No bananas were harmed in the making of this Hall effect drift-proof joystick replacement. OK, not really — two bananas were turned to mush. But it’s OK, they’re just bananas, after all.
Why bananas, you ask? Because [Marius Heier] uses them to demonstrate what we all intuitively know — that rubbing something over and over again tends to wear it away — but engineers seem to have forgotten. Wear such as this, with resistance material rather than fruits, is what causes the dreaded drift, a problem that the world collectively spends $20 billion a year dealing with, according to [Marius].
While numbers like that seem to be firmly in class-action lawsuit territory, sometimes it’s best to take matters into your own hands and not wait for the courts. The fix [Marius] shows here is to yank the potentiometers off a PS4 joystick and replace them with contactless Hall effect sensors. The end of the shaft for each axis gets a diametral neodymium magnet attached to it, while a 3D printed bracket holds a tiny custom PCB in close proximity. The PCB has an AS5600 Hall sensor, which translates the shaft angle to an analog voltage output. After programming the chip over its I2C bus, the sensor outputs a voltage proportional to the angle of each shaft, just like the original pots, but without all the wear and tear.
While [Marius] is selling these as drop-in replacements for PS4 controllers, he plans to release all the design files so you can build one yourself. He also has his sights set on replacements for PS5 and Xbox controllers, so watch for those. This isn’t his first foray into joystick hacking, having shared his 3D Hall effect and haptic feedback joysticks with us previously.
There’s an even simpler solution. Don’t buy consoles, don’t play video games. Life is much better when you don’t waste it on corporate entertainment.
But non-corporate entertainment is not a waste of time?
There is plenty of non-corporate video games… even FOSS ones.
… but using your time complaining to others about what they choose to do with their own time and money isn’t a waste of time?
its my life and i will waste it in the most efficient way possible.
You do you, champ, but stop judging people for their hobbies. I work as a helping professional, read, don’t watch television shows, and play video games in moderation, but if a developer pulls some shady sh!t, I’m quick to stop supporting them.
I have no consoles (well a Wii in a box somewhere and a WiiU nobody ever turns on), yet I have many controllers. Xbox 360 and USB receiver. And PS4 works natively on SteamDeck OS (on deck or HoloISO).
Also controllers are commonly used for flying drones or controlling other equipment.
That said I am more concerned about physical wear of the bearing and self centering surfaces. Which the hall+effect purveyors conviently neglect to mention. My Guliket stick has a problem getting hung up just off center and at the very edge. (The center isn’t terrible, its still 1/5 the deadzone of original stick, but irritating, ofc they were fairly cheap)
i mean they use thumbsticks, they aren’t really at the pinnacle of joystick technology. il take a good set of cnc machined gimbals over anyone’s fancy sensor tech any day. having calibration points built into the hardware is also nice. a clean signal path if using analog, or at least keep the adc as close to the sensor as possible, perhaps with a couple of tweakable voltage references.
i got a number of military surplus potentiometers that are still the best sensors ive ever used, they were made in the usa at a time when analog was king. years of military service never impinged on their linearity or precision by any measurable quantity. i tested them against a home brew hall sensor and they had measurably superior linearity. of course such parts are hard to come by and not something you can base a production line on.
As long as you have consistency you can get linearity with software curves on a hall.
However, are the military grade pots able to be utilized for modern console joypads, or something roughly akin to arcade joysticks only? … or simply used to trim lighting levels in your home office or whatever?
Tell me about the non-corporate device you sent this reply with. I’m on pins and needles.
Coming here to a tech site to proselytize about technology seems both egotistical and wasteful of time.
Or stick to non-analog consoles? Like SNES and Sega Genesis? Analog stick started getting lots of use with PSX, N64, and XBox. (analog have been in use since 70s Pong but other than an Atari 5200 it wasn’t the main feature)
Not our fault you’re too poor to experience them properly. Try applying outside of McDonald’s mate.
🤣🤣🤣 maybe you need to play some vr and get teabagged
I’ll do whatever I want, chief. I don’t need your permission to live my life as I see fit.
This is neat, but I’m still somewhat confused by the problem. I’ve never had a thumb stick wear out. Are people abusing their gamepads, or gaming for far too long?
I’ve had a few used Xbox controllers zero drift on the sticks which makes most any game or even navigation in the gui unusable (it throws dpad tracking off too).
yes
N=1. To counter your n=1 with mine: I got a Switch a couple years ago, same time and batch as some friends of mine. They’re more active gamers than me, yet mine started drifting only months in. Did a DIY fix and it’s been fine since. Not all parts are created equal.
Not just the quality control in the parts, some of it is going to be the users – hands covered in dusty crap you work in that makes the contact perform worse but if cleaned will work alright for a while longer etc… I’ve really not had trouble with wear on any of the consoles I own at a pace that seems unreasonable, which does include a Switch (though ergonomics being awful it never gets huge normal use).
I also don’t really thing a moving part wearing out is at all a problem the whole “class action lawsuit” from the article seems really daft overall – or are we going to class action against the automotive company in general because some big number are spent on new brake/clutch parts every year… Some brands do make surprisingly low quality parts, so they should perhaps have a class actions. But moving parts always wear – even the ‘hall effect’ doesn’t stop the pivot and centring springs from wearing out creating the same result as the dirty/worn potentiometer traces/brushes, and equally no point engineering it to last 20 years of heavy use when its going to be getting dusty on a shelf in 2 years. That is just unnecessary expense and probably makes the device actively worse to use – more wear proof tends to end up meaning heavier and/or with active high cadence maintenance requirements.
So your N=1 counters their N=1 which counters someone else’s N=1?
I guess I’ll counter-counter-counter with my N=1. My switch’s left joycon started drifting somewhere around a year after I bought it. Relatively modest amounts of use, a generally gentle usage pattern, and I never play with grubby hands.
It’s a problem that many people have experienced far before what one would expect the lifespan of a controller to be. Luckily 3rd-party joycons are cheap.
Didn’t say the joycons were good – IMO they are garbage even when they are not drifting, and I have had some drifting – though never actually needed a replacement just a little blast of contact cleaner did the job (but replacements sticks are cheap so I did that pro-actively and those replacements have been bulletproof so far).
I was just saying there is more to a joystick wearing out than bad production quality with one example, but there are others – for instance tossing the controller in a bag/box with no stick protection can be great for wearing them out or outright ripping them off without ever using them for their intended purpose. A human interface device generally doesn’t have an easy life and a gamepad is perhaps the most abused by its nature and how impossible they are transport/store well protected.
Probably only an issue if you play competitive type games. If not, you can just turn up the deadzone up a little bit and it still “works”. I’m sure low deadzone feels much better though.
I never used to worry about my mouse and monitor on PC until I started playing competitive counterstrike. Now I do. You can save a couple of milliseconds here and there if all your gear has the lowest possible latency. If you and your opponent both have reaction times in the 300mS range, then shaving 10mS can give you a very noticeable edge.
I’m sure the same sorts of issues apply to console gaming. Maybe even more so, because almost everyone else is playing with the stock controller.
“are people abusing their gamepads” lol
the non-parent has logged in :)
i think the problem is that consoles don’t have the calibration facilities that pcs tend to have. they want the sticks to wear out so you have to buy a new controller. but most of these problems can be calibrated out. you might have situation where the carbon strip wears down, and i got some milspec parts where that doesn’t happen (seems a problem with the unsealed parts you find in consumer electronics). this happens on a game pad you use with a pc, you can run the calibration on it and you can at least re-center. you usually lose some linearity when this happens though. if you can accurately characterize the pot you could build a linearization table, but that usually requires firmware level hacking.
Nintendo Switch joysticks are well-known for their exceptionally high failure-rates. It’s to the point where Nintendo offers a free repair service for them – and some people have sent their controllers back multiple times.
I’m an occasional gamer (usually 1-2 hours a week), and yet my Switch’s “joycons” failed within 6 months.
Rather than sending them back to Nintendo for repair & having to deal with this again next year, I just replaced the sensors with Hall-effect ones, and hopefully that solves the problem going forward.
(Honestly, I’m surprised that somebody hasn’t sued Nintendo over this yet … or maybe they have…)
There are multiple active lawsuits. The issue is that – thus far – US courts have sided with Nintendo and forced arbitration due to the EULA displayed when setting up a Switch.
This is a cool in place replacement. I’m fond of these kinds of magnetic sensors, and use the 5601 in a USB arcade spinner.
That said, battery life on wireless gamepads will take a noticeable, though not enormous, hit. The AS5600 in constant running mode uses 6.4ma, while a 10k pot at 3.3v uses 0.3ma. And we have four of the AS5600s in a typical game pad. A gamepad with 13 hours battery life and a 1050mah battery draws 81ma, and 24ma is a noticeable fraction.
I also actually haven’t had joypads wear out in the way this fixes. All my failures were internal springs breaking, and this won’t help.
What I wondered when seeing this is if there might be a different configuration possible like a singe 3 axis hall effect sensor per stick but I suspect some complex math, reliability, and form factor constraints are big parts of the reason for this particular configuration using 2 sensors.
That would require changing the rest of the controller electronics. Using one Hall effect sensor in analog output mode makes for a dropin replacement for the pot.
I started out wondering why you used such a complicated hall effect sensor when most commercial applications use simple 3 pin chips, that basicly act like a standard potentiometer. Then it hit me – calibration! You solved the biggest issue other people have failed at by using a more complicated sensor that you can use to correct for whatever calibration the controller has
For those not familiar, the reason that it isn’t as easy as just replacing the sticks is that every thumbstick gimbal and every potentiometer is a little bit different. Brand New sticks will have some variance, so when the controller is manufactured the company calibrates it out. If the stick originally reads -4%X +7%Y, they can offset the values the controller reads to be +4% -7% and it will read perfect center. If you replace the pots or full thumbstick module, especially if you get one that is variant in the opposite direction, you can end up with stick drift that’s due to the miscalibration rather than physical wear. I used to correct this by taking the wiper out and very precisely moving the middle bump back or forth to adjust it physically. Works but takes forever. At least Nintendo realized the joycon sticks were garbage and let us actually use the calibration menu ourselves
Having a single chip that is programmable is exciting. You would need to sell or design a programming tool that also calibrates it. You could have something like a cap with pogo pins that has the same screw holes and screws down over the sensor, then you push a button and move the stick all the way forward, all the way back, it takes what it reads and applies the calibration. The main issue here is the movement range of the stick in the controller is not the same as the total travel distance of the stick. Ex the physical stick might be able to move 45 degrees in any direction, but inside the controller shell of a PS4 moves 30 degrees in any direction, and inside the shell of an Xbox one controller moves 38 degrees, etc. If you calibrate to maximum extents = 3.3v/0v, and the controller only expects it to move within 80% of that range, once assembled it would only go to 2.7v/0.6v. this is a an extreme example but makes the point – you really need a way to calibrate them with the controller top shell and stick caps on.
One possible solution is add a vertical ffc/ribbon cable connector (like the one they use for the cable on the ps4 controllers going to the USB board, or the dual sense microphone connectors) and mount it close enough to the PCB side that colliding with the thumbstick cap isn’t an issue. You can connect the programmer by an fpc flex cable, and then assemble the top half of the controller so you can calibrate it with the correct stick caps and shell
Another thought is getting a sensor that has a built in min max calibration routine, and having it enter like a calibration mode on startup for the first 1 second, if the sticks aren’t moved then use the current calibration, if the sticks start moving then record the min and max values and after 3 seconds use that as the total range. I remember some older controllers used this and it seemed to work fine. The issue is finding a sensor that has this function built in – you might have to rely on having 2 chips, a simpler more basic hall effect sensor and a small microcontroller that applies the min max and programming logic (this is what guilikit does with the stream deck replacement stick modules)
I was also wondering, what about a sensor that has 2 pins for min and max setting, and like 2 tiny and pots on the back of the sensor to adjust them manually? You would need a sensor that uses the resistances of the pots to determined your upper and lower limit. End user could manually tweak them once which would kind of suck, but a dab of hot glue or silicone to keep them from ever moving again and it’s done, and the sensor wouldn’t need any external programmer to set the limits. No idea how you would calibrate this while it’s in the shell though
i like single chip solutions for thumbsticks. a vector hall can do that with the right kind of calibration routine. to make that work you want a part with an onboard dsp that can be factory calibrated for linearity. you could offload the math to the mcu but that will add latency.
in fact ms sidewinder line used an optical system, which was just a mouse sensor but with the dsp programmed for point tracking (a board with 2 leds on it was the target, just inside the ball gimbal), you could get 3 axis tracking on a single sensor. doing something similar with hall sensors seems the optimal way to go (or something on a modern mouse sensor perhaps).
I appreciate his research, but I find it hard to believe he can’t use pre-existing Hall joysticks and sensors. Also if he is planning to produce these he will probably find he is better off with a cheap micro-controller and a couple sensors /under/ the stick on a thin “flexible” Kapton PCB.
He also neglected to mention most cars are controlled by potentiometers in the gas pedal and throttle body. And the only ones I’ve heard of with high failure rate are a 90s-00s Volvo with Italian made throttle body.
What I’m getting at is the best fix might be getting some new high quality swiss made potentiometers that fit the original space. In the time frame (and considering the quality of the joysticks axis bearings) the result might be nearly indistinguishable. 🤷♂️
problem is that most of the better hall sensors are digital, you can get purely analog hall sensors you can run to any ordinary adc pin on your mcu. but the interesting ones are the ones with a digital interface so your analog signal path stays on chip. i did some experiments with ss49e, which you can replace a pot with, however you have to give up a lot of range compared to a 60 degree pot. this sensor has a 180 degree range with only about 160 linear enough to be usable, so you leave over half your adc range on the table.
this is where i normally get to go on about how much of a gimmick hall sensors are and how high quality potentiometers and adcs exist that are effectively immortal, and there are other options that are cheaper (such as optical, like the sensors old sidewinder sticks used). and that hall sensors come in a wide veriety of options from freaking expensive to so cheap a cheap pot looks like an upgrade.
but instead id like to say there is a place for it in the lowly thumb stick. something with high wear and low precision requirements is perfect for hall sensor applications. they do not require high cost components which is a huge selling point. though i wish they would start using the vector hall sensor. preferably one with an on board dsp for linearization and a 100% digital interface. a single chip solution would really get the costs way down.
Now that you say this I’m wondering if high DPI mouse sensors are getting cheap enough to build into thumbsticks.
The problem is definitely shifting from the quality of the sensor to the quality of the mechanical centering. See N64/Dreamcast for very real examples of that.
Once you cure that you are left with the programming of the game engine for smooth frametimes, the acceleration and damping curves applied to the inputs. Etc etc etc. Then you know why some people don’t even bother playing. I don’t much, and when I do it is not reaction based games (EG Myst, Subnautica, SnowRunner). Or rarely a good racing game like Trackmania or Rallisport 2 (aka the game that ruined NFS crap molasses physics for me forever); sadly Rallisport was made an Xbox Exclusive. I did try Rallisport 2 on the X360 “emulator” for OG Xbox, which only had minor graphical issues (it does require a modded X360, which isnt much of a hurdle these days, sometimes only needing a NAND flash tool and a jumper wire)
So, this is just using the same technology as PS3 controllers, at least the majority of them.
joystick and triggers from Sega Dreamcast controller uses hall effect sensors too
Gulikit has been making these for a while for the Switch and last last year or so for the SteamDeck. While the video is cool, joysticks using Hall effect aren’t really a new thing. Pretty sure the first PS3 controller had them back in 2006/2007.
But what is new is retrofitting a Hall effect sensor onto a pot based stick, without having to remove the old stick from the board.
I’ll just leave this here as it is already being done
https://www.amazon.com/AKNES-Joystick-Replacement-Controller-Thumbstick/dp/B0BPS1ZW1H/ref=sr_1_9?keywords=hall%2Beffect%2Bgimbal&qid=1684951643&sprefix=hall%2Baffect%2Bgimba%2Caps%2C308&sr=8-9&th=1
Trim pots should be on any resistance based controller. I made a pitch bender with a stiff leaf spring on a synth that has no dead band which uses a coil and slug. Vibrato as well as pitch bending is all enabled in a one dimensional control. There is nothing I’ve seen like it on keyboards. Dead band on a controller? On a plane or bicycle? They all need to be trimmed to center.
Now for volume control. I have used various resistance devices to express on a steel guitar with my fingertip on the slide itself. Old Baldwin variable pressure spring “contacts” for their pedals seem the best I’ve found, made to wear. These are followed by 70’s Yamaha 60 degree stop pots sort of a lever pot. Sadly they wear out fast. What’s to do today that’s simple? Optical with banned photo resistors or go full active volume control with hall sensor or optical interface?
That Lectronz looked interesting at first glance, but then I noticed that it’s like a US site: totally lacking in information that they should supply, like what freaking payment options they support – or what shipping methods for that matter.
Like with many US site you would first have to fill in all your information and details before you find out.. it’s idiotic.
They mention ‘EU cards’ and ‘non-EU- card’ fees, so I’m guessing it’s all CC only.
They have a link to their payment service, called ‘stripe’, but even there finding relevant info is a bit of a pain.
Even if you think CC is the only thing that exists, then you’d still need to tell people WHAT cards are supported don’t you? Or are we just to assume everybody not only has a credit card but in fact has 20 of them. I guess it makes sense to americans… But if you make a site aimed at the EU then perhaps realize that the EU isn’t inhabited by americans huh, and that there are many payment systems in ‘modern times’ in the EU that are not linked to any CC.
Thanks for the space to rant HaD :)