Goal

The main objective with this project is to design and build a small form factor Linux computer, let's draw some high level requirements :

It has to be a true Linux computer and not just an other Android Smartphone (Gemini/Cosmo from Plant Computer, PinePhone, Astro Slide...)
It needs to be powerful enough to run a Linux Desktop environment (while there is no need for it to be comparable to a smartphone level of performances)
It needs to have a decent display while not being too high resolution otherwise it will be not confortable to use (Sharp Netwalker main issue)
It needs to be small (pocketable)
It needs a keyboard (that's probably the most difficult part)
It needs an USB-C for charging
It needs at least 1 USB-A for accessories
It needs to be cheap (99$ should be achievable)
Two version should be designed to cover most users needs : 5” and 7” (similar to a Sharp Zaurus for the 5" and a HP Jornada for the 7")
Nice to have :
- GPIOs for expansions boards & more features
- High quality keyboard
- Wireless charging
- It’s own Linux Distribution optimised for the hardware & screen resolutions

While I have a clear idea of what I would like to achieve I do not have the skills to design such hardware from scratch appart from the basic understanding of the needed components and how they are connected to each others. That’s why I hired Electronics Engineers to help me design and produce the first prototype:

Adam Nowak : https://www.linkedin.com/in/nnowakadam/
Marcin Turek

They did an amazing job and suggested improvements to the design and embedded features, they are also looking forward to work on the next iteration of the project.

A bit of history

Here is a list of the few devices from which this project will draw inspirations from, they are all very distinct and were released in different eras but they were all great pocket computers.

Psion 3a

Psion 5Mx

HP Jornada 720

Sharp Zaurus SL-C700-750...

I don't think I need to present any of these devices. While the Psion 5 has one of the most recognisable keyboard (it is membrane based but looks like a mechanical) I don't think it had the best typing experience, the Psion 3 was probably similarly conformable to type on while being very different (membrane based with spaced keys). The HP Jornada had probably the best typing experience out there but they were super THICC devices. The Sharp Zaurus were my favourites, super small, running a real Linux operating system with a gorgeous display (at the time), but their keyboard was quite bad. The end goal is basically to take the best features from all these devices and combine them into a single pocket computer.

Master Plan

Step 1 : Create a first prototype with a 7” display to define project feasibility and identify possible issues
Step 2 : Share this first status update to the community & gather feedback & ideas
Step 3 : Design a second prototype most likely in 2 sizes around 5” and 7” displays
Step 4 : Design a great keyboard & a nice looking shell once the hardware has been finalised : get that thing as slim as possible
Step 5 ; Run a first batch of 50-100 devices, share with the community, gather extra feedback & ideas
Step 6 : Mass Produce (if there is demand)

Articles & News

- Hackster.io

- Tom's Hardware

- Its Foss

Project Logs

Collapse

Step 1 : First Prototype
Icepat • 06/22/2024 at 10:10 • 0 comments

Like explained earlier for the keyboard we are using part of the Ti-92 layout (and silicone membrane & keycaps) to speed up development & reduce cost (two other Hackaday project were used for the PCBs, linked in the Project details). The keyboard is managed by an Atmega32U4 directly connected to the Pi's USB-OTG port (there is pins under the Pi). The Atmega32U4 is visible by Linux as an USB-HID device but we still have to write the software for it (currently waiting to get a Ti-92 from Ebay).
There is room on this version to have a touchpad area, we could integrate this later if there is demand for it (cost should be low). On top of the keyboard we have some action button and a d-pad.
The display is a 7" 1024x600 24 Bits DPI RGB with a Capacitive Touch Panel (GT911 Controller) connected directly through the PI's GPIOs, reducing energy consumption & costs as we don't need extra hardware to drive these kind of displays apart from the backlight controller. The only "problem" is that this display takes up all the Pi's GPIO's so there is none left available for accessories. And the touch controller can't be connected to the Pi directly so we connected it to the Atmega32U4 act as an adapter between the touch controller and the Pi. A similar project is available here and we will most likely base our work on this code (not done yet).
Here is the back of board, with a large emplacement for the Li-Po battery, the connectors, flex cables and the Pi itself. There is not many components and it doesn't take much space so we can definitely create a 5" version of this same board, the only difference will be that the display connector will be 40 pins instead of 50 and that we would have an extension port somewhere with the 8 available GPIOs left.
Here is most of the power management components, handling the main power of the boards, the Pi, display's backlight, battery charge/boost.
Here we can see the 50 pin display connector that's directly connected to the PI's GPIOs, the Capacitive Touch Pannel connector that's connected to the Atmega32U4 present just next to it, which also handles the keyboard and key presses and will send all this over USB-HID to the Pi through the USB-OTG connector (connected to it under the Pi itself).
Here is a video of the prototype booting a Raspberry Pi OS Lite install customised with XFCE & Conky. There is probably many ways to make that thing boot way faster, nothing has been optimised at the moment. The Pi Zero 2W is definitely not the fastest out there, but it is very cheap, available anywhere, has great support and it has a DPI Display interface reducing the overall cost of the computer.
The display is quite good for the price (15$ with the Capacitive Touch) has acceptable viewing angles, contrast and colours. The Linux desktop could be more optimised, some windows will be too big to fit in that screen and it will be more of a problem on the 5" with the 840x480 resolution. But the goal here is not to play 4K HDR videos, but to have a super portable hacking linux computer.
Here is a few pictures to get an estimation of the size compared to a HP Jornada 680 and a Sharp Zaurus SL-C7500c. This 7" version of the prototype will be a great mini linux laptop, while I believe a 5" version (comparable to the Sharp Zaurus in these pictures) will be the ultimate pocket hacking device especially with the 8 GPIOs (& I2C) available for expansion.
Design
Icepat • 06/22/2024 at 09:23 • 0 comments
Following these requirements, here is the hardware that was picked for the first design :
- Raspberry Pi Zero 2W : While not being the fastest SBC out there it has great advantages
- 7” 1024x600 DPI RGB Display 24 Bits + Capacitive Touch Screen
- Keyboard : Atmega32U4
- Power Management
- Connectors
What did we learn ?
- The connector we used for the PI is nice for a prototype but will most likely have to be replaced with something better for a production build
- The Pi Zero 2W is definitely a slow device, but the direct support of a DPI display stills makes it the most relevant option ATM
- The 7” display is nice but is also quite “big”. A 5” version of the device the size of a Sharp Zaurus would be definitely better portability wise
- Having direct access to flashing the Atmega32U4 through the USB-C would be a nice improvement
- We could add an HDMI connector BUT the PI doesn't support HDMI while the display is set on the DPI Interface, so even tho the device could output on HDMI the user would have to edit some stuffs in the config.txt and reboot to get it to work
We are currently using the TI-92 layout and will be using the silicone membrane & the keycaps for this phase of the prototype (reducing cost). It would be nice to have our own keyboard eventually, but I'm not sure what type we should have. Building a good keyboard is really hard and a low price point is still one of the main goal. I don't think we should go down the Psion 5MX/Gemini/PinePhone road as it will be too complicated and costly. O ordered a Psion 2 and will gut it to see how it is build, I believe it should be similar to what we have now just with larger keys and slightly different layout (still a silicone membrane based keyboard). Welcoming feeback & suggestions on this from anyone.