MyGpadTester tests controllers in your browser, free, with no download. A test reads what the browser can see through the W3C Gamepad API. It does not read the circuit board directly, and it does not bypass your operating system. This page explains the method behind every guide on this site, so you can repeat a test and trust a comparison.
The authority reference for how the browser reads a controller is the W3C Gamepad API specification: https://www.w3.org/TR/gamepad/
A browser test measures Gamepad API timestamps and reported axis values. Each stick reports an X axis and a Y axis from minus 1.0 to 1.0, with 0.00 at centre. A button or trigger reports its own value as the tester shows it. The test shows the reported value, not the physical track inside the stick, so a calibration error can look like a hardware fault until it is retested.
A browser result is gated by display refresh and frame timing. On a 60Hz display, a new frame arrives about every 16.7ms, so a browser observed rate can sit near that frame rate even when the hardware rate is higher. That is why a browser result is a comparison tool, not a hardware certificate. It can compare wired against wireless, or one controller against another in the same browser. It cannot certify a true USB hardware rate, and it cannot measure true end to end lag from thumb to screen.
Your test data is processed in your browser to run the test. It is not uploaded to our server to run the test.
These rules apply to polling rate, drift, lag, and deadzone alike. A result that breaks them is not a verdict yet.
Polling rate is how often a controller reports input, in hertz (Hz). The nominal report interval is maths, Tier 1: 1000 divided by Hz. So 125Hz means an 8ms interval, 250Hz means 4ms, 500Hz means 2ms, and 1000Hz means 1ms. That interval is the expected time between reports, not the total delay from a button press to a screen change.
Method: connect wired first, open the Polling Rate Test, press any button to wake the Gamepad API, then move a stick in slow, continuous circles for 20 to 30 seconds. Read average Hz, peak Hz, and jitter together. A single peak proves nothing. Judge by the average plus how steady the gaps are, after a repeat run.
Typical ranges, Tier 3, browser observed, not manufacturer specifications: a PS5 DualSense is about 250Hz native wired, and reaches 1000Hz only with PC tools, wired only. An Xbox Series controller is about 125Hz, lower and more variable on Bluetooth. A Nintendo Switch Pro controller is about 125Hz. A wired figure near the typical range is normal, not a fault. A fault is instability: wild jumps, long dropouts, or a number far below the range for your connection, confirmed after changing one variable and retesting.
Bluetooth can throttle when sticks are still, on some devices, firmware, and browsers. That is why the test uses continuous movement. A still stick drop on Bluetooth, on its own, is not proof that a controller is slow.
Stick drift is when a controller reports stick movement even though you are not touching the stick. The test is simple on purpose. Put the controller flat on a stable surface, hands off, open the Stick Drift Test, press any button, and leave both sticks untouched for 30 to 60 seconds. Read X and Y at rest for each stick separately, then repeat after a restart, or wired if the first result was borderline.
Use the house scale for every reading. These are tester conventions, Tier 3, not a Sony, Microsoft, or Nintendo specification. Flickers within plus or minus 0.02 to 0.03 are normal sensor noise, not drift. A steady offset from plus or minus 0.05 to 0.10 in the same direction is measurable mild drift. A steady offset above plus or minus 0.10, or a value that keeps creeping in one direction, is clear drift. Random jumps, even untouched, point to debris, a loose stick, or a failing sensor, so clean carefully and retest wired before deciding.
A deadzone hides drift, it does not repair it. Raising a deadzone stops visible movement, but it reduces fine aim precision, and the underlying wear is still there. Test drift first, before you raise a deadzone. A single glance is not evidence. A repeatable number in the same direction is drift, and a random flicker that disappears is noise.
Input lag is the total delay from moving a stick or pressing a button to seeing the action on screen. The chain has six links: controller sensor and firmware, polling interval, connection, operating system and software layers, game processing, and display. The slow link is found by changing one link at a time, not by guessing.
A browser test can measure the controller's contribution only. Method: connect wired first and open the Input Lag Test. Move a stick or press a button continuously for a full 10 seconds. Read four numbers together: minimum, average, 95th percentile, and maximum browser intervals. Then repeat wireless with the same motion, and repeat wired once more. The gap between the wired and wireless pairs is the connection's contribution. A low average with a high 95th percentile or maximum means spikes, not steady slowness, and that needs a different fix.
Use these bands only as a browser estimate guide, never as end to end totals. Under 8ms is excellent, 8 to 16ms is normal, and above 16ms means investigate the controller link. A browser test cannot measure the instant your thumb moves, game engine time, or display time. True end to end measurement needs hardware, such as a high speed camera or a photodiode. Do not buy a new controller on one browser run.
Polling interval is one slice of the chain, and it is maths, not a measurement of the whole chain. Refresh rate is a separate variable. A higher refresh display shortens the gap between frames, but it does not fix a bad wireless link on its own.
A deadzone is the centre area a game ignores, so a resting stick does not move your character. Inner deadzone decides where input starts to count. Outer deadzone decides where input reaches full speed. Axial handling decides whether X and Y are judged separately or together. Deadzone size, sensitivity, and response curve are three different settings, and changing one does not replace the others.
Method: open the Stick Deadzone Test, leave the stick untouched for 30 to 60 seconds, and read the resting dot first. Then move one stick slowly in a full circle to the edge and read the outer edge trace. Set the inner deadzone just above the resting value, test small tilts to confirm precision is still there, then repeat the whole process for the other stick separately. Retest after a restart or wired connection if the first result was borderline.
Starting guides, labelled as tester guides, never vendor specifications: start from resting value plus 0.01. Use an outer deadzone of 0.95 to 0.98 as a starting guide for clipped corners. If you need an inner deadzone above about 0.20 to hide movement, the stick needs repair or replacement, not more tuning. The same number behaves differently per stick, per game, and per response curve. That is why this guide sets a value per stick by testing, not by copying one universal number.
A browser test diagnoses, it does not repair. It cannot identify the exact worn part inside a controller, and it cannot measure inside the controller. It cannot certify a hardware rate, a firmware version, or a lab end to end total. It cannot settle a repair or replacement decision on its own, and it cannot replace a maker's own software or a hardware tester where a certified number is needed.
If a result is borderline, retest wired, in the same browser, with the same motion. Confirm it a second way, such as a second game or a platform check, before you open anything. If several faults appear together, run the Gamepad Health Checker for a full screening before you buy or repair anything.
Check warranty before you open anything. If a controller is still covered, claim it first, because opening it can void it. This site does not recommend opening a controller. Power off and disconnect the controller before cleaning around the stick base. Use compressed air, and only a small amount of electronics contact cleaner if you are comfortable using it. Do not flood the module, and do not use WD-40 or household cleaners. If the battery is swollen, stop, do not charge it, never open it, and contact the maker.
Marcus Sterling, Lead Engineer and Founder, MyGpadTester, Austin, Texas, USA. Marcus Sterling is a Certified Hardware Analyst with over 12 years of hands-on experience in gaming peripheral diagnostics. He is certified by iFixit as a Master Tech. His affiliations include Associate Member of IEEE, Consumer Electronics Society. His specialization is Advanced TMR Magnetic Sensor Calibration and Latency Testing. He was formerly Lead Tech at Performance Labs TX in Austin. His certifications include ESD Professional and Master Hardware Diagnostic Certified. He built MyGpadTester to give every gamer the same diagnostic tools he uses on his own workbench.
Follow Marcus Sterling on X: https://x.com/marcussterlin
Tests follow this methodology: same browser, same connection, wired first, retest before deciding. Results are browser estimates via the W3C Gamepad API, not lab measurements.
It measures Gamepad API timestamps and reported axis values in your browser. It compares runs, it does not certify hardware rates or lab end to end lag.
That is the typical wired range for Xbox controllers, Tier 3 typical, browser observed. Judge stability, not the number alone, and use a wired run as the control.
That is the typical native Windows wired rate, Tier 3 typical. On PC, 1000Hz usually needs PC tools, wired only, and it does not change how the controller behaves on a console.
Near 0.00 with flickers within plus or minus 0.02 to 0.03 is noise. A steady offset at plus or minus 0.05 or higher in the same direction is drift on this site's house scale, which is a tester convention, not a vendor specification.
No. A browser cannot see your thumb move or pixels change. It can measure the controller's browser observed contribution and compare wired against wireless. True end to end needs hardware measurement.
No. A test diagnoses and measures. Fixing needs cleaning, recalibration, a deadzone as a bridge, or module repair or replacement, each in its own guide, with warranty checked first.
Last updated: 27 October 2026. Figures are either Tier 1 maths, Tier 3 typical ranges, or labelled tester guides, as stated in each section. Browser tests measure Gamepad API timestamps, not laboratory end to end latency. Retest after any fix and compare before and after numbers recorded the same way.