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.
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.
Attached: 2 images My AU$10-including-postage brand-free NES controller has arrived. The cable is heavier than the gamepad itself and only four of those six holes on the back have screws in them. It feels pretty awful and I'm actually mildly surprised the case isn't 3D printed. I would be surprised if the PCB has more than 1.000 layers. There should be absolutely nothing inside but the bare minimum to communicate up, down, left right, B, A, start and select to a NES, and that makes it perfect to test my adapter.
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.
Um, okay - it works perfectly! This is a 3D printed socket I modeled in negative from the plug of a NES MAX controller, with 1.0mm PCB pins soldered to a breakout board of my own design.
It was a bit of a struggle to plug in this $10 repro controller - it was clearly designed for the tolerances of the original console's socket, which my breakout here does not at all have - but the dead MAX's plug slides on flawlessly.
Time to collect a *working* authentic NES controller and validate with that!
I ended up collecting a used NES-039, the new-style or dogbone controller, and that worked perfectly. I've since collected a Four Score as well, here's a thread on that: https://digipres.club/@timixretroplays/114494458860844483
It's now time to design and make a dual NES socket with matching breakout board to match this bad boy.
Attached: 2 images Time for a closer look at 1990's version of a USB hub. This is #Nintendo's Four Score, a 4-port multiplexer for the original #NES. Mine is the NES-034A model, which reportedly says "NTSC/PAL" on the board; there is also a NESE-034 variant, which I think is the weird Spanish type of NES hardware with extra resistors and diodes as an early region locking mechanism - mine should be universal. I collected one sealed new-in-box for a couple of research reasons, which I'll get to. 🧵 #retrogaming
Unexpectedly, the socket pitch on a NES seems to be 19.5mm - I had fully expected it to be 20, since I've seen a lot of very neat metric measurements in these plugs and sockets already.
I know the spacing here will match the spacing on an original NES console because the Four Score can physically plug into itself - that wasn't just a silly thing to do for a funny photo.
Well, this is the first pass to replicate that geometry, anyway. I don't have time to 3D print this tonight, but that's the next step.
The two screw holes will be used to firmly mount this socket to the PCB.
The Four Score's double plug perfectly fits my 3D printed double socket. Next up is a circuit board to match.
The two plugs on that cable actually wiggle around slightly - it's clearly the same single-plug part in a different housing. So while I've measured the Four Score's sockets at 19.5mm apart, this plug would also certainly work on 20mm pitch sockets.
All of which makes me very keen to see what the NES Satellite's receiver plugs looks like.
And the PCB for this I actually designed last night, on the assumption my socket would fit. I'm choosing not-the-cheapest postage option this time because I'm sick of waiting weeks for projects to advance in between steps.
I've broken out the power, latch and clock signals in pairs so I can experiment with bridging them together - I'm fairly sure the original NES hardware had them all bridged, but maybe I'll discover a reason why it might be useful to page controllers separately at some point.
Five boards ordered on Tuesday, arrived on Monday - not bad for AU$20-ish. Would probably have been here last week if I hadn't opted for black PCBs out of vanity - I might just leave prototyping stuff to the default green in future, which apparently is quicker to turn around.
Anyway. My dual NES socket breakouts have arrived - they'll be a little work to fully assemble and solder up, but they look glorious so far!
First time trying this out. I originally thought I'd be putting the pins into the back of the 3D printed socket and slotting that assembly onto the circuit board, but the fit between the pins and those holes is so precise I had to carefully line them up and pop them into the PCB one by one. I'll be printing a little tool to make doing this by hand easier in future.
The socket screws to the PCB in two places for maximum stability and to stop NES pads from yanking it off the board if they stick.
I have had to partially break it back down again to make it work on my little board soldering attachment. The exact order of operations for assembly is a work in progress.
Again, this is just my little remix of @rasterweb's little soldering board: https://www.printables.com/model/598207-prodoehls-solder-board-v2
Looks like I'll need to make a bigger one to do stuff like this in one go!
And finally, the whole shebang - an #Arduino Pro Micro clone, my own custom dual #NES controller breakout with a #3DPrinting socket, and the #nanoDLA logic analyser ready to help me learn how to talk to a NES Four Score - and later, to help document how to do the same with a NES Satellite!
This is the confluence of a decade plus of hobby learning and making.
A note mainly for myself: CLK and LAT are pins 5 and 6, and channels 0 and 1; P1D0 is pin 7 and channel 2, P2D0 is pin 8 and channel 3.
The straight-through NES sockets have arrived from AliExpress. They are very clearly using the exact same shell as the 90-degree ones, just with straight pins instead of bent ones.
Unfortunately, while the 1mm pins in these connectors would individually fit the 1.1mm through-holes in my breakout board, they won't do so all together - so I'll need to order a separate version of this breakout board to test off-the-shelf sockets like this. That might work better for my 3D printed sockets, too.
With 1.2mm holes, and the extra pins clipped off, the AliExpress NES sockets sit comfortably and snugly in my prototype breakout, with the correct spacing for a Four Score's double plug. This project is now waiting on the arrival of my NES Satellite to validate that plug fitting as well (I think it's more rigid than the FS's so less room for error).
This might be the end of the road for my 3D printed socket, just for now - I'll focus on the better quality outcome for the V1.0 #NES breakout.
My #NES Satellite has arrived - and what a beautiful chonker of vintage tech it is. The entire right half of it is just space for *six* C-cells. Tapping the side of this thing makes it ring like a bell due to all the giant springs in the battery bay.
It looks like my dual NES breakout needs its sockets about half a millimetre closer together - this would make the connector pitch 19mm, vs the 19.5mm I measured on the Four Score. I'll order another board to test this.
Soldering to pins on a 4mm pitch is actually kinda zen. I'm really pleased with my consistency there.
In other news, turns out a 19mm pitch between connectors feels perfect for both the Four Score and the Satellite. I think this is the final version of my dual NES breakout, unless coding for controllers reveals some terribly incorrect assumption.
And here's my setup for implementing and testing support for controllers on the Four Score.
Latch (yellow) is pin 15, ch 7.
Clock (green) is 14, ch 5.
Port 1 data (blue) is 16, ch 3.
Port 2 data (blue) is 10, ch 1.
Purple (pin 9, channel 0) is so I can trigger an interrupt for the logic analyser - I can take that line high, wait for the analyser to notice and start capturing, then start flicking latch and clock to grab data.
Tonight's result. This is a #NES controller plugged into the 4th port on a Four Score, with Up and Right held down - that's the first two dips in P2D0 in the screenshot.
The dips in P2D0 and P1D0 at the 18th and 19th clock pulses respectively represent the signature of a Four Score, exactly as described here: https://www.nesdev.org/wiki/Four_player_adapters#Four_Score
And see that very short section below, about the NES Satellite, where it says "specific details are not yet known"?
That's this weekend's mission.
I did plug my Satellite's receiver into this contraption and run the logic analyser capture again, but not much happened - both data lines were held low, which was interesting, but there were no ID pulses like I saw with the Four Score. Possibly the Satellite does something different entirely, or only produces output when it's got a connection to the transmitter.
I bought this Satellite specifically because the seller said it was tested working, so I'm not expecting to find I bought a dud.
I'm home sick today, and this stuff is all still set up on my desk, so I'm asking the question: What did the 2/4 switch on the Four Score actually do?
The manual just says to switch it to "2" if you have one or two players, or "4" for three or four. But it doesn't actually explain why, or what the consequences might have been if you got it wrong.
The troubleshooting table says if the game can't be controlled, the switch might be in the wrong setting. Time to find out what that means, exactly.
Oh, okay. If you flick the switch to "2", it just passes ports 1 and 2 straight through to the console, and ignores 3 and 4 entirely, bypassing the Four Score's logic and protocol.
That'll be for compatibility - according to the manual, the Four Score only works with the original controller, the MAX pad and the Advantage joystick, so if you wanted to plug a light gun or dance pad into the Four Score, it has to be in passthrough mode.
The NES Satellite has a "Mode Switch" that's labeled CTLR/GUN on the unit itself, and the manual says it has to be in the GUN position for the NES Zapper to work. I bet it's identical to the 2/4 switch on the Four Score, and just dumbly passes through inputs on the first two ports.
It'll be interesting to see how the wireless comms work. Does the Satellite receiver itself flick the latch and clock signals for the four ports? How are they encoded into IR LED blinks? That's a question for later.
Did you know AA batteries are the same length as C ones? This cursed fact means you can 3D print simple adapters to install one in the others' place - and it meant not buying AU$17 of batteries I'll use a handful of times, ever, or waiting for cheaper ones to be mailed to me.
It was not fun getting them in there, and don't ask why the 3D printed parts are so shocking, but my NES Satellite lives!
And yup, it works. Screenshot shows me pressing up and A on the controller plugged into port 1 of my Satellite.
This is probably the first time in decades anyone's looked at data coming from a NES Satellite this way, and... it's not quite what I expected. No signature blips like the Four Score, the data lines just kinda die after the 24th-ish clock pulse without ceremony.
Ok, so:
- with one or zero controllers plugged in, both data lines go high for 24 ticks. Player 1 input works entirely as expected, except that holding right (the last bit transmitted) leaves D0 low, like it's not expecting any more bits on that line (from, say, P3).
- With a pad in port 2, P2D0 drops after eight ticks, so it's detecting there's something there.
- In no case do I see inputs from pads in P3 or P4. I thought it might need tighter ticks, but that hasn't made them appear.
- It behaves the same way regardless of whether the mode switch is on CTLR or GUN. I think I need a Zapper now to test if there's any difference and if this switch even works.
- Maybe it's like SEGA's 6-button protocol, and flicking the latch again within a frame will page across to P3/4 input?
- Two of the top ten google results for "nes satellite protocol" are posts I've made in this thread. I've @foone'd myself.
The manual (https://archive.org/details/NESManuals/NES%20Satellite/page/n1/mode/2up) does say that "the NES operates differently in the three or four-player mode than in the one or two-player mode. Because of this, you should only plug in the necessary controller(s) for the game you are playing". Also that the Zapper (in GUN mode) and the Power Pad only work in port 2. So it's detecting what's plugged into #2 in case it's a Power Pad.
It does not describe *how* it acts differently in 3/4 player mode, or what might trigger that.
Well, this will come in handy when I'm up to #SNES stuff: https://www.backoldgaming.com/console/super-nintendo/entretien-super-nintendo/read/super-nintendo-service-manual-snes
This guy has a very tantalising, low res photo of an original #NES technical manual, but no links or much more information about it other than a list of contents: http://www.limbofunk.com/nes-technical-manual.html
I think that would contain exactly what I need. Too bad it's not 1993 anymore, there might not even still be a copy in my country. I will shout anyone who can help me find a copy, physical or digital, many beers.
I've never successfully found anything of use by directly searching the Internet Archive. If I've cottoned on to a specific website once existing with the information I want, sure; but searching it directly, no.
Take this for example: https://web.archive.org/web/20161025144738/http://abracing.us/book-name/super-nintendo-entertainment-system-technical-manual.pdf
It's one of the first results for "nintendo technical manual" searching in the PDF section. It's a 3-pager SEO spam PDF file full of links to a long-dead website that probably never even had the file to begin with.
In fact, here's exactly the same look-and-feel PDF for exactly the file I want, for a completely different site that was never archived either: https://web.archive.org/web/20170629175651/http://improvechiparks.com/openbook/download/nintendo-entertainment-system-repair-manual.pdf
There's some tantalising strings in there that look like the serial number for the manual, which might be further searchable, but I think that was auto generated as well.
I don't blame IA for this - it's an awful problem to solve - but there's so much spam in there it's impossible to even know if what I want is in there.
I have my answer. Nintendo's own 1990 "World Class Service" Authorised Servicer Technical Training Workbook says I should replace the CTLR/GUN switch or check the soldering on ports 3 and 4.
The seller said my Satellite was working - fair enough, he probably plugged one pad into it and got to world 1-1.
I found this manual thanks to a reddit post that mentioned Avery Beckett's Nintendo collection on the Internet Archive: https://archive.org/details/@avery_beckett?query=nintendo
The Workbook is here: https://ia801603.us.archive.org/15/items/nintendo-world-class-service-technical-training-workbook/Nintendo%20World%20Class%20Service%20Technical%20Training%20Workbook.pdf
No major progress on this today. I did poke around in my Satellite with a multimeter, one pair of pins under the mode switch goes from zero resistance in one position to some resistance in the other. Without another known-good one to compare with, or any in-depth technical manual, I have no way to tell if that's normal.
I have ordered a NES Zapper - if that's readable in port 2 with the switch in both positions, that'll prove my Satellite is faulty and I'll need to collect another.
@timixretroplays sometimes I find child/leaf node devices wait very passively for an all-call inatead of reporting in. It depends on the dev.
I HATE that. It's so hard to work with.
@timixretroplays hey do you have a reliable* source for NES and SNES controllers? I need one of each for… shenanigans…
* not necessarily fast, preferably cheap, and non-OEM is fine
@jpm it depends. If you just want *something* to plug into a thing and see if things work, go on eBay and buy the cheapest shittest thing with the right plug on it.
I've collected a weird bunch of coconuts over time. I have a dogbone, a broken MAX, that awful $10 pad, an official NES joystick made by Gravis, and a Retro Fighters JAB - the latter is $30 from The Gamesmen, also features a USB plug, and is a *really* nice controller to hold (I don't think I've posted about it yet).
@timixretroplays the basic way to do it on a microcontroller is vaguely-pseudocode:
set-gpio(LATCH, HIGH);
wait(1ms);
set-gpio(LATCH, LOW);
read-gpio(DATA); // data 0 may not need a clock
wait(1ms)
loop 7 times {
set-gpio(CLOCK, HIGH)
wait(1ms)
set-gpio(CLOCK, LOW)
read-gpio(DATA)
wait(1ms)
}
The final wait inside the loop is VERY IMPORTANT, without it you’ll put the clock high for 1ms, put the clock low, then almost immediately put the clock back high again.
@timixretroplays now there’s a bit-shifting trick here that you can use to populate a variable full of bits as they are read. Vaguely pseudocode:
unsigned 8-bit int variable = 0
Counter = 0
loop {
Variable <<= 1 // left-shift 1 bit
Variable |= GPIO bit // logical-OR to set lowest bit
counter++
} while counter < 8
@timixretroplays there’s an extra-extra trick here too, if your variable can hold more bits than the number of bits you want to read then you can set the original value to 1, and when the 1 gets to a certain point then you’ve read the number of bits you want!
unsigned int16 variable = 1
loop {
variable <<= 1
variable |= GPIO
} while (variable & (1 << 8)) == 0
@jpm the first data bit does indeed not require a clock flick - it appears the instant latch goes high, and the rest of the bits appear when clock goes low. In this capture I'm holding A (the first data bit) and Up (the fifth bit).
Also for my testing at the moment I'm using a 1 millisecond delay between changes for consistency, but the code I wrote weeks ago has 6 microsecond pauses in between - from memory that's as low as I got it to consistently work.
Tim 🎮
Jake in the desert
I love this, so I
Krystian Majewski
aeberbach
Craig P