Part of what makes these slightly hard to find, I think, is the rarity of the sockets. The NES is now over forty years old and I doubt anyone's caused a factory to produce more parts for them since at least the year 2000, so anyone making commercial adapters will be doing so in very small quantities.

Mayflash, for example, seem to have given up on that approach entirely, and their adapter uses a 9-pin connector and (I suspect) a butchered extension cable to make it work: https://www.mayflash.com/product/PC053.html

So, the goal for this mini project will be an Arduino 32U4-based adapter with a couple of 3D printed NES controller sockets on it. That could be replicated for well under AU$10 by someone with the right machinery and skills if someone (I'm the someone!) wants to burn some spare time in making a design that will work, and most of that cost is in the Arduino-compatible board - the rest is cents of plastic.

Worst case, I'll build a few with the official sockets, and make those available too.

It'll be weeks before I have any electronic parts to play with, but for now I have a 3D printed socket that perfectly fits the plug on one of my NES controllers. A 0.4mm nozzle isn't fine enough to print walls between the pin shrouds in the connector, so my version will look open like this.

Do you have a 3D printer, a NES controller and half an hour to kill? DM me, I'd love to test this part on other printers and with other random controllers people have in the cupboard.

#3DPrinting

I think, with a 32U4 Pro Micro clone, I can support up to four controllers with all pins wired up. That's clock, latch and 3x data lines per controller, which is 20 in total, but I think I can make the latch pin common to all ports - that's the one that causes controllers to reload their states into their shift registers for reading.

I think the MVP for my adapter will be two ports - both the Advantage joystick and Four Score adapters have two plugs - but a four port variant should be possible.

Reading from a NES Four Score looks trivial.

For a regular controller, which is basically an 8-bit shift register, after strobing the latch pin, strobing the clock pin 8x will reveal its D-pad and button states, and you do the same again on the second port for the second controller.

The Four Score appears like a 16-bit register, so you flick the clock eight times to read player 1's pad, then another 8x to read 3. Then you do the same to port 2 to get players 2 and 4. https://www.nesdev.org/wiki/Four_player_adapters

The Hori 4 Players adapter (15-pin "Famicom Expansion" plug) looks identical to the Four Score in terms of protocol, and while nobody appears to have documented exactly what the NES Satellite does, I'd be very surprised if it did anything differently at all.

Satellites are also not particularly difficult or expensive to get, either. I foresee a new mini collection brewing... but I'll force myself to wait and validate my NES MAX pad working with my adapter before I start acquiring more hardware.

Here's my vision for the NES socket: a #3DPrinting frame for wires bent at a 90 degree angle to be pushed into, with one end pointing out into the socket for a plug to slide onto, the other out of the frame to go through holes in a PCB to be soldered down.

I will also make some sort of jig for precisely bending... some sort of wire. I know not to use anything galvanised as the fumes from soldering it is toxic, so my first attempt will be with 1mm copper jewellery wire - any other suggestions?

That settles it - the pins in a #NES controller socket are of 1mm thickness. This validates my measurement from the photo I saw.

This is with the current reproduction socket, of course; there is still the guy who measured his at 1.2mm (method unknown), and the other guy who made his own controller using sockets that took 1.3mm pins. It's possible the original hardware used thicker pins, but the 1mm jeweller's wire I bought on a whim seems to make a solid connection in this OG controller plug.

The pins in the socket itself are on a 4x4mm pitch, the two rows of pins that go to a circuit board seem to be intended to be 3mm apart.

The spring clips seem to need a 4mm hole. The centre line between the two clips seems to be 10mm from the centre of the nearest row of pins, and the clips themselves seem to be 20mm centre to centre.

Reverse-engineering stuff made to metric measurements is so much more pleasing than those made in old money. Japanese hardware just hits different.

It didn't escape me that 1mm pins are thicker than what I'm used to - if you buy typical 0.1" header pins, you'll get ~0.6mm square pins that perfectly fit the apparently-default 1mm through-holes.

I'm *fairly* confident 1mm round pins will fit into 1.1mm PCB holes, but I know for sure they'll fit the next size up in software which is 1.5mm. I can also set the holes to whatever diameter I want.

I need to order a board that's just different pin pitches and hole sizes to test parts on my desk.

And that's it - another Very Specific Adult Fingerpainting Adventure (or so it feels) on its way to manufacturing.

I've heard submarine warfare being described as weeks of sheer boredom punctuated by moments of extreme terror. Hobbyist electronics is similar - hours of hyperfocusing, interspersed by weeks of "aww, still in Memphis".

A quick test print with holes that are 1.0, 1.1, 1.2, 1.4 (because I can't count) and 1.5mm in width. The goal is to find a sweet spot I can push a 1mm pin through - when you 3D print a circle, it's not exactly the width of the geometry; it'll shrink a little bit, so a 1.0mm hole won't actually admit a 1.0mm pin through it.

It looks like I nearly under-estimated it - 1.5mm seems like the goldilocks zone, anything lower would require significant force and it just needs to sit/hold.

#3DPrinting

Okay, I think that's it for now.

The next major step is to validate the PCB I designed with the sockets from AliExpress, and use that to make sure I can actually make a working NES adapter (and breakout board) of any kind - that board is a couple of weeks away.

I've also ordered some 1mm PCB pins that should be the right size for my 3D printed sockets - with those I can design my own custom NES socket breakout and everything.

Time to box this project up for another couple of weeks. 🌛

Here's a thought - once I'm up to the point of plugging a NES controller into an logic analyser and an Arduino to take a look at its communication and programming something to read it, would anyone be interested in watching a livestream of that happening (or even the recording after the fact)?

Would a video of the process of reading basic stuff from an analyser and using that to write a program to talk to a thing be of interest to folks here?

Pleased to announce I am apparently competent at operating calipers and Fritzing. These aftermarket NES sockets from AliExpress appear to fit perfectly with my boards, and I should now be able to start talking to NES controllers.

Don't ask why there's no D1 or D2 pin - the pinout on nesdev.org shows D0, D3 and D4 for this connector: https://www.nesdev.org/wiki/Controller_port_pinout

There are days I look at the crap spread over my desk and think, man, I really have no idea what I'm doing in this hobby.

Then there are days I look at the tiny spiky hovercraft sculpture I've just willed and soldered into being, and think, man, I really have no idea what I'm doing, but this looks kinda pretty.

Lessons learned so far: The extra "hackable" pins need to be slightly further away from the Pro Micro footprint so they're easy to plug in to, and I need to put pin labels outboard as well.

What's going on here is pins 5, 6 and 7 on the Arduino, which are clock, latch and data for the NES socket, are also connected to channels 0, 2 and 4 on my logic analyser. Channel 6 (pin 10 on the Arduino) is just an extra line I can take high to trigger captures in PulseView.

Is this #NES controller busted or am I just doing something fabulously wrong? The data line just echoes whatever the latch line does, maybe suggesting the two are bridged somehow, but is this something a shift register might do in some circumstances?

Data line should only go high if I'm holding a button at an appropriate time in the clock line's cycles, I'm not holding anything in this capture but even if I do there's no change.

Data is just high the whole time without the pad plugged in.

The $10 pad arrived, and is pretty terrible: https://digipres.club/@timixretroplays/114296167646130444

It does work, though - here's my prototype program's output while I'm pressing A, Start and right on the D-pad.

I'm feeling confident enough about this project now to try ordering PCBs to match my 3D printed NES controller sockets.

I found a plugin for Rhino 3D that lets me export geometry to SVG, so I can print the exact outline of my 3D parts on the PCB very easily - how cool does this look!?

My NES breakout boards arrived - I've soldered one up with my 3D printed socket and I'm about to give it a go. Even if this doesn't work at all, it's still a good run of the process and I'm very happy with how it's gone and what I've learned.

The extra big hole in the corner is for an M3 screw to hold the 3D printed socket to the PCB, so yanking the controller's cord doesn't cause the socket itself to slip over and off the pins.