I knew that if I went down this rabbit hole I would become obsessed with it until I finished it, but fuck it... here I am AGAIN with another crazy project.

But this one is my favorite so far. By far.

It was Sunday. I was playing with my daughter when, in the middle of that, it just struck me out of nowhere:

The MX518 was a really good mouse.

I have an MX Master 3S now, paired with MX Keys, and I really enjoy and love that setup.

But I miss my MX518.

So I, of course, dug it out, and immediately remembered why I loved that mouse.

I got it in 2008. It was the Dark Knight edition and it served me really well until around 2016/2017, when I switched to Mac (again) and got myself a Magic Mouse.

I did use my MX here and there when I needed precision and when I was gaming, but since I was working more than gaming, I found myself using the Magic Mouse and trackpad more often while mx518 stayed in my studio.

By 2020 I had completely switched to Bluetooth mice. Mostly some random corporate stuff.

I always had my MX with me, but it was not practical anymore. The cable became redundant, and I had already repaired the cable connection once after it desoldered from the board.

At some point I got myself a Rapoo MT750S/L — back when I was still considering whether to buy an MX Master.

Once I got it, somewhere around 2022, I fell in love with horizontal scrolling. It is such a useful thing for me that I could not see myself without it anymore. I did use horizontal scrolling a lot on the Magic Mouse too, to be fair.

I used the Rapoo as my main mouse for quite some time and I was impressed by the battery life. I literally charged it twice a year, and I used it a ton across multiple devices.

After I got the MX Master 3S, I continued using the Rapoo as my mobile/office mouse. So the Rapoo became a permanent citizen of my backpack, while the MX Master stayed at home chilling on the cozy desk.

Then my job shifted more toward working from home, so I barely used the Rapoo anymore.

And then I started looking at my MX518.

At that point the idea was already accepted, whether I wanted it or not.

My brain had decided:

Let’s make the old MX518 useful again.

I wanted to convert it into my mobile mouse, and for that I needed to make it wireless.

Since I already have a bunch of other projects involving ESP32 controllers, Raspberry Pis, batteries and similar stuff, I thought it could not be that complicated to make some kind of USB-to-BLE board that would have a battery, USB charging port, read the mouse and send the data wirelessly to the PC.

Then I thought there must already be DIY devices like this ready to buy.

And there are plenty on AliExpress.

The problem: almost all of them are designed for keyboards.

I contacted a couple of sellers and they confirmed that their boards were keyboard-only.

Flashing some random proprietary chip was not really an option, and I didn’t want this to turn into an “ask AI to reverse-engineer unknown firmware” project.

So I decided:

Fine. ESP32-S3.

It already gives me most of what I need. Add charging, battery, firmware, done.

But then I realized that the MX would be a great nostalgic device, but I was turning this into a lot of work just to make an old mouse wireless.

So I went back to looking at DIY boards.

And while thinking about what to do, I went all the way back to square one:

What do I actually want from this?

I want my old buddy back.

Refreshed.

Modernized.

And usable as my mobile mouse.

So...

Let’s sacrifice my current mobile mouse for this.

I opened the Rapoo.

And this was not the first time I had opened it. I had already repaired the DPI/Bluetooth switches once after they broke loose and fell inside.

Then I opened both mice.

And at first it was obvious that the Rapoo electronics could not simply fit inside the Logitech body. The Logitech shell is much narrower toward the front.

BUT...

Apparently corporate Cthulhu had not sunk its tentacles too deep into the mouse division yet, because the Rapoo turned out to be surprisingly modular.

The main board is mounted to the bottom shell, while almost everything else lives on separate daughterboards.

That changed everything.

I drew a sketch of how I thought the whole thing should work.

All of the original Logitech click mechanics would stay.

I would replace the main electronics with the Rapoo board and rewire everything so each Logitech control talks to the corresponding Rapoo input.

And this is the point where this became my favorite project so far.

Challenge 1: The board does not fit

The Rapoo board has its sensor farther back than the original Logitech sensor.

Because of the MX518 scroll-wheel mounting posts, I could not push the Rapoo board far enough forward.

That left the new sensor almost 20 mm closer to the palm compared to the original position.

Quick fix:

Drill a new hole for the sensor.

Done.

Except...

That changes the way the mouse tracks.

I did a quick investigation using basically every mouse I had around me, and noticed that the sensor is generally somewhere around the area below the knuckles.

If I moved it almost 20 mm toward the palm, the mouse would behave more like you were holding it much deeper in your hand.

I did not want to overthink it, so I connected the Rapoo board outside the shell and tested the tracking.

And honestly, it did not feel that bad.

But wrist/palm movement became more noticeable.

Over the years I had trained myself to keep my palm resting on the mouse pad and mostly move the mouse from the hand/wrist. Moving the sensor backward changed that relationship enough that I noticed it.

So I wanted to do it properly.

Proper fix:

I needed to push the Rapoo board underneath the MX518 scroll wheel somehow.

I briefly thought about rotating the whole Rapoo board 180 degrees.

But then I would make my life a living hell by having to reverse X/Y tracking in software on every computer that uses the mouse.

So that idea died quickly.

I drew a ton of sketches and thought through a lot of different options.

Eventually I decided that I absolutely wanted to keep the original MX518 scroll wheel.

I wanted to preserve as many original parts as possible, especially the wheel, because I really like that old-school narrow Logitech rubber scroll wheel.

The idea was to 3D print a bridge connecting all four original scroll-wheel supports.

I went through multiple versions trying to get:

• enough structural support
• the correct height between the Logitech click board and Rapoo main board
• enough clearance underneath for the Rapoo PCB
• a tight fit against the original support posts

Once I got a version I liked, I installed it and glued the original scroll supports to it.

Yes, I know. Glue is usually the enemy of repairability.

But this is a one-way project.

There is no going back.

So glue is allowed.

Once the bridge was solid, I took the Dremel and removed the lower half of two of the original scroll supports.

That created just enough of a tunnel for the Rapoo motherboard to slide underneath.

And it worked.

The Rapoo sensor landed exactly where the original Logitech sensor hole already was.

Success. Challenge one solved.

I connected the battery, paired the mouse with the computer and moved it around.

The cursor moved.

At that point I knew the project was real.

Next step: wiring.

I bought a multimeter because I did not want to guess which wire did what.

I mapped Left Click, Right Click, Middle Click and the scroll wiring.

And then I finally had to deal with the problem I had known about from the beginning but had conveniently swept under the carpet, hoping it would somehow solve itself.

lol.

The MX518 has an optical scroll encoder.

The Rapoo uses a mechanical rotary encoder.

Even if I can map all the wires, the signals are not the same.

So the original optical scroll system had to go.

Challenge 2: Replace the optical scroll with a mechanical encoder

This was probably the hardest challenge.

The Rapoo encoder opening/shaft arrangement is much smaller than the original Logitech optical mechanism, so the MX wheel simply could not fit into it.

I tried to figure out a way to make a new axle, unsuccessfully.

At one point I put the MX scroll wheel into the Dremel and tried sanding the center thinner.

It broke.

Which, honestly, I was expecting.

So I left the whole thing alone overnight.

Then another night.

After thinking through a ridiculous number of possible solutions, I finally decided that the only sensible way forward was:

Make the Rapoo encoder as short and compact as possible.

Then disassemble the original Rapoo scroll wheel and steal its center hub and axle.

The idea was to take the original Rapoo axle together with its hub, machine the hub down, and implant it inside the Logitech scroll wheel.

So that is what I did.

Carefully.

Slowly.

A lot of sanding, trimming and test-fitting.

And somehow...

It worked.

The wheel turned straight.

It was balanced.

And the cherry on top:

the encoder actually fit inside the modified Logitech wheel.

Next step was mounting the encoder so the entire wheel assembly could still move vertically for Middle Click.

Aaaaand...

The original guide broke.

Challenge 3: Mount the Frankenstein scroll wheel and still keep Middle Click

At this point I had a lot of thoughts.

One of them was:

“Fuck this. Buy a newer Logitech scroll wheel and adapt that.”

I considered a bunch of completely different approaches.

But once I had the MX wheel spinning properly with the transplanted Rapoo axle, I realized:

I am unstoppable now.

So I 3D printed a new mount/guide for the encoder.

Then another.

Then another.

A lot of versions, sanding, cutting, fitting, throwing things away and printing again.

Eventually I got one where the encoder can slide vertically.

The Middle Click switch itself is actually strong enough to push the whole assembly back up.

Technically it works without anything else.

But then Middle Click becomes way too sensitive.

While scrolling, it becomes very easy to accidentally click the wheel.

Absolutely not.

So I kept the original Logitech return spring.

Now the switch and the spring work together, just like I wanted.

The wheel comes back properly, scrolling does not accidentally trigger Middle Click, and the whole mechanism feels much closer to a proper mouse rather than a prototype.

And since I was already this deep into the insanity...

I decided to transplant the Rapoo horizontal scroll too.

AND IT FITTED FLAWLESSLY.

Seriously.

The position is great.

It works.

It feels really good.

And I actually think the position is better than on the Rapoo and mx masetr 3.

You can keep a normal grip on the mouse while using it. You do not have to lift your thumb upward to reach the wheel, it is not in the way during normal use, and it sits naturally underneath the original Back/Forward buttons.

And somehow...

It actually looks good too.

So now, since I am already going completely overboard, I ordered silent switches from AliExpress and decided to replace all the clicks with silent ones.

And this is where the project goes on hold for a little while.

Next step

I am waiting for a new encoder because I abused this one pretty badly during all the bending, pushing, fitting and experimentation, and it now scrolls harder than it originally did.

I am also waiting for the silent switches.

Once those arrive, I will continue with:

• final encoder installation
• rewiring all the buttons
• horizontal scroll wiring is just plug and play yaay
• DPI/Bluetooth controls
• silent Left/Right
• silent Back/Forward
• silent Middle Click
• USB-C charging port
• final assembly

When I finish this, the MX518 might actually become my main mouse again.

Who knows.

There is also one more possibility.

While working with the Rapoo board I noticed that it has exposed development/debug pads.

And since multiple Rapoo mouse models seem to use the same board, I suspect that the Bluetooth device name may simply exist somewhere in the firmware as a string.

So obviously my next stupid idea is to get into the firmware and rename the mouse to: (imagine bricking the device after all this work lol)

MX518 RMG Edition

Or something more violent.

We’ll see.

I’ll keep you updated.

What I expect to have when this is finished

• original MX518 shell and ergonomics
• sensor upgraded from the original optical system to the Rapoo sensor
• wireless conversion: 3 Bluetooth profiles + 1× 2.4 GHz receiver mode
• vertical scroll converted from optical sensing to a mechanical encoder
• horizontal thumb scroll
• silent Left/Right clicks
• silent Back/Forward clicks
• silent Middle Click
• functional DPI control
• functional Bluetooth/device switching
• rechargeable battery
• USB-C charging

Let’s hope the parts arrive soon.

I knew that if I went down this rabbit hole I would become obsessed with it until I finished it, but fuck it... here I am AGAIN with another crazy project.

But this one is my favorite so far. By far.

It was Sunday. I was playing with my daughter when, in the middle of that, it just struck me out of nowhere:

The MX518 was a really good mouse.

I have an MX Master 3S now, paired with MX Keys, and I really enjoy and love that setup.

But I miss my MX518.

So I, of course, dug it out, and immediately remembered why I loved that mouse.

I got it in 2008. It was the Dark Knight edition and it served me really well until around 2016/2017, when I switched to Mac (again) and got myself a Magic Mouse.

I did use my MX here and there when I needed precision and when I was gaming, but since I was working more than gaming, I found myself using the Magic Mouse and trackpad more often while mx518 stayed in my studio.

By 2020 I had completely switched to Bluetooth mice. Mostly some random corporate stuff.

I always had my MX with me, but it was not practical anymore. The cable became redundant, and I had already repaired the cable connection once after it desoldered from the board.

At some point I got myself a Rapoo MT750S/L — back when I was still considering whether to buy an MX Master.

Once I got it, somewhere around 2022, I fell in love with horizontal scrolling. It is such a useful thing for me that I could not see myself without it anymore. I did use horizontal scrolling a lot on the Magic Mouse too, to be fair.

I used the Rapoo as my main mouse for quite some time and I was impressed by the battery life. I literally charged it twice a year, and I used it a ton across multiple devices.

After I got the MX Master 3S, I continued using the Rapoo as my mobile/office mouse. So the Rapoo became a permanent citizen of my backpack, while the MX Master stayed at home chilling on the cozy desk.

Then my job shifted more toward working from home, so I barely used the Rapoo anymore.

And then I started looking at my MX518.

At that point the idea was already accepted, whether I wanted it or not.

My brain had decided:

Let’s make the old MX518 useful again.

I wanted to convert it into my mobile mouse, and for that I needed to make it wireless.

Since I already have a bunch of other projects involving ESP32 controllers, Raspberry Pis, batteries and similar stuff, I thought it could not be that complicated to make some kind of USB-to-BLE board that would have a battery, USB charging port, read the mouse and send the data wirelessly to the PC.

Then I thought there must already be DIY devices like this ready to buy.

And there are plenty on AliExpress.

The problem: almost all of them are designed for keyboards.

I contacted a couple of sellers and they confirmed that their boards were keyboard-only.

Flashing some random proprietary chip was not really an option, and I didn’t want this to turn into an “ask AI to reverse-engineer unknown firmware” project.

So I decided:

Fine. ESP32-S3.

It already gives me most of what I need. Add charging, battery, firmware, done.

But then I realized that the MX would be a great nostalgic device, but I was turning this into a lot of work just to make an old mouse wireless.

So I went back to looking at DIY boards.

And while thinking about what to do, I went all the way back to square one:

What do I actually want from this?

I want my old buddy back.

Refreshed.

Modernized.

And usable as my mobile mouse.

So...

Let’s sacrifice my current mobile mouse for this.

I opened the Rapoo.

And this was not the first time I had opened it. I had already repaired the DPI/Bluetooth switches once after they broke loose and fell inside.

Then I opened both mice.

And at first it was obvious that the Rapoo electronics could not simply fit inside the Logitech body. The Logitech shell is much narrower toward the front.

BUT...

Apparently corporate Cthulhu had not sunk its tentacles too deep into the mouse division yet, because the Rapoo turned out to be surprisingly modular.

The main board is mounted to the bottom shell, while almost everything else lives on separate daughterboards.

That changed everything.

I drew a sketch of how I thought the whole thing should work.

All of the original Logitech click mechanics would stay.

I would replace the main electronics with the Rapoo board and rewire everything so each Logitech control talks to the corresponding Rapoo input.

And this is the point where this became my favorite project so far.

Challenge 1: The board does not fit

The Rapoo board has its sensor farther back than the original Logitech sensor.

Because of the MX518 scroll-wheel mounting posts, I could not push the Rapoo board far enough forward.

That left the new sensor almost 20 mm closer to the palm compared to the original position.

Quick fix:

Drill a new hole for the sensor.

Done.

Except...

That changes the way the mouse tracks.

I did a quick investigation using basically every mouse I had around me, and noticed that the sensor is generally somewhere around the area below the knuckles.

If I moved it almost 20 mm toward the palm, the mouse would behave more like you were holding it much deeper in your hand.

I did not want to overthink it, so I connected the Rapoo board outside the shell and tested the tracking.

And honestly, it did not feel that bad.

But wrist/palm movement became more noticeable.

Over the years I had trained myself to keep my palm resting on the mouse pad and mostly move the mouse from the hand/wrist. Moving the sensor backward changed that relationship enough that I noticed it.

So I wanted to do it properly.

Proper fix:

I needed to push the Rapoo board underneath the MX518 scroll wheel somehow.

I briefly thought about rotating the whole Rapoo board 180 degrees.

But then I would make my life a living hell by having to reverse X/Y tracking in software on every computer that uses the mouse.

So that idea died quickly.

I drew a ton of sketches and thought through a lot of different options.

Eventually I decided that I absolutely wanted to keep the original MX518 scroll wheel.

I wanted to preserve as many original parts as possible, especially the wheel, because I really like that old-school narrow Logitech rubber scroll wheel.

The idea was to 3D print a bridge connecting all four original scroll-wheel supports.

I went through multiple versions trying to get:

• enough structural support
• the correct height between the Logitech click board and Rapoo main board
• enough clearance underneath for the Rapoo PCB
• a tight fit against the original support posts

Once I got a version I liked, I installed it and glued the original scroll supports to it.

Yes, I know. Glue is usually the enemy of repairability.

But this is a one-way project.

There is no going back.

So glue is allowed.

Once the bridge was solid, I took the Dremel and removed the lower half of two of the original scroll supports.

That created just enough of a tunnel for the Rapoo motherboard to slide underneath.

And it worked.

The Rapoo sensor landed exactly where the original Logitech sensor hole already was.

Success. Challenge one solved.

I connected the battery, paired the mouse with the computer and moved it around.

The cursor moved.

At that point I knew the project was real.

Next step: wiring.

I bought a multimeter because I did not want to guess which wire did what.

I mapped Left Click, Right Click, Middle Click and the scroll wiring.

And then I finally had to deal with the problem I had known about from the beginning but had conveniently swept under the carpet, hoping it would somehow solve itself.

lol.

The MX518 has an optical scroll encoder.

The Rapoo uses a mechanical rotary encoder.

Even if I can map all the wires, the signals are not the same.

So the original optical scroll system had to go.

Challenge 2: Replace the optical scroll with a mechanical encoder

This was probably the hardest challenge.

The Rapoo encoder opening/shaft arrangement is much smaller than the original Logitech optical mechanism, so the MX wheel simply could not fit into it.

I tried to figure out a way to make a new axle, unsuccessfully.

At one point I put the MX scroll wheel into the Dremel and tried sanding the center thinner.

It broke.

Which, honestly, I was expecting.

So I left the whole thing alone overnight.

Then another night.

After thinking through a ridiculous number of possible solutions, I finally decided that the only sensible way forward was:

Make the Rapoo encoder as short and compact as possible.

Then disassemble the original Rapoo scroll wheel and steal its center hub and axle.

The idea was to take the original Rapoo axle together with its hub, machine the hub down, and implant it inside the Logitech scroll wheel.

So that is what I did.

Carefully.

Slowly.

A lot of sanding, trimming and test-fitting.

And somehow...

It worked.

The wheel turned straight.

It was balanced.

And the cherry on top:

the encoder actually fit inside the modified Logitech wheel.

Next step was mounting the encoder so the entire wheel assembly could still move vertically for Middle Click.

Aaaaand...

The original guide broke.

Challenge 3: Mount the Frankenstein scroll wheel and still keep Middle Click

At this point I had a lot of thoughts.

One of them was:

“Fuck this. Buy a newer Logitech scroll wheel and adapt that.”

I considered a bunch of completely different approaches.

But once I had the MX wheel spinning properly with the transplanted Rapoo axle, I realized:

I am unstoppable now.

So I 3D printed a new mount/guide for the encoder.

Then another.

Then another.

A lot of versions, sanding, cutting, fitting, throwing things away and printing again.

Eventually I got one where the encoder can slide vertically.

The Middle Click switch itself is actually strong enough to push the whole assembly back up.

Technically it works without anything else.

But then Middle Click becomes way too sensitive.

While scrolling, it becomes very easy to accidentally click the wheel.

Absolutely not.

So I kept the original Logitech return spring.

Now the switch and the spring work together, just like I wanted.

The wheel comes back properly, scrolling does not accidentally trigger Middle Click, and the whole mechanism feels much closer to a proper mouse rather than a prototype.

And since I was already this deep into the insanity...

I decided to transplant the Rapoo horizontal scroll too.

AND IT FITTED FLAWLESSLY.

Seriously.

The position is great.

It works.

It feels really good.

And I actually think the position is better than on the Rapoo and mx masetr 3.

You can keep a normal grip on the mouse while using it. You do not have to lift your thumb upward to reach the wheel, it is not in the way during normal use, and it sits naturally underneath the original Back/Forward buttons.

And somehow...

It actually looks good too.

So now, since I am already going completely overboard, I ordered silent switches from AliExpress and decided to replace all the clicks with silent ones.

And this is where the project goes on hold for a little while.

Next step

I am waiting for a new encoder because I abused this one pretty badly during all the bending, pushing, fitting and experimentation, and it now scrolls harder than it originally did.

I am also waiting for the silent switches.

Once those arrive, I will continue with:

• final encoder installation
• rewiring all the buttons
• horizontal scroll wiring is just plug and play yaay
• DPI/Bluetooth controls
• silent Left/Right
• silent Back/Forward
• silent Middle Click
• USB-C charging port
• final assembly

When I finish this, the MX518 might actually become my main mouse again.

Who knows.

There is also one more possibility.

While working with the Rapoo board I noticed that it has exposed development/debug pads.

And since multiple Rapoo mouse models seem to use the same board, I suspect that the Bluetooth device name may simply exist somewhere in the firmware as a string.

So obviously my next stupid idea is to get into the firmware and rename the mouse to: (imagine bricking the device after all this work lol)

MX518 RMG Edition

Or something more violent.

We’ll see.

I’ll keep you updated.

What I expect to have when this is finished

• original MX518 shell and ergonomics
• sensor upgraded from the original optical system to the Rapoo sensor
• wireless conversion: 3 Bluetooth profiles + 1× 2.4 GHz receiver mode
• vertical scroll converted from optical sensing to a mechanical encoder
• horizontal thumb scroll
• silent Left/Right clicks
• silent Back/Forward clicks
• silent Middle Click
• functional DPI control
• functional Bluetooth/device switching
• rechargeable battery
• USB-C charging

Let’s hope the parts arrive soon.

I built RPAi Daily because my notes apps kept becoming another job.

I did not need another system that asked me to design a database, maintain a complex taxonomy, build dashboards, or organize the organization system itself.

I needed something simpler:

Open the day.

Write what matters.

Keep a few notes nearby.

See unfinished work.

Return tomorrow without configuring everything again.

RPAi Daily is a local-first desktop workspace with:

• calendar-connected Daily notes
• regular Markdown notes in tabs
• search and nested folders
• a lightweight Board with a Parking Lot
• ordinary files stored in a local Workspace selected by the user

There is no account, cloud synchronization, telemetry, subscription, or proprietary note format.

Your notes remain ordinary Markdown and text files that can be opened outside the application.

The first public version is free and available for Windows and macOS Intel.

I launched it on Product Hunt today, but what I need most is honest feedback from people who actually try it:

Where did you hesitate?

What did you expect to happen?

Which part felt more complicated than it should?

Product page: https://rpa.rs/en/tools

Product Hunt: https://www.producthunt.com/products/rpai-daily?launch=rpai-daily

I built RPAi Daily because my notes apps kept becoming another job.

I did not need another system that asked me to design a database, maintain a complex taxonomy, build dashboards, or organize the organization system itself.

I needed something simpler:

Open the day.

Write what matters.

Keep a few notes nearby.

See unfinished work.

Return tomorrow without configuring everything again.

RPAi Daily is a local-first desktop workspace with:

• calendar-connected Daily notes
• regular Markdown notes in tabs
• search and nested folders
• a lightweight Board with a Parking Lot
• ordinary files stored in a local Workspace selected by the user

There is no account, cloud synchronization, telemetry, subscription, or proprietary note format.

Your notes remain ordinary Markdown and text files that can be opened outside the application.

The first public version is free and available for Windows and macOS Intel.

I launched it on Product Hunt today, but what I need most is honest feedback from people who actually try it:

Where did you hesitate?

What did you expect to happen?

Which part felt more complicated than it should?

Product page: https://rpa.rs/en/tools

Product Hunt: https://www.producthunt.com/products/rpai-daily?launch=rpai-daily

Small update on my homemade Cardputer Zero / Cardputer-compatible DIY dev rig.

This is the next step after my first “first pulse” post where I finally got the 1.9-inch ST7789 display working with a Raspberry Pi Zero 2 W. The goal is still the same: build a small Cardputer Zero-style dev rig so I can develop, test, and port apps for this form factor before the official device starts shipping.

Like with my Cardputer ADV + Pi setup, I’m treating this as a project launcher before executing my tasks.

Aim small, miss small.

The latest version is now much cleaner physically, even though internally it is still very much a dev rig. I printed a small case for the Pi Zero 2 W and the 1.9-inch display only. The display is mounted, the Pi is inside, and the ports are exposed. It is not perfect, and it is definitely not the most DIY-friendly build because I ended up gluing some parts in place, but after many hours spent fighting with this thing, I decided this is good enough for what I actually need.

I do not want to spend another week polishing the case. I want to build apps.

The biggest headache was the Unit Card KB2 keyboard.

I really wanted to use it as the keyboard for this build, but I had trouble getting it to work reliably. The only thing I managed to get working was BLE, but in my testing it was slow, unreliable, and buggy enough that I did not want to depend on it for testing the apps.

So I changed the plan.

For now, the Cardputer ADV is acting as the keyboard, and the Pi Zero 2 W is running the actual Cardputer Zero-style system and apps on the external 1.9-inch display. It is not the final fantasy version I had in my head, but it works, and that is more important.

The ADV gives me a comfortable keyboard, the display shows the apps and firmware UI, and the Pi holds everything together.

One important note for anyone thinking of going down this route: Cardputer USB alone was not enough to reliably power the Pi + display in my setup. I currently need both power and USB connected to the cardputer adv. I might eventually make a Grove-to-USB power cable so the Cardputer can power the Pi/display setup more cleanly, but since this is only a dev rig, I am not sure it is worth the extra work right now.

The software side was also a lot of calibration and firmware setup. I had to adjust the Cardputer Zero-style firmware environment and remap the keyboard behavior so it makes sense with the Cardputer ADV layout instead of the Zero setup.

Big thanks to Codex here, because without it helping me reason through the keyboard mapping and firmware adjustments, I would probably still be losing my mind over tiny input issues.

The good news is: it works now.

I can boot the system, navigate the launcher, test apps, and continue developing without having to wait for the final hardware.

Apps I am testing / porting on this setup include:

• Void
• Creative Pressure
• Signal vs Noise
• Void Note
• Song Check
• Noirsonance Toolkit
• Control Center
• Some small experimental music and writing tools

I still strongly believe this form factor can be much more than pentesting and RF tools.

That stuff is cool, of course, but I also think tiny keyboard devices like this are perfect for notes, journaling, creative pressure, field tools, music tools, system control, tiny dashboards, and weird personal utilities.

This build is not polished. It is not a clean tutorial yet. It is not something I would recommend unless you enjoy headaches.

But it is now usable.

And honestly, that was the point.

For anyone considering a similar route: prepare for a lot of calibration, keyboard weirdness, power issues, and small “why is this not working” moments. But once it starts working, it becomes a really fun little development setup.

Now I can finally continue building apps in peace lol.

For anyone interested here are previous posts for more context information and progress

• Cardputer DIY First post idea in the shaping
• Cardputer DIY Second post parts arriving
• Cardputer ADV + Raspberry pi zero w 2

Till next crazy idea.

I finally managed to get a real pulse from my homemade Cardputer Zero / Cardputer-compatible DIY dev rig.

For context: this is part of my ongoing project where I am building a small Cardputer-style device using a Raspberry Pi Zero 2 W, a 1.9-inch ST7789 display, and soon the Unit Card KB2 keyboard. The goal is to develop, test and port apps for this form factor before the official device starts shipping.

Previous posts for context:

• Cardputer DIY First post idea in the shaping
• Cardputer DIY Second post parts arriving
• Cardputer ADV + Raspberry pi zero w 2

This step was tough, at least for me. A lot of ups and downs, and a lot of “why is this not working” moments, but I finally managed to make the display work properly with my system.

And of course, the first thing I did was patch the system so I can exit the Store xD
In this pre-beta version, once you enter the Store, you are basically stuck there lol. But now it is alive.

I also started porting some of my existing apps, because I strongly believe devices like this can be much more than pentesting and RF signal analyzing tools.

Do not get me wrong, that stuff is cool, but I think this form factor is also perfect for notes, creative pressure, field tools, small utilities, music tools, system control, and weird little personal apps.

Apps I ported so far, and plan to publish to the app store:

Void: Speak to the void because you know the void will never talk back.

Creative Pressure: Put yourself under pressure to finalize the thought you have in your head.

Signal vs Noise: Distill messy brainstorming ideas into signal worth keeping, or noise that only bloats your mind.

Void Note: I am a huge fan of Obsidian, daily notes, and journaling, so of course I need notes on a Cardputer too.

Song Check: A small audio checking tool. This is also coming as a desktop app soon for music producers to check signal, levels, and overall audio health.

Noirsonance Toolkit: A bunch of small tools for music producers and audio engineers.

Control Center: The same control center idea I released for Pi Zero and Cardputer ADV, now running here too.

Also, for those who haven’t had a chance to see the 1.9-inch display in action: this thing is great.

I was a little suspicious when I first saw “1.9-inch display” in the cardputer zero specs. Knowing the original Cardputer has a 1.14-inch screen, I wasn’t sure how much of a difference around 0.8 inches would really make. It is still under 2 inches, after all.

But after playing with it for a bit, let me tell you: it feels perfect for this form factor.

It still keeps the whole device small, which is the point, but the screen gives you just enough extra space to show a decent amount of information. And yes, the resolution may not look impressive on paper in a world of bazillion-K displays, but in real use it looks great.

I think people will not be disappointed once the official device starts shipping.

It is still messy, still wired like a small red spider, and still far from finished, but this was the first real pulse.

Next step is Unit Card KB2. Once I get that working, I’ll start putting everything into the case.

I carry Mesh and Porkchop with me everywhere and after a few months for Porkchop (days for Mesh) I started noticing scratches on the screen and the body, so I got some cheap hard case with dimensions just enough for two devices plus a LoRa module.

BUT I started noticing that Porkchop is heating badly, so much that it freezes and smells like it's melting, so I started thinking about cooling options and naturally I am on the road to overengineering it.

I had a hard time making these old MacBook fans spin, so I will probably get plain old regular slim laptop fans with plain old motors with 2 wires and connect them to 2x AAA batteries, but that's yet to be finished.

What are your carrying experiences? Heat, design, protection..?

So, all the parts finally arrived, and now that I can measure everything properly, I've run into the first real challenge.

My original plan was to desolder all the buttons from the KB2 and place them where I wanted. As you can see, the keyboard is almost the same size as the Cardputer itself, but the buttons are 6 mm × 6 mm, which doesn't leave much room for the display. Since I don't want to change the original size (I'm building a Cardputer Zero, not a cyberdeck 😄), I ran into space limitations.

Then I realized I could unglue the display from its PCB, move the board slightly lower inside the case, and gain just enough room. Tight, but enough.

Unfortunately, that's when I noticed the clever keyboard design. The diodes are placed directly between the bottom legs of each switch. With my soldering skills, I would almost certainly knock off or damage at least one diode while desoldering 42 switches, and then I'd be hunting keyboard matrix problems for days.

So I started looking for another solution.

Since I actually have quite a bit of free space inside the case (I'm not carrying all the ports and extra electronics found in the original Cardputer Zero), I decided to order a new set of switches from AliExpress. This time I went with 4.5 mm × 4.5 mm switches, which should fit comfortably below the display.

The idea is simple: solder the new switches directly onto the existing ones. The original keyboard electronics stay untouched, the matrix remains intact, and the new switches should short the contacts exactly the same way as the originals. I'll also 3D print a small spacer/cover to prevent accidental presses on the original switches underneath.

If this works, I've solved the biggest mechanical challenge so far and made my life much easier. Soldering new switches onto existing ones is still a lot of work, but it's far less risky than removing 42 switches and hoping I don't destroy a diode somewhere along the way.

While I'm waiting for the new switches to arrive, I'll focus on getting all the components working outside the case. Once everything is tested, I can spend some more time redesigning the enclosure internals so it properly fits and secures all the hardware.

Progress is slower than expected, but at least the project is still moving forward.

Mom can we have Cardputer Zero?

We have Cardputer Zero at home...

After M5Stack announced shipping starts in September, I realized the earliest I’d probably get one in my hands is October… realistically November. So… let’s go for Cardputer Zero DIY Part 2 lol.

I’ll keep you posted, but the current goal is to build one with:

• Raspberry Pi Zero W2
• Battery + all the necessary power components and proper on/off switch
• CardKB2 keyboard (where I’ll need to de-solder ALL buttons… god help me with this assignment)
• 1.9 inch screen

The plan is to make it as clean as possible, ideally not a single port visible… or MAYBE I push the Pi to the edge and expose the ports intentionally.

I’m also considering using a Pi UPS with an 18650 battery, but I need to calculate dimensions and see how everything fits once the keyboard arrives. Honestly, fitting the keyboard will probably be the biggest challenge of the entire build.

Missing parts are already ordered from Ali, so hopefully in a couple of weeks I’ll have everything needed to start assembling.

Once the build begins, I’ll post updates, progress pics, files, and schematics again.

I built a tiny desktop app called Void Room.

It does almost nothing.

And that’s the point.

https://noirsonance.com/product/void-room/

Sometimes you don’t want AI.
Sometimes you don’t want productivity systems.
Sometimes you don’t want notifications, collaboration, optimization, or another app pretending to change your life.

Sometimes you just want to say something into the void.

I think I will call it a project here (when it comes to hardware)
I’ve been messing around with the Cardputer (v1 and ADV) for some time and I knew and I felt like it could be something more practical, and since I had lot of stuff going on my raspberry's I decided to turn ADV into control device for pi and put all together and to push it a bit and see where it goes.

I made a 3D printed adapter so everything sits inside the case without completely ruining the original look and to be reversable at any point. It’s not perfect (fit could be tighter, screws are a bit off), but it holds together well and actually looks like a single device now.

The Pi is running a small system I’ve been building:
https://github.com/rimedag/VaultPi-Control-Center

It’s basically a lightweight control center where I can run scripts, trigger things, manage small workflows, repo's, etc. Running it inside the browser is much nicer and better looking but with custom cardputer firmware I can access it via cardputer, It’s very terminal-first, nothing fancy visually (on cardputer), but it fits the screen and the whole concept pretty well.

I also added a few things that made it feel way more “real” than I expected.

For example, I put a w3m browser on the Pi and integrated it into the firmware, so you can actually browse the internet from this thing. It’s obviously minimal, but it works. There’s also a terminal exposed so you can directly access the Pi, not just go through menus.

The GO button on top is now used for lock/unlock, and I added an optional PIN if you want some basic safety. There’s also a small thing like keeping the screen on while charging, which sounds minor but makes a big difference in usability (When it sits on my desk and I can with glance of my eye see if my websites and services are running or if anything is down).

Communication between the Cardputer and the Pi is currently pretty simple, they talk over the same network. The control center listens on a port and the Cardputer connects to it with a password. It works reliably enough for now.

I did think about going lower level. Those two unused data wires (from grove) are basically just sitting there and could be wired directly to the Pi’s GPIO, which would allow full control even without network. That was the original idea.

But I stopped myself there.

I know how that goes, one more layer, then another, then suddenly it’s a never-ending project. So I left that as a “maybe later” and kept this version focused and working.

Power-wise, same story. I originally wanted to go all-in with a second battery and extra modules to properly power the Pi. I even got advice to do that. But again, I simplified it.

Right now I’m powering the Pi through Grove. For normal use it works fine. It’s not meant for heavy loads, but for what this is, it’s stable enough. I’ll test battery life properly and share results once I have something real.

Where it stands right now is somewhere between prototype and usable device. It works, I actually use it, but it’s not polished. Things that could be better:

• smaller screws
• tighter 3D print
• cleaner internal layout
• Additional battery and modules for having pi with its own powersupply (but then it needs switch bla bla bla too much hustle + not intuitive)

Firmware also isn’t “clean” yet. It works, but there are hiccups. I’ll make that repo public once I fix the obvious stuff. The Pi side is already up if anyone wants to try it.

I had plans to go much deeper with this, more integration, maybe push it towards a small OS-like system, but I intentionally stopped before it turns into something I never finish.

So this is it for now. Not perfect, but real and working.

Lets call it a v1

Curious what people think, if you are inpatient to get your hands on CP Zero this might be worth a try, even having raspberry separated from cardputer is good idea, since they have to be on the same network raspberry could sit somewhere in the house while you access it and manipulate it via cardputer

For those wanting to have raspberry on keychain I have modified 3d print in github repo beside the app

Oh I almost forgot, 3d model of this adapter can be found here
https://www.printables.com/model/1711895-cardputer-adv-adapter
(with ender 3v3 and basic PLA I found it best when printing on 75% speed, nozzle temp 212 bed temp 63)

Firmware for cardputer can be found here
https://github.com/rimedag/Vaultpi-Cardputer

But as of now it is buggy and probably not gonna work properly (there is source code and firmware builded for you to put on SD card and install)
I am giving link just if you want to bookmark it or save for later

Lets have fun

I already have my setup running on a Pi Zero 2 W and a custom firmware on Cardputer ADV that controls it.

So I decided to merge everything into a single device.

I designed and 3D printed an extension piece that sits between the Cardputer and its original back cover, I want to keep the stock shell on the outside for a cleaner finish instead of a fully printed case.

Current plan

• Fit Raspberry Pi Zero 2 W + battery + power module inside the extension
• Keep Cardputer fully functional (screen/keyboard/UI)
• Use Cardputer as a terminal/controller for the Pi

Power architecture

• 1x LiPo battery (limited to ~80x30x6mm -> probably ~1800–3000mAh)
• 1x power module (IP5306 or similar)

Wiring:

Battery → power module → 5V rail
                        ├── Pi (direct 5V)
                        └── Cardputer (5V0 pin)

• Cardputer internal battery stays
• External 5V acts like “always plugged in”

Communication

UART via Grove:

Cardputer TX → Pi RX (GPIO15 / pin 10)
Cardputer RX → Pi TX (GPIO14 / pin 8)
GND shared

Hardware mods

Raspberry Pi

• Remove bulky connectors (USB + HDMI) to save space

• Power Pi via test pads:

  5V → PP1 GND → PP6

(no USB needed at all)

Cardputer

• Inject 5V internally:

  5V → 5V0 (or 5V1) pin GND → GND

• Grove used only for UART

Goal

• One device
• One charging port
• No external cables
• Instant-on terminal for my Pi control center

Next steps

• Find largest LiPo that fits (80x30x6mm constraint)
• Pick stable 5V boost/charge module (IP5306?)
• Finalize internal wiring layout
• Clean enclosure routing (battery + Pi stacking)

If anyone has:

• better power module suggestions (compact, stable under Pi load)
• experience running Pi Zero from small LiPo setups
• cleaner way to inject power into Cardputer

I’d love feedback before I lock the design.

p.s. once cardputer zero comes out I will use this as a backup device

in case you wondered if it would fit YES IT WILL!

I had to take of connector from gps module and solder wires directly and I had to adjust the housing a bit to fit the cable and module