I bought 9 PS3 controllers for $40... how bad could they be?

While MattKC videos aren't the only topic of discussion here, they are probably the primary way people found out about this place, so here's a forum for discussing them.
Post Reply
User avatar
MattKC
Site Admin
Posts: 323
Joined: Mon Aug 22, 2022 1:05 am
Contact:

I bought 9 PS3 controllers for $40... how bad could they be?

Post by MattKC »

Merry Eggsmas everyone! I guess this is my first (inadvertent) Christmas special. It wasn't supposed to be, it was actually supposed to come out a few days earlier, but hey now I can call it the lump of coal I've left in your stocking EZ

Metadata:
  • This video was actually shot alongside the PS3 repair (there's no way I could have done all of this in between), and you can actually see some of the half-finished PS3 project behind me while I'm trying out the controllers.
  • These were just bought online, also over a year ago when I realized I had no controllers for the PS3. Likewise, they've sat around since then.
  • I really did buy 8 and receive 9. In all fairness, it took me a while to even notice that, so I can see how they would have miscounted by accident.
  • As I mentioned in the previous video's metadata, technically these are the real reason I repaired the consoles. I wanted to sell off another PS3 I had that I didn't need, but didn't have any controllers to go with it, so I wanted to fix these, and figured it made sense to fix the consoles first if I was going to turn them into videos.
  • Again there's a lot of time compression. The section where I'm testing them was filmed in August, and then the second part where I'm "reviewing" how the repairs went was filmed in December.
  • If you see ZIP codes in the video, don't worry, they're actually not mine! The screenshots were taken in a browser isolated inside a VPN, so it's just some place in the middle of Los Angeles (which is my closest server). I could have photoshopped them out regardless, but I figured since they weren't actually dox, it didn't really matter.
User avatar
12OVER12
Posts: 5
Joined: Sun Dec 25, 2022 4:59 pm
Location: yo
Contact:

Re: I bought 9 PS3 controllers for $40... how bad could they be?

Post by 12OVER12 »

Quick thing.
If you shake the controller and there's random button inputs it's because there's a piece of foam that deteriorates. you wrap it in electrical tape then it fixes it!
User avatar
andreizabest
Posts: 55
Joined: Sat Dec 10, 2022 4:49 am

Re: I bought 9 PS3 controllers for $40... how bad could they be?

Post by andreizabest »

I found this "original used" listing on Alibaba although you need to order at least 50 here
There's also this from zed labz but it's like 10$

I've also seen some eBay listings that advertise it as a 4 pin but in the photos it's only the 3 pin which makes me wonder if you can fit a 3 pin joystick to a 4 pin controller. Would it be possible to solder one or do the pins not even align properly? Maybe it can be modified to fit. Does the controller even make use of all 4 pins?
User avatar
MattKC
Site Admin
Posts: 323
Joined: Mon Aug 22, 2022 1:05 am
Contact:

Re: I bought 9 PS3 controllers for $40... how bad could they be?

Post by MattKC »

andreizabest wrote: Sun Dec 25, 2022 11:37 pm I've also seen some eBay listings that advertise it as a 4 pin but in the photos it's only the 3 pin which makes me wonder if you can fit a 3 pin joystick to a 4 pin controller. Would it be possible to solder one or do the pins not even align properly? Maybe it can be modified to fit. Does the controller even make use of all 4 pins?
I've seen a lot of this too, "4-pin" listed in the title, but then when you look at the photos or read the description, it says 3-pin. I have wondered about whether they can somehow be made to work, but some comments I got suggest that 4-pins use a completely different technology and that's why they're incompatible. Also as far as I can tell the pins don't line up so it wouldn't be easy to get it to align flush even if the signals were interchangeable.

EDIT: Someone just informed me in an e-mail that the 4-pin modules were actually exclusive to Sony controllers, a design they introduced mid-way through the PS3's life and then abandoned a few years later. I haven't found any info to verify that yet, but that would definitely help to explain why they're so rare.
User avatar
12OVER12
Posts: 5
Joined: Sun Dec 25, 2022 4:59 pm
Location: yo
Contact:

Re: I bought 9 PS3 controllers for $40... how bad could they be?

Post by 12OVER12 »

MattKC wrote: Mon Dec 26, 2022 7:34 pm
andreizabest wrote: Sun Dec 25, 2022 11:37 pm I've also seen some eBay listings that advertise it as a 4 pin but in the photos it's only the 3 pin which makes me wonder if you can fit a 3 pin joystick to a 4 pin controller. Would it be possible to solder one or do the pins not even align properly? Maybe it can be modified to fit. Does the controller even make use of all 4 pins?
I've seen a lot of this too, "4-pin" listed in the title, but then when you look at the photos or read the description, it says 3-pin. I have wondered about whether they can somehow be made to work, but some comments I got suggest that 4-pins use a completely different technology and that's why they're incompatible. Also as far as I can tell the pins don't line up so it wouldn't be easy to get it to align flush even if the signals were interchangeable.

EDIT: Someone just informed me in an e-mail that the 4-pin modules were actually exclusive to Sony controllers, a design they introduced mid-way through the PS3's life and then abandoned a few years later. I haven't found any info to verify that yet, but that would definitely help to explain why they're so rare.
You could say fuck it and get a bunch of shitty broken sixaxis controllers and use the 4pin boxes from that. since sixaxis controllers are useless 5Head
User avatar
andreizabest
Posts: 55
Joined: Sat Dec 10, 2022 4:49 am

Re: I bought 9 PS3 controllers for $40... how bad could they be?

Post by andreizabest »

MattKC wrote: Mon Dec 26, 2022 7:34 pm
andreizabest wrote: Sun Dec 25, 2022 11:37 pm I've also seen some eBay listings that advertise it as a 4 pin but in the photos it's only the 3 pin which makes me wonder if you can fit a 3 pin joystick to a 4 pin controller. Would it be possible to solder one or do the pins not even align properly? Maybe it can be modified to fit. Does the controller even make use of all 4 pins?
I've seen a lot of this too, "4-pin" listed in the title, but then when you look at the photos or read the description, it says 3-pin. I have wondered about whether they can somehow be made to work, but some comments I got suggest that 4-pins use a completely different technology and that's why they're incompatible. Also as far as I can tell the pins don't line up so it wouldn't be easy to get it to align flush even if the signals were interchangeable.

EDIT: Someone just informed me in an e-mail that the 4-pin modules were actually exclusive to Sony controllers, a design they introduced mid-way through the PS3's life and then abandoned a few years later. I haven't found any info to verify that yet, but that would definitely help to explain why they're so rare.
Looking at this website it seems that it might be the case
In ALPS webpage are only available sticks with 3 pins pots, there are two posible explains for this, maybe ALPS manufactured the sticks without pots and sony ordered this special version of the pots to other company then sony asembled the pots and sticks together.... or... ALPS made an special production of sticks with 4 pins pots. anyway, this pots with 4 pins are a bit special
Also looking at the same webpage at the design it seems like these 4 pin pots are, or at least were supposed to be more reliable than the 3 pin
In the boards where this component is present (or his replacement upgraded versions) every stick has two pots (to meassure rotations of X and Y axis of that stick), and every pot has 2 pins that are carrying the signals related with the rotation of the axis (pin 1 and 2 of the pot). This signals are voltages, but there is also another pin of the pot that has an additional voltage (pin 3). Inside the pot there are two separated voltages, this seems to create a magnetic field with the hall effect. Also this allowed them (in further versions of the pots) to separate the mobile parts inside the pot by an intermediary "seal" that prevents dust and degradation of the parts

So... this component is the initial version that gives support for this "special sticks with 4 pins pots"
PS3_controller_stick_sealed_pots.jpg
Seeing this diagram, I'm not 100% sure if this would work but if you were hypotethically able to install a 3 pin stick in the place of the 4 pin one. Then bridge the 4th pin and 3rd pin where the poles are together it may work. Seeing that they're all fed into one multi purpose controller which I'd imagine then spits out 4 signals for the 2 sticks instead of 8 which are then carried to one or two amplifiers which in total can only take 4 inputs such as 2 of the toshiba 5w54 which are featured on some PS3 boards.

Capture.PNG
Edit: Also looking at the potential interchangeability, I'm not sure if the technology is very different. The website I listed above says that the stick is in a range from 0 to 2.8V with the center point at 1.4V and while I can't see any graphs on Alps data sheet on some of their joysticks I don't see why that won't be the case with a 3 pin Joysticks as well but at the same time I could be completely wrong. I have a joystick from one of those Raspberry Pi kits that's definitely 3 pin, I'll take some measurements to see if it's any different.

Edit 2: Nope, it's the same. Feeding the 3 pin joystick 5.12V got me 2.56V at rest and about 0V when pushed all the way to the left and 5.12V when pushed all the way to the right so it's possible they're interchangeable assumming the output at both poles is similar to each other and not two very different values although I'm not sure why it would be useful since they get combined anyway before being amplified.
User avatar
MattKC
Site Admin
Posts: 323
Joined: Mon Aug 22, 2022 1:05 am
Contact:

Re: I bought 9 PS3 controllers for $40... how bad could they be?

Post by MattKC »

Hmm interesting... if the 3-pins can be repurposed that would definitely be helpful. Though I'd still be concerned about how they'd align on the board. My guess is the 4 pin holes wouldn't perfectly align with the 3-pins on both sides, which would cause the whole module to be off center (or might simply mean both sides can't be attached at the same time).
xaytana
Posts: 2
Joined: Tue Jan 03, 2023 3:38 pm

Re: I bought 9 PS3 controllers for $40... how bad could they be?

Post by xaytana »

I don't have one of the few models of controllers that have 3pin sticks, but in theory a 3pin pot should be compatible with with a 4pin board and vice versa. The issue is how do you connect them. You might get away with bending pins 90° and putting some electrical tape on the pads, and using jumpers to VCC, ground, and the test points that go to the microcontroller inputs. You might also be able to get away with cutting slots and passing the pins through these slots, then jumping things again. Similarly you could just cut that entire portion of the PCB and use a daughterboard, with the relevant connections and fixing any severed traces not relating to the sticks. But all of these are extremely jank solutions, a more proper one would be a reproduction of the proper board and using donor chips from the other board, but this is a sunken cost and a third party clone or alternative is a better investment.

As for the 4pin units themselves, they should basically be without fail, any failure would be upstream or downstream from the modules themselves. They're literally just inductive components, a PCB with some traces and an accompanying target, and in this case they're DC components (yes, I know, very odd for inductive circuits, which are typically AC) with a magnet for a target, probably for a more defined magnetic field and as a sort of amplification for weak inductance. The units themselves are very odd and reverse engineering them is definitely a unique challenge, but I think I'm getting closer to figuring them out, including production method. Unfortunately I cannot source Sony's patents on these, or really anything PS3 era for actual products, and usually that's a good first step. The devwiki is a good starting point, but it's also utterly incomplete and their theory on the sticks doesn't make sense, there's really no way to fit a wiper and carbon under the potting material and still have the wiper rotate.

I don't have the first versions of these modules, yet at least, so they're more unknown to me. They're clearly a PCB, more than likely produced on a sheet connected by the top, right, and left tabs. Notice the ground loop on the back and how the traces on the front go to a central point.

What I am more familiar with, and the modules I do own, are the second versions. They have fat traces on the front, with rear access, that's what all those little points are, with the ground 'loop' now on the front side and now only surrounding the other pins. I believe the manufacturing method for this is injection molding (or some kind of casting) around the copper, because the greenish teal plastic is not a PCB nor fiberglass, and the top, right, and left tabs are copper and connect to ground, so I think they produced these in sheets as well with very thin wires holding the other pins in place (which I'll get into later) with plastic casted around them, then snipped apart. But these, as far as I can tell, are not PCBs at all, just copper in set plastic.




I already went over basic operation, but I'll get into numbers now. Pulsed DC seems to be 2.8v at 200μs pulses, I'm not sure what valley to peak is or pulse width is, I don't have a scope on hand so I reached out for comment from the creator I grabbed the scope picture from, though I haven't gotten a reply yet; I could just measure for pulse width, but I'm not sure what the voltage scale is; Input 1 is the board with the module still attached, Input 2 is the board without the module.

As for my own measurements, I have a few. I know my meter has some amount of deviation, probably only -0.01v off.
First off, voltages. On the module itself before processing, 1.39v center, 1.36v and 1.42v extremes, and the two pins switch these values depending on direction, so for example at full left it'll read something like 1.36v/1.42v but at full right it'll read 1.42v/1.36v, and the next oddity is that going from VCC to pole shows these values positively but ground to pole shows these values negatively. After processing, at the MCU or test point at least, the values are 1.39v center with 0.23v and 2.5v extremes; you may have also noticed from the devwiki that there are second inputs on the MCU, LX/LY/RX/RY '_V ?' inputs, I believed these are some kind of reference voltage, though I was reading 1.2555v which appears nowhere in documented schematics, so I believe this may be tied to processing, the TI SN84001 chip, I don't have access to a VX2 or before or a VX5 to test the Toshiba 5W54 or the TI A6044A0; all three methods use an op-amp per stick, the pole inputs into it with a singular output for the MCU, as shown by the Toshiba chip, I'm just not sure where 1.255v is coming from as these chips should be running off of 2.8v.
Second off, resistance, which behaves differently if the module is in the controller or not. In the controller, the only resistance is 10kΩ. But removed from the module they show 10kΩ between VCC and ground, and 7.25kΩ between every other pin combination. Back to having the modules in the board, they do show diode values (with the positive lead on the first pin and the negative lead on the second pin) ground-VCC showed -0.718v, ground-P1 and ground-P2 showed +0.388v, and P1-VCC and P2-VCC showed +0.421v, all other combinations had no value. Testing for diodes with the module removed from the board had no values. Circling back to manufacturing method, and the idea that there may have been thin wires to bridge the copper traces during production, this might be where the strange disparities are showing themselves, I also believe the 10kΩ value between P1 and P2 while the module is installed in the board shows itself from the processing side of things, and not within the module itself.
Capacitance values didn't read as anything. And I don't have an LCR meter to test inductance, I also reached out for comment from the previous content creator to see if they had one while I was asking other questions, again no reply.

These are interesting little modules and provide a benefit over hall sensors, as inductive sensors don't have the issues hall sensors do. It's just a shame I can't find the Sony patents to reproduce these, as reengineering these otherwise will require either complete destruction or an x-ray imager. The potting compound really only exists, to my knowledge, to locate the magnet and as a bearing surface, which is genius as you have the four pins to constrain placement in one way and now a convex-concave coupling to further constrain the module to a location, plus the accompanying clips on the black plastic. From what I've found, these should be fairly easy to reproduce, as they're literally just some traces, this setup honestly looks like a pulsed DC TX antenna, a couple of RX antennas, and a grounding antenna for the EM field around the other antennas, all manipulated with a segmented (along the vertical axis) magnet. The magnet will be the hardest part of reproduction, as it's a fairly custom piece. I don't think there'll be much difference between a PCB and the strange plastic molding method Sony used for the second version. Though I would need to produce some and test them.

While I was patent searching, I did come across a few other things. Someone had patented a modern inductive position sensor to work with a stickbox in 2019, an idea I've been thinking about it, no idea if this ever practically existed. And in '96 someone had patented a multi-sensor inductive stick, for coordinate-like values, another idea I've thought about, though inductive coil designs have come a long way since then, and there's similar coordinate-like arrangements that're more recently novel than a grid of individual sensors. Inductive sticks are a better way forward than hall sensors, given their issues. Another option is optoelectronics but those can be more costly and mostly unexplored within this use case. I also want to experiment with capacitive sensors, like a very near range proximity sensor and find a component to throw into the stick's centering mechanism, and see if this provides a better solution.
User avatar
andreizabest
Posts: 55
Joined: Sat Dec 10, 2022 4:49 am

Re: I bought 9 PS3 controllers for $40... how bad could they be?

Post by andreizabest »

xaytana wrote: Sun Jan 08, 2023 5:32 am It's just a shame I can't find the Sony patents to reproduce these, as reengineering these otherwise will require either complete destruction or an x-ray imager.
It may not be a sony patent it could be a patent by alps or another company, or just a patent that expired.
I found this patent while looking around which seems to be at least similar to what you might find in the PS3. It's a Noncontact potentiometer and has 2 output terminals similar to the one in the sony controller. And the abstract says this.
A potentiometer includes a magnetoresistive element made of ferromagnetic material. A device applies a magnetic field to the magnetoresistive element. An absolute value of the magnetic field applied to the magnetoresistive element is equal to or greater than a saturation magnetic field with respect to the ferromagnetic magnetoresistive element. The magnetic field applying device is movable relative to the magnetoresistive element.
Another reason why I think this might be the patent for it is that it expired 8 or so months after the PS3 released and looking at the PS3dev wiki the first six axis controllers were 3 pin for the most part until the third "V2" revision which got the 4 pin pot.
xaytana
Posts: 2
Joined: Tue Jan 03, 2023 3:38 pm

Re: I bought 9 PS3 controllers for $40... how bad could they be?

Post by xaytana »

I hadn't considered the idea of magnetoresistance, considering how they're much more complex than just a copper PCB, as they typically require ferrite alloys. But this seems to be what they are, and ALPS ALPINE seems to have an unsearchable amount of patents (when I glanced on Google patents, the number was in the thousands with only searching 'ALPS ALPINE' as the assignee under the broad term of magnetoresistance and magnetoresistor), but the patent would seemingly belong to ALPS and not Sony; though this doesn't clear the fact that I can't find any relevant product patents for the PS3 era, which is still bizarre, if I could've at least found the produced SixAxis or DualShock 3 patents this revelation could've come sooner. Though the pulsed DC VCC line is still odd, as is the design of the second variant of module, where the traces otherwise don't lead to a chip, more reason to throw it under an x-ray I guess; at surface value these things really do come off as some kind of inductive component. I also find it a bit odd that out of all the various MR patents ALPS has, seemingly only two stickbox modules ever existed and only under one customer for a relatively narrow span of time; I wonder how hard it would be for a company to get ALPS to produce these again.

But I got around to probing a bare module. The things aren't easy to test, and numbers aren't exact considering the plane of the magnet has to be parallel to the plane of the sensor, any skew throws numbers off. They also seem somewhat non-linear when nearing 0°/180° from 90° being center, as testing horizontal is easier that attempting to test the ±23° arc or whatever it is these stickboxes were originally rated for. Also any disparity in my dmm produces slightly off numbers.
Combinations of pins for G-P1, G-P2, V-P1, and V-P2: 90° (typical center) was 7.64Ω, 0° (X+, Y+) was 7.30kΩ, 180° (X-, Y-) was 7.84kΩ. Inverting the magnet in rotation (270°) produced the same 7.64kΩ reading. Edit: V-P1 flips the values for 0° and 180°, V-P2 may do the same but I'll have to check this again at some point. These things would be a hell of a lot easier to test if I either had an eval PCB for them (the pins don't match any pitch of breadboard or strip PCB I have on hand) or some very thin hook probes for my DMM. It also seems that V-P1 (and again possibly V-P2 as well) had a bare (no magnet) resistance of 4.83kΩ while G-P1 and G-P2 had a reading of 7.23kΩ bare. I did try my best to measure ±45° from center by aligning the magnet with the stickbox alignment pins, and got about ±0.18kΩ from center, so these do become much more linear in output towards the narrower range of what the stickbox physically allows; I'd need a more scientific setup to test the ±23° (or whatever the spec for these stickboxes were), but I'd hazard a guess within a ±0.10kΩ range from center.
G-V and P1-P2 both resulted in 10kΩ, rotation didn't seem to matter.
Also very susceptible to foreign fields. I don't know the strength of my magnet, I just have one of those magnetic bolt grabbers mechanics use, but I started noticing drift within 7.25mm of the back of the module with the stick getting maxed out at around 3mm of distance; the magnet isn't super strong, just strong enough to grab a bolt. Shielding might help but that'd also mess with the intended magnet's fields and how the magnetoresistors act.
Finding an MR sensor of these specs isn't going to be a fun time, assuming a readily available product exists.
Edit: I'm not even sure how these sensors are properly specced; are they specced bare as most specifications include a minimum magnet strength, but we all know that magnet strength and distance will adjust output.

As for the patent you found, that's more or less a throwaway paper that was already nearing its 20 year lifespan by time the controllers were coming out. Maybe novel in implemented design, but there's thousands of other magnetoresistior patents. Good lead, though. Replicating Sony's design would obviously need a custom fab, there might be some off-the-shelf sensor that could be adaptable to a DIY module, I just don't know where to find affordable modules of the correct spec as Sony's 10kΩ-7.64kΩ sensor seems to be extremely custom. I might just continue to chase the idea of an inductive stick, considering how much simpler those are to produce in comparison to an MR sensor, or more novel designs that're still approachable. A better design for the MR sensors would be to have the sensor and op-amp on the same module, and any other relevant circuitry to adapt controller VCC to what the module wants to run at and modify the module output (if a simple amp doesn't suffice, an ADC-DAC setup could work) to what the controller expects; though Sony's design seemingly hit the correct numbers to make the MR and pot modules electrically interchangeable. I do have a couple of industry contacts, I could see if they want to dive down the rabbit hole of making a modern solution, but as an individual this project is too far out of reach; making a drop-in three-pin solution as a replacement to pots that fully adapts controller to MR input and MR output to controller seems fairly novel and shouldn't infringe on any standing patents.
Edit: ALPS does seem to have a line of off-the-shelf magnetoresistors, and I doubt their specs would line up with what's required for a stickbox, I also wouldn't be surprised if attempting to implement these would run into clearance issues with the stick cap itself. They do have a tiny module that's only 1.8×1.05×0.65mm that may not cause physical interference depending on how the magnet is supposed to be set up in relation to the sensor, as this is one of the smallest I've come by so far; I've seen some datasheets for other sensors that seemingly imply that magnet placement is not equal between all products). This might be a decent first step towards a solution, I still don't think any off the shelf product is going to have the spec for a drop-in replacement though. A custom fab is still probably the best bet all things considered, especially when considering the idea of putting the dual-input op-amp onto the package itself for a 3-pin stick. Given the price of any stickbox on the market, especially when looking at something like the Steam Deck replacement boards (one of the few sources of getting guaranteed branded parts), I think a proper company with a proper manufacturing contract could probably generate a solution at a competitive price and still be profitable. I still need to dive down the rabbit hole of durability, see if there's any quiescent issues that eventually pop up, such as the hall sensor's issue with age-related voltage drift.

I wish there was an ALPS warehouse that just has a ton of these laying around. Or a treasure trove of them buried out in a desert somewhere. Adapting what is more readily available to replicate the design seems like a pain in the ass.
Post Reply