Turning the Raspberry Pi 5 + 3.5″ LCD into a Portable OTS Cyberdeck

In my previous post, I walked through assembling a Raspberry Pi 5 (8GB) with an active cooler and a 3.5-inch resistive touch screen inside a compact enclosure. Once the core hardware was up and running, the next logical question was simple: how do I turn this into a self-contained, field-ready portable computer without custom 3D printing or complex soldering?

Keeping with the same 100% off-the-shelf (OTS) philosophy, I have put together a complete pocket cyberdeck loadout inside a hard-shell carrying case. With this hardware build, a quick swap of a microSD card, a swap of USB hardware peripherals; this compact kit shifts between a portable Linux terminal, a standalone IoT demo server, and a digital workstation for amateur radio.

Just look like another portable gaming console setup
Inside the portable gaming console setup

Additional OTS Components & Accessories are Keebdeck wired keyboard and Hagibis electronic organizer.

Continue reading →

Adding a 3.5″ LCD to Raspberry PI 5

End result, a Raspberry PI 5 with 3.5″ inch touch screen
Size of the Raspberry PI 5 with 3.5″ touch screen compared to iPhone 14 Pro

This is a relatively simple project; all the parts are off-the-shelf (OTS) or in simple language it can be procured or bought from any electronics store or webstore.

This project requires understanding of how to load Raspberry PI compliant OS and any PI OS that is compatible with driver of the 3.5″ touch screen LCD. Familarity to use command shell or terminal or SSH shell.

Continue reading →

How to use AI to create android app

Android Studio 11 Medium Android Phone emulator
Finally a customed WordPress Client developed using AI Assistance and AI Agent in the flesh

The biggest hurdle of creating an App for android device is learning how to use its tool chain.

This is how I did it using AI and AI Agent or AI Assistance.

Continue reading →

Smoking gun that AI will lie to you, gaslight you, and tell the half truth

It all started with a small experiment to restore an image I have using ChatGPT.

No argument on the capability of ChatGPT to enhance image

The curiosity as an engineer that profess system thinking wanted to know more on how ChatGPT enhanced the image. So I asked a follow up question “are the image rebuilt using known images that matches this image ?”

ChatGPT trying to lawyer its user with very eloquent gaslighting

The answers feels almost perfect for folks that does not have domain knowledge about digital image. To top of that from my post graduate degree I learned an important concept about AI or Machine learning. The quote from Tom Mitchell “A computer program is said to learn from experience (E) with respect to some class of tasks (T) and performance measure (P), if its performance at tasks in (T), as measured by (P), improves with experience (E).“

Experience (E): The data or learning experience the program receives.

Class of Tasks (T): The specific task or class you want the computer to perform.

Performance Measure (P): The testing and evaluation phase used to verify if accuracy is improving.

In this case I realized and very aware that the glare on my photo has no more data, which makes it impossible to be restored without understanding of the logo and character. To dig deeper I need to use to proper keyword “data wipe out from the glare” in the prompt.

ChatGPT acknowledge the challenge posed to it

Fair explanation on the first part. Until you realized it is using lawyer tactics to misdirect and to eloquently gaslight in the second part of the explanation.

ChatGPT long version of agree to disagree

I can see that ChatGPT uses a wily constructed sentence used by a sleazy colleague or boss to project power. Gaslighting and dismissive of your counter point. ChatGPT would be able to convince folks that has no domain knowledge.

Conclusion, make sure you do a “doverjai, no proverjai” when using of AI or “generative AI“.

Using ESP32 to upload data into ThingSpeak

ESP32-Wrover does have SPI or SDMMC, that will require additional purchase of modules for the microSD or other SD physical card module.

The need to create proper data acquisition (DAQ) using ESP32-Wrover will be easier by leveraging free service offered by ThingSpeak from Mathlab.

From the previous project https://chow.karmeng.my/2025/03/28/programming-esp32-wrover-with-dht22-am2302-humidity-and-temperature-sensor/ , I am intent to make use of the light sensor. L1B01 light module to collect data of ambience light.

L1B01 light sensor
Continue reading →