C1's sense line to Trigger and Threshold) and added the
transistor.</p>
</a>
</a></p>
The dimmer in a plastic case (left) and 3 walls done (right)</em></p>
The next step was to solder the 4 connection wires and build a mini plastic case. I decided to use a
plastic angle bar to form 2 walls of the case. I cut the two next walls out of the same plastic,
drilled holes for the wires and squeezed the circuit between them. Then I applied hot-melt glue to
temporarily connect the pieces together. In the image on the left the pot is laying face down and
there's a hole in the angle bar to allow for brightness control. After the left wall fell out for
some reason, I added the top one and filled everything with even more glue, treating it as a (very
poor) potting compound.</p>
</a>
</a></p>
All walls but one (left) and the last wall glued (right)</em></p>
I re-added the left wall, filled everything with as much glue as I could and pushed the precisely
crafted front wall inside. Making this wall required some patience as I wanted it to fit very
tightly to make sure there were no holes.</p>
The final step was to cut all the remaining bits of plastic and make the whole device into a nice
box. I used a drill with a grinding disc to cut the plastic and then sandpaper to finish the rough
edges. This took a while but I was left with a not-very-nice-looking final product which I'm still
pretty satisfied with. Below, you can see pictures of the final stage:</p>
</a>
</a>
</a>
</a></p>
As you can see, it's not a very pretty device, but hopefully it will work. Now I'll wait for the LED
strip to arrive to check it. My only concern is that when the transistor gets warm it might start to
melt the glue, and then the whole thing might fall apart. But time will show...</p>
Testing</a></h3>
</a>
</a>
</a>
</a>
</a></p>
From left to right: minimum brightness, duty cycle < 50%, ~50%, > 50%, maximum brightness</em></p>
I hooked the dimmer up to some 3 white LEDs and a limiting resistor to test if everything worked
okay.</p>
</a></p>
It did;)</p>
On the right side I present the voltage plots of the LEDs and resistor in series. As you can see,
the frequency changes a bit with the duty cycle (the lower the duty cycle, the higher the
frequency), but it's still within acceptable range. You can also see the plots for the edge
positions of the pot (minimum and maximum brightness) on the left.</p>