Complete Controller Health Report

Most controller problems are diagnosed one symptom at a time, which is a slow way to find out that three small things went wrong together rather than one big thing going wrong alone. This page does the opposite. It runs seven short checks in a single sitting, scores each one against what a healthy pad should manage, and hands you a single number with the working shown underneath. Two minutes of your time gets you a report you can act on, argue with a manufacturer about, attach to a resale listing, or file away and compare against six months from now.

Run The Health Check

Results

Area by area

What to do next

Report

How To Run It Properly

1. Plug the controller in with a cable if you own one. A wireless link adds its own faults on top of the controller's, and separating those two later is much harder than avoiding the problem now.
2. Close Steam Input, DS4Windows, reWASD and any vendor software first. Every one of those sits between the pad and this page, and a profile you set up months ago can change the numbers without ever telling you.
3. Press any button on the controller so the browser starts reporting it. Nothing appears until you do.
4. Press Start the report, then put the controller flat on a desk and take your hands off it completely. The first stage measures what your sticks do when nobody is touching them, and holding the pad adds your own tremor to the reading.
5. When the travel stage begins, push both sticks into all four edges in turn. Push until the stick physically stops against the housing rather than until it feels far enough. That gap accounts for more bad readings on this page than any hardware fault does.
6. For the release stage, push a stick fully to one side and let go cleanly. Do not guide it back. The tool is measuring how far past centre the stick throws itself on the way home, and steering it defeats the measurement.
7. Work through the buttons stage until every input has registered at least once, then roll your thumb around the D-pad, then pull both triggers to the bottom and release them.
8. Save the report. A score on its own is useful and a score next to one from six months ago is evidence.

If a stage is not relevant to your controller, skip it rather than forcing it. A skipped stage is left out of the total and clearly marked in the report, which is far more honest than a zero for something you never tested.

What Each Of The Seven Stages Measures

Hands off

The most informative six seconds on the page, because it measures two completely different things at once. First, where your sticks sit when nothing is touching them, which is drift. Second, how much that reading wanders while it sits there, which is noise. Those two get confused constantly and they have different causes and different fixes. If either one looks bad here, the stick drift test is the page that pulls the resting position apart properly.

This stage also quietly measures your connection, by watching how evenly readings arrive. A pad that reports in a steady rhythm looks different from one that stutters, and a stutter here explains odd results in every stage that follows.

Full travel

How far each stick actually reaches in each of the four directions. Games assume you deliver a full 1.00 when you push all the way, and plenty of sticks do not. A stick that tops out at 0.90 cannot reach full steering lock or maximum turn speed, and nothing on screen ever explains why. This stage also compares the two ends of each axis, because a stick that reaches full travel one way and falls short the other is a specific fault rather than general ageing.

Release

What the stick does on the way back to centre. A healthy stick returns and stops. A tired one overshoots past centre before settling, which in a game reads as your aim flicking the wrong way for an instant every time you let go. This is a spring and damping problem rather than a sensor problem, which is why it gets its own stage rather than being folded into the travel score.

Buttons

Every button the controller reports, checked for three things. Whether it registers at all, whether it releases cleanly, and whether it drops out while being held. That last one is the interesting one, because a contact that fails only under sustained pressure passes every quick check anybody ever runs. If something looks wrong here and you want to dig into individual inputs, the button test gives each one its own attention.

D-pad

All eight states rather than just the four directions, because a D-pad is one plastic cross on a pivot and the diagonals are built from two contacts closing together. Diagonals fail long before cardinals do, which makes them the earliest warning a D-pad gives you that it is wearing out.

Triggers

Where each trigger rests, where it bottoms out, and how much of its travel is actually usable. A trigger that registers slightly before you touch it, or that stops reporting before it physically stops moving, is losing you range at one end or the other. The trigger pressure test goes further into how smooth that travel is between the two ends.

Rumble

The only stage the tool cannot score for you, because there is no way for a web page to detect whether a motor actually spun. It fires both motors and asks you what happened. Your answer goes into the report as a note rather than into the score, which is the honest way to handle a subjective check. If nothing happens and you want to test the two motors separately, the vibration test drives them independently.

How The Score Is Worked Out

No hidden formula. Each area scores from zero to a hundred on its own, then the areas are combined using the weights below. Skipped stages are removed from the calculation entirely and the remaining weights are scaled up to fill the gap, which is why the report always tells you what percentage of the checks your score is actually based on.

The weights are opinionated on purpose, and it is worth saying why rather than pretending they fell out of a formula. Sticks carry the top three weights because stick faults are the ones that ruin games silently, cannot be worked around, and get worse over time. Buttons rank high because a dead button is unarguable. D-pads and triggers carry less weight not because they matter less to you, but because they fail less often and their faults are more obvious when they do, which means you are less likely to need a tool to notice.

If you disagree with the weights, the area by area table gives you every underlying measurement, and you are welcome to reach your own conclusion from those. The single number is a summary, not a verdict.

What The Grades Mean

One caveat that stops people misreading a good score. A high number means nothing measurable is wrong at the moment you measured it, which is not the same as nothing being wrong. Intermittent faults are the hardest thing to catch on any test bench, and if your controller misbehaves once every twenty minutes, a two minute report can easily miss it entirely. Run it again when the fault is happening rather than when it is not.

Reading The Weakest Area First

The single most useful line in the results is the weakest area, and it is worth explaining why that is more valuable than the total.

Controllers rarely fail evenly. Almost every pad that scores badly is scoring badly because of one thing, with the other seven areas sitting somewhere between fine and excellent. That matters enormously for what you do next, because a single failing area usually points at a single part, and single parts are cheap and replaceable. A controller failing in five areas at once is a different conversation entirely, and usually a shorter one.

So the practical reading order is weakest area first, then the score. If the weakest area is stick travel and everything else is above eighty, you have a range problem in one stick and you know exactly where to go next. If the weakest area is connection and everything else is high, your controller is probably fine and your cable is not.

There is one pattern worth watching for specifically. When resting position and signal noise are both poor on the same stick, that combination usually points at contamination on the sensor track rather than wear, which is genuinely good news because contamination cleans off. When resting position is poor and noise is clean, that is more often physical wear, and cleaning is less likely to help.

What A Score Cannot Tell You

Worth being direct about the limits, because a single number invites more confidence than it deserves.

It cannot see intermittent faults. Two minutes is a snapshot. A stick that drifts for ten seconds every half hour will pass this comfortably. If you are chasing something occasional, run the relevant single test for a much longer stretch instead of running this one repeatedly.

It cannot measure what your games receive. Everything here has already passed through the controller firmware and your operating system driver, both of which normalise the values before you ever see them. A reading of 1.00 means the value saturated, not that the stick hit its mechanical stop. This makes comparisons far more reliable than absolutes. The same pad today against six months ago is meaningful. Your number against a stranger's number is not.

It cannot tell you whether a repair is worth it. That depends on the part price, your soldering confidence and what a replacement costs where you live, none of which this page knows. It can tell you what needs fixing, which is the part that is genuinely hard to work out alone.

It cannot judge feel. Stick tension, button travel, trigger resistance and how a pad sits in your hands are all real and none of them are measurable from a browser. A controller can score ninety five and still feel wrong to you, and you are not imagining that.

It cannot separate your hands from your hardware. Several stages depend on you pushing properly, releasing cleanly and holding still. The tool flags the obvious cases, like a controller that moved during the hands off stage, but it cannot tell an under pushed stick from a short one. When a result surprises you, running it a second time is almost always worth the ninety seconds.

Turning The Report Into Something You Can Act On

A score is only useful if it changes what you do next. Here is the honest mapping from each weak area to a first move, in the order most likely to work.

If resting position scored badly

Start with compressed air around the base of the stick while working it through its full range, because debris on the sensor track causes more resting offset than wear does and it costs nothing to rule out. If the offset survives that, check whether it is stable by measuring it three times. A stable offset can be corrected in software, and the controller calibration tool produces the exact numbers to feed into Steam Input or a game settings menu. An offset that changes between runs cannot be corrected that way and points at a part rather than a setting.

If signal noise scored badly

Change the port first, moving to a rear motherboard port and off any hub, then change the cable. Both are free and both solve more noise problems than people expect. If the noise survives both, it is coming from the stick itself and the stick jitter test will tell you whether it is a steady hiss or occasional jumps, which have completely different causes.

If stick travel scored badly

Push harder into the gate and measure again, because technique accounts for most short readings. Then close any remapping software, since a range setting in a profile can quietly cap your output. If the shortfall is real, either clean the gate in case something is obstructing it or scale the range in software, and check what your deadzone is costing you on top with the joystick deadzone test.

If return to centre scored badly

There is no software fix for overshoot, because it is a spring and damping problem in the gimbal. Raising a deadzone hides how far the overshoot travels without stopping it happening. If it appeared suddenly after a cleaning attempt, contact cleaner stripping the damping grease is the usual explanation and it is worth knowing before you clean the other stick too.

If buttons or the D-pad scored badly

Cleaning fixes far more of these than people expect, because most button faults are contamination on the contact pads rather than wear. If cleaning helps for a week and the fault returns, the rubber membrane has worn out, and replacing it is cheap and needs no soldering. Both jobs share a teardown with the drift repair, and the disassembly steps and safety notes in the controller drift repair guide apply directly.

If connection scored badly

This one is rarely the controller. Test wired, move off any hub, and check the pack, since a controller running low reports less reliably than a charged one. The connection stability test is the direct measurement when dropouts only happen wirelessly, and the battery health test rules out the obvious before you go looking for anything clever.

Using The Report For Warranty, Resale And Buying Used

The exported report is written to be readable by somebody who has never seen this page, which is deliberate, because the three most useful things you can do with it all involve showing it to somebody else.

Warranty and support

Support agents deal with vague descriptions all day. A report with a date, a controller name, specific measurements and a clear weakest area moves a conversation along considerably faster than saying the stick feels wrong. Two reports taken a month apart, showing the same fault getting worse, is stronger still, because it demonstrates a trend rather than a bad day.

Selling a controller

Attaching a health report to a listing does two things. It answers the question every buyer wants to ask, and it quietly signals that you actually checked rather than assuming. A pad with a documented eighty two and a noted weak trigger sells more easily than an undocumented one described as working fine, because the buyer knows exactly what they are getting.

Buying used

This is where the two minutes pays for itself most obviously. Run the check before a return window closes rather than after. Drift and short range are both invisible in a listing photo and invisible in a quick press check, and neither improves on its own. If you are buying in person and can bring a laptop, running this in front of the seller settles the question entirely.

Tracking A Controller Over Time

The single most valuable thing you can do with this page costs nothing and takes ninety seconds, and almost nobody does it. Run the check on a controller when it is new, save the report, and forget about it.

The reason is that controller wear is gradual and your hands adapt to it. A stick that loses two percent of its range every few months never feels different from one week to the next, because you compensate without noticing. By the time it feels wrong it has usually been wrong for a long time, and you have no idea what it used to be.

A baseline solves that completely. It turns every later reading into a comparison rather than a guess, and comparisons are where the useful information lives. A resting offset of 0.03 means very little on its own. A resting offset of 0.03 on a pad that read 0.004 last spring means something specific, and it means it clearly enough to act on.

Twice a year is plenty for a pad in normal use, plus one run after any repair so you know whether the repair actually helped. Keep the JSON exports rather than screenshots, since they hold every underlying number rather than just the summary, and they are small enough that a folder of them costs nothing.

Method And Honest Limits

Sampling

This page reads your controller in step with your display refresh, so a 60Hz screen gives about sixty readings a second and roughly sixteen milliseconds between looks. That is plenty for measuring positions, offsets, ranges and button states, all of which last far longer than one frame. It is not enough to measure how fast a signal oscillates, and it means the connection score describes how evenly the browser received readings rather than how often the pad actually reported. The polling rate test measures the rate underneath this one directly.

Normalisation

Every value here has already been through the controller firmware and the operating system driver, both of which reshape it. That is why the report emphasises comparisons over absolutes, and why a reading of 0.94 might be a genuinely short stick or a driver making a scaling choice. It does not undermine the score, because your games receive the same reshaped values this page does, but it does mean you should not treat these numbers as physical measurements of the hardware.

Skipped stages are not failures

If a controller has no D-pad, no triggers or no rumble motors, skip those stages. They are removed from the weighting rather than scored as zero, and the report records the coverage so nobody reading it later mistakes a skipped stage for a failed one.

The rumble check is subjective by design

There is no way for a web page to sense whether a motor spun, so the tool fires the actuator and takes your word for what happened. That answer is recorded as a note and kept out of the score, because a number built partly on somebody's opinion is worse than a number that admits where it stops.

Run it twice when something surprises you

Several stages depend on you doing your part properly. A short travel reading is more often an under pushed stick than a worn one, and a poor release score can come from steering the stick back instead of letting go. When a result does not match how the controller feels, the second run is nearly always the more accurate one.

Browser And Platform Notes

• Chrome and Edge report controllers most consistently. Firefox works but sometimes places buttons and axes on different indices for the same pad, so check the setup selections match your controller before reading anything into a low score. Safari is the least predictable of the three.
• Keep this tab focused throughout. Background tabs get throttled and sampling stops, which will cost you a peak in the travel stage and produce a false stutter in the connection score.
• If nothing registers at all, the browser may not see the device rather than the pad being faulty, and the controller checker confirms that in a couple of seconds.
• Third party pads with an XInput and DirectInput toggle, usually a button combination involving Home, expose completely different indices in each mode. Test in the mode you actually play in, and use the controller mapping tool if you are unsure which axis numbers your pad is using.
• Rumble needs a browser that exposes the vibration actuator. Chrome and Edge generally do, others often do not, and the tool will say so rather than pretending the motors failed.
• Everything runs locally in the page. Nothing about your controller or your report leaves your device, and the exports are files your own browser writes for you.

Questions People Actually Ask

How long does the whole thing take?

About two minutes at a normal pace, most of which is you moving sticks rather than the tool waiting. The quick check trims it to under a minute by running only the hands off, travel and button stages, which between them catch the large majority of real faults.

My controller scored 68. Should I replace it?

Not yet. Read the weakest area first, because a score in that band is usually one area dragging down seven healthy ones. Most controllers in the sixties respond to cleaning, and the ones that do not usually need a single inexpensive part rather than a replacement.

Why did my brand new controller not score 100?

Because parts are made to a tolerance rather than to a value, and a fresh stick reaching 0.97 in one direction is sitting at one end of that tolerance rather than being faulty. Anything above ninety is a healthy pad. Chasing the last few points is not a good use of an evening.

Can I trust the score if I tested over Bluetooth?

Broadly yes for positions and ranges, less so for the connection score and for anything that depends on catching a peak. If a wireless run produces a surprising result, repeat it wired before acting on it, because a dropped packet and a genuine fault look identical from a browser.

The score changed a lot between two runs.

That usually means one of the runs was contaminated rather than that the controller changed. The most common culprits are holding the pad during the hands off stage and not pushing the sticks fully into the gate during travel. Run it a third time and take the two that agree.

Does the report include anything personal?

Only the controller name your browser reports, which is a model identifier rather than anything about you. Everything runs locally and nothing is uploaded, so you can share the file freely or strip that line out before you do.

Should I run this before or after calibrating?

Before, always. Calibration corrects measurements, so calibrating first and testing afterwards gives you a report on your settings rather than on your hardware. Measure the raw controller here, then decide what to correct.

Does it work on a phone or tablet?

Yes, with a controller connected over Bluetooth or through an adapter. Everything scales down, though rumble support is much less common on mobile browsers and that stage will usually report itself as unavailable.

Related Tests On Gpadtester

This page is the overview. When one area scores badly, these are the pages that take that area apart properly and tell you what to do about it.

Analog sticks

Buttons and triggers

Connection and response

Read next

How to fix controller drift covers the teardown, the cleaning and the part replacement. The same disassembly reaches the sticks, the button contacts and the D-pad membrane, so one guide covers most of what a poor report will ask you to do.

Other hardware

Away from the controller: keyboard test, mouse test, mic test, frame rate test and refresh rate test.