stuntstein.dk

Tuesday, March 17, 2026

Arcade Machine

This started in February 2017. It is my own design of a bartop arcade machine. Heavily based on a design found on the internet.

The build

Made out of MDF and some 1x1.5" wood. Very simple.

Joystick controller

I purchased a cheap joystick controller, named Xin Mo.

It is by far perfect but got the job done, sort of. First issue was the Linix kernel didn't recognized the controller. I found a patch but had to apply it everytime I updated the kernel. Finally I submitted a change to the Linux kernel. My one and only Linux commit :-)

https://github.com/torvalds/linux/commit/9257821c5a1dc57ef3a37f7cbcebaf548395c964

Over the years I noticed that sometimes the controller would register multiple inputs from the joystick. It wook me 9 years before I decided to fix it.
I added a little RC filter on each input. 120nf + 13KOhm to filter out the rebound from the micro switches. Working really great now.

Arcade OS

I started out using RetroPie. This worked great and you can customize everyting. This has been running for 10 years on the machine.

It was time to try something else so it is currently running Batocera. This is working really well and starts up fast.

ToDo

Wanna make a custom intro video some day.

Some random photo of the arcade in action

Hobbywing XR10 PRO G3

This has been by far the most challenging ESC to repair. So many components on this device.

It wouldn't turn on. Measuring the power side everything looks fine. Later I found this was not true, because the power board has a 3 mosfet ground switch. Why? The negative battery connector is isolated from the ground on the boards because of this switch. So it turns out one of the low side mosfets on port A was shorted. All the mostfets on the power board is tphr6503pl.
I didn't have a 1:1 replacement but has some AON6512 laying around from another broken ESC.

The mosfet closest was shorted between gate, source and drain

But that was the easy part to find. The bigger problem was the logic board wouldn't turn on.
Usually there is a latch circuit that once you press the ON/OFF botton the circuit would latch the power circuitry until the CPU turns it off via a gpio/transistor.
Going over the main components to investigate:

There is a buck converter which is power by a 12V boost converter. Odd to first boost then buck and the 12V boost doesn't seems to be nearly as powerful as the buck.

A 7805 regulater which is power via a single low power SOT23 PNP transistor. Agains seems odd to have a 2A regulator powered by a 500mA transistor. The transistor may also be a P-channel mosfet. I couldn't find any datasheet for this.

When pressing the ON/OFF switch the base voltage to the PNP transistor is driven low. However the whole circuit wouldn't stay locked meansing powered on. Feeding power directly to the 7805 I was able to turn on the cpu to confirm it was working. If the CPU is bad then the it is a no-fix.

Power directly to 7805 turns on the CPU

The emitter of the PNP is connected to the positive of the battery. The collector did go high but only to around 2V. After check all around I determine that it has to be the PNP that was bad. Replacing with a common PNP it is now working :-)