News Some Early Steam Deck Customers Report Joystick Drift

InvalidError

Titan
Moderator
The JoyCon drift is caused by the analog controls using a simple variable resistor to sense joystick position which is extremely susceptible to any variation in resistor wiper contact characteristic, carbon/polymer track wear, contamination, etc. If the Steam Deck used the same thing, then premature failure is inevitable. Older analog controls used potentiometers where the wiper reads voltage along the resistor track, which makes it far more resilient to most crap at the expense of requiring constant power.
 

peachpuff

Estimable
Apr 6, 2021
571
591
2,760
The JoyCon drift is caused by the analog controls using a simple variable resistor to sense joystick position which is extremely susceptible to any variation in resistor wiper contact characteristic, carbon/polymer track wear, contamination, etc. If the Steam Deck used the same thing, then premature failure is inevitable. Older analog controls used potentiometers where the wiper reads voltage along the resistor track, which makes it far more resilient to most crap at the expense of requiring constant power.
So a financial tradeoff?
 

Sleepy_Hollowed

Distinguished
Jan 1, 2017
501
195
19,070
This is sadly almost expected.

The only controller I used in the past that did not suffer from this was the gamecube, but that console was built like a tank on the controller front.

On more recent consoles, the PS4 controller MAYBE, but then again I abuse it much less (have much less time to play) than older consoles.
 
Apr 1, 2020
1,391
1,046
7,060
Something I don't understand is why controllers aren't designed to have easily replaceable joysticks? Sure it may add a little into the cost but to be able to quickly unlock, remove, and plug in a new replacement joystick module would solve a lot of headaches...
 
This is sadly almost expected.

The only controller I used in the past that did not suffer from this was the gamecube, but that console was built like a tank on the controller front.

On more recent consoles, the PS4 controller MAYBE, but then again I abuse it much less (have much less time to play) than older consoles.
I would not expect it from new which is the case here. I’ve had many XBox controllers over the years and none had it from new, a few did develop it after much use.
 

InvalidError

Titan
Moderator
Something I don't understand is why controllers aren't designed to have easily replaceable joysticks? Sure it may add a little into the cost but to be able to quickly unlock, remove, and plug in a new replacement joystick module would solve a lot of headaches...
It is engineered obsolescence: these things are designed as consumables with a clearly wear-defined lifespan: the wiper contact applies a predetermined pressure to the resistive contact pads with a predetermined smoothness, which causes a very repeatable amount of wear per amount of wiper travel, a repeatable amount of dust and a fairly repeatable travel time to failure.

In the two-terminal analog circuit, that dust ends up in series with whatever the conductive film resistance at the wiper position should be and becomes part of the readout. In a three terminals potentiometer arrangement, the voltage at the wiper is the same regardless of the amount of crap under the wiper as long as the op-amp reading the voltage has much higher impedance than the potentiometer and wiper dust combined.

While old-school analog sticks may have required some in-game calibration to reset the dead-zone location and range limits, they hardly ever required replacement.
 
I do not believe there is a meaningful cost difference between a variable resistor and a potentiometer, especially when printed on a flex PCB that does X+Y+button. All you need is an extra ground or power trace.
A variable resistor and a potentiometer, is just a different name for the same thing.
The JoyCon drift is caused by the analog controls using a simple variable resistor to sense joystick position which is extremely susceptible to any variation in resistor wiper contact characteristic, carbon/polymer track wear, contamination, etc. If the Steam Deck used the same thing, then premature failure is inevitable. Older analog controls used potentiometers where the wiper reads voltage along the resistor track, which makes it far more resilient to most crap at the expense of requiring constant power.
Most potentiometers nowadays use a carbon/polymer track although you can get wire wound potentiometers that cost more.
 

InvalidError

Titan
Moderator
A variable resistor and a potentiometer, is just a different name for the same thing.
While they can be physically identical, they are functionally different: a variable resistor is exactly what the name says with any garbage getting between the wiper and film heavily affecting total resistance while a potentiometer is a voltage divider where both ends of the resistive film are tied to different voltages and the wiper outputs something somewhere in-between based on position along the film and as long as the sense circuit has sufficiently high impedance, garbage between the wiper and film has little to no effect on reading.
 
Why oh why don't they use encoders? A micro slit encoder is literally about 1x1cm. Much more reliable and accurate than analog circuits.

Steam index controllers had the same issue. Still no fix. Eventually you'll have to replace them.

Also there's an issue with the ear speakers and display port cable. None of which you can easily fix or get parts for.

Valve has a problem with hardware quality.
 
Last edited:

InvalidError

Titan
Moderator
Why oh why don't they use encoders? A micro slit encoder is literally about 1x1cm.
Too large. I repaired my nephew's JoyCons a few months ago and took one of the stick module apart to look at how they worked. Those things have only ~2mm of travel after the angular-to-linear conversion mechanism that leaves practically no z-height in the ~3mm the mechanism is crammed into to put anything on the flex PCB. Putting optical encoders in there would require a much thicker package to accommodate optics and that package would need to be dust-proof to prevent dust from ruining the encoders.
 
Too large. I repaired my nephew's JoyCons a few months ago and took one of the stick module apart to look at how they worked. Those things have only ~2mm of travel after the angular-to-linear conversion mechanism that leaves practically no z-height in the ~3mm the mechanism is crammed into to put anything on the flex PCB. Putting optical encoders in there would require a much thicker package to accommodate optics and that package would need to be dust-proof to prevent dust from ruining the encoders.
The z height is only about 4 mm. And it is dust proof. the 2mm travel is no issue with encoders this small. They are a fairly simple mechanism with 3 LEDS, 3 IR sensors, 1 moving part (the wheel attached to linkage) in the package.
 

InvalidError

Titan
Moderator
The z height is only about 4 mm. And it is dust proof. the 2mm travel is no issue with encoders this small. They are a fairly simple mechanism with 3 LEDS, 3 IR sensors, 1 moving part (the wheel attached to linkage) in the package.
You aren't going to fit much of a wheel in a package with ~3mm of internal clearance (the package may be 5mm thick but you have over 1mm of top plastic, a 0.5mm flex PCB and 0.5mm steel back cover in there) and additional linkages would add backlash which in turn makes the dead zone larger while also adding more mechanical failure points and manufacturing complexity.

If you wanted to build an optical encoder in there, you would mould the slits into the sliding racks (no extra part or backlash) and decode that, though those slits would need to be crazy small to provide any sort of resolution on ~2mm of travel, you get maybe 10 ticks from end-stop to end-stop with mouse wheel type slot pitch. Position encoders used in things like mouse wheels have only one LED and two detectors, using slits in the optical housing to shape the beam(s) and receiver window whichever way the intended application requires.
 
You aren't going to fit much of a wheel in a package with ~3mm of internal clearance (the package may be 5mm thick but you have over 1mm of top plastic, a 0.5mm flex PCB and 0.5mm steel back cover in there) and additional linkages would add backlash which in turn makes the dead zone larger while also adding more mechanical failure points and manufacturing complexity.

If you wanted to build an optical encoder in there, you would mould the slits into the sliding racks (no extra part or backlash) and decode that, though those slits would need to be crazy small to provide any sort of resolution on ~2mm of travel, you get maybe 10 ticks from end-stop to end-stop with mouse wheel type slot pitch. Position encoders used in things like mouse wheels have only one LED and two detectors, using slits in the optical housing to shape the beam(s) and receiver window whichever way the intended application requires.

LEDs and IR's are about .65 mm each (edit: Looked it up). The wheel is photo etched and < 1mm And the armature attached to the sliding rack. I've seen it done in specialty applications. Dead zone comes down to "slop" tolerance on the rack and armature.
 

InvalidError

Titan
Moderator
LEDs and IR's are about .65 mm each (edit: Looked it up). The wheel is photo etched and < 1mm And the armature attached to the sliding rack. I've seen it done in specialty applications. Dead zone comes down to "slop" tolerance on the rack and armature.
I doubt Sony, Microsoft, Nintendo, etc. are interested in micro-machining parts for their$50 game pads.