The Sensor

Introduction

• Sensor = the 'eye' of the mouse. It sits on the bottom and tracks movement.
• Your mouse has a tiny camera. It takes THOUSANDS of photos of your pad every second, compares each photo with the last one, and works out which way you moved.
• Bad sensor = crosshair jumps. Like aiming with blurry glasses.

How it works (side cut view):

1. Optical vs Laser

• Laser: older sensors (around 2012, e.g. Avago ADNS-9800). Tracked on more surfaces but read too deep into the pad texture -> unwanted acceleration.

  ~ like: a camera that sees every fibre of the cloth and gets confused.

• Optical: what almost every serious gaming mouse uses today.

2. Who makes the sensors?

• PixArt (Taiwan): the KING. Nearly every gaming mouse uses a PixArt sensor. Became a patent powerhouse after buying Avago's optical sensor division.

  e.g. Zowie, Vaxee, Pulsar, Lamzu + dozens more built their wireless mice around the PAW3395.

• Logitech: makes its OWN sensor called HERO. Only in Logitech mice.
• Razer: 'Focus' sensors, co-developed WITH PixArt, sold under Razer's name.
• Ninjutso: 'AimNinja' sensors, developed with PixArt (more in section 7).
• Glorious: 'BAMF' sensor in the Model O Wireless, also made with PixArt.

-> Razer Focus, AimNinja, BAMF = brand names on PixArt-made tech.

• SteelSeries: 'TrueMove' sensors - SteelSeries' own name for its sensors.
• Corsair: 'Marksman' sensor - spec lists show it as PixArt-based.
• Budget brands: cheap mice sometimes use lesser-known sensors -> check before you buy!

~ like: PixArt = big engine factory. Razer pays for a special engine first. Others buy the regular engine. Logitech builds its own.

3. Specs: real vs marketing

• DPI / CPI: how far the cursor moves per inch. Boxes say 30K-50K, pros mostly play 400-1600. MARKETING
• IPS: max tracking speed. 650 IPS = ~16.5 m per second - no human arm moves that fast. MARKETING
• LOD: lift-off distance: how high you lift before tracking stops. REAL
• Motion sync: syncs sensor data with the PC -> smoother, tiny latency cost. REAL
• Glass tracking: does it track on glasspads? REAL (important if you use a glasspad).
• CPI deviation: set 800 DPI, real result 760? Most listings don't mention this - you can test it yourself.

Hidden settings (should usually be OFF for gaming):

• Smoothing: sensor averages movements -> feels floaty + slightly delayed. Flagship sensors have none.
• Acceleration: cursor goes further when you move faster -> same hand move, different result. Bad for aim.
• Angle snapping: straightens your lines. Nice for drawing, bad for aim -> keep OFF.
• Ripple control: hides jitter at very high DPI but adds smoothing -> usually OFF.

~ like: autocorrect on your phone - helpful for typing, annoying when you want EXACTLY what you typed.

DPI inflation: the 'fake' big numbers

• Every sensor has a RATED (native) resolution. Going above it means the MCU/firmware 'guesses' extra counts = interpolation.

~ like: zooming a small photo on your phone - it gets BIGGER, not SHARPER.

e.g. PAW3950 is rated 30,000 DPI, yet some listings advertise it as 42,000 DPI -> the extra is beyond the sensor's rating.

• Native steps: DPI values the sensor was designed to output cleanly (often round steps like 400 / 800 / 1600 / 3200). Odd values like 1030 may be scaled.
• Why it can hurt: guessed counts can add tiny jitter or inconsistency - some firmware adds smoothing to hide it.
• Honest update: sources DISAGREE on flagships: some testers say modern sensors (3395 / 3950 / Razer) track cleanly even at their high RATED numbers. The clearer red flag = numbers ABOVE the sensor's rating, or a budget sensor with huge DPI.

-> Before you buy: check the RATED DPI, check the sensor name, and remember pros mostly use 400-1600.

Sensor angle (rotation) - why your 'straight' swipe goes crooked

• Sensor angle / rotation: a setting that rotates the sensor's output a few degrees so YOUR natural side-to-side swipe becomes perfectly horizontal on screen.

  e.g. Razer has a free online rotation tool: swipe side to side 10 times, it recommends an angle to enter in Razer Synapse.

• Not the same as angle snapping: angle snapping FORCES lines straight (bad for aim). Angle rotation just CORRECTS your grip tilt.
• Sensor position matters: a small tilt on a front-mounted sensor causes a bigger cursor drift than on a centre sensor.

~ like: a crooked camera on a tripod - you don't change the view, you just straighten the camera.

Motion Sync vs Frame Sync

• Motion Sync: lines up the MCU's reports with the moments your PC asks for data -> smoother, more consistent tracking (tiny latency cost). Common on PAW3370 / 3395 class sensors.
• Frame Sync: goes one step further: also lines up the SENSOR's photos with the MCU + polling cycle. Before, sensors still took extra photos between polls.

  e.g. Razer Focus Pro 50K Gen-3 (Viper V4 Pro) - Razer calls Frame Sync the industry's first, with faster, more consistent aim + better battery (no wasted sensor photos).

  e.g. Ninjutso AimNinja 2 Pro also advertises 'FrameSync' - described as next-gen MotionSync.

~ like: Motion Sync = the waiter brings your plate exactly when you sit. Frame Sync = the chef also cooks it at exactly the right moment - no food wasted.

~ like: same idea as G-Sync / FreeSync on monitors: sync the hardware to the data.

4. PixArt: naming + ladder

How to read a PixArt name, e.g. PAW3955:

P	PixArt
MW / AW	MW = mostly wired, AW = made for wireless
3955	just a model code

!! Bigger number does NOT always mean better. PAW3311 is budget even though it came after the PMW3360.

PMW vs PAW - do wired mice use PMW sensors?

PMW (older)	made mostly for wired
PAW (newer)	made for wireless, now used in BOTH

• PMW3360
• PMW3366 (Logitech only)
• PMW3389
• PAW3311 (budget)
• PAW3370
• PAW3395
• PAW3950 / 3955

• Old days: PMW for wired, PAW for wireless (PAW = low power).
• Today: the newest + best sensors are all PAW, so WIRED mice use PAW too. Low power doesn't hurt a wired mouse!

Rough tier match (PMW vs PAW):

PMW3389	↔	PAW3370	similar class (~400 IPS era)
PMW3360	↔	PAW3311	older / budget class (rough)
no PMW match	↔	PAW3395 / 3950 / 3955	PAW only

Endgame Gear wired history (PMW -> PAW):

XM1 (original)	→	PMW3389	widely listed
XM1r	→	PAW3370	refresh = switched to PAW
OP1 / OP1 8k	→	PAW3395
OP1 8k v2 / XM2 8k v2	→	PAW3950 'egg'

-> Endgame Gear is one of the best-known WIRED brands - and even they moved to PAW sensors.

(they also make wireless now: XM2we, OP1we, XM2w 4k v2, OP1w 4k v2)

PixArt ladder:

1. PAW3955 - 2026, newest flagship
2. PAW3950 - 2024, flagship
3. PAW3395 - 2022, still excellent
4. PAW3370 - older flagship
5. PMW3360 / 3389 - old classics
6. PAW3311 - budget

Mice using them:

• 3955: G-Wolves (Fenrir Pro, Lycan 8K, Warg 8K, VUK Medium, HTS Plus Pro, HTX Mini Asym) + VGN, Akko, ATK, Xiaomi
• 3950: Lamzu, Endgame Gear, WLMouse, ASUS
• 3395: Zowie, Vaxee, Pulsar, Lamzu, Ninjutso
• 3311: many budget mice

What improved each time:

3370	→	3395	+ hardware motion sync
3395	→	3950	LOD 1.0 -> 0.7mm, ~8% more efficient, reliable on glass
3950	→	3955	40K DPI, 60G, adjustable virtual sensor centre

-> for ~80% of buyers, the 3395 is the best-value choice.

PAW3950 has many versions (confusing for buyers!):

• PAW3950: the normal one: 30K DPI, 750 IPS, 50G.
• PAW3950 MAX: spec lists show 42K DPI, 750 IPS, 70G.
• PAW3950 Ultra: spec lists show 45K DPI, 900 IPS, 85G.
• PAW3950 HS: used in WLMouse Strider.
• PAW3950 🥚: Endgame Gear's custom version (the 'egg' sensor).

~ like: same phone model with different storage + camera versions - check WHICH version you're buying.

Endgame Gear wired mice:

XM1r	→	PAW3370	Kailh GM 8.0 switches
OP1	→	PAW3395	STM ARM Cortex-M4 MCU, 1000Hz
OP1 8k / XM2 8k	→	PAW3395	Nuvoton MCU, 8000Hz
OP1 8k v2 / XM2 8k v2	→	PAW3950 🥚	Nuvoton, 8000Hz

5. Logitech: naming + ladder

• HERO: High Efficiency Rated Optical -> the goal is saving battery.
• 16K / 25K: = max DPI. HERO 25K = 25,600 DPI. (Different from PixArt's code!)
• HERO 2: a brand-new generation, not just a DPI bump.

Logitech ladder + mice:

1. HERO 2 - G Pro X Superlight 2
2. HERO 25K - Superlight, G502 X, G502 HERO, G703, G903, G309
3. HERO 12K - G305 (budget wireless king)
4. PMW3366 - original G Pro, G502 Proteus, early G403/G703/G903
5. 8K budget sensor - G203 / G102 (not HERO)

What improved each time:

HERO (2016)	→	HERO 16K (2018)	new lens, no accel/smoothing
HERO 16K	→	HERO 25K (2020)	more DPI via FREE software update
HERO 25K	→	HERO 2	888 IPS, 44K, dual array for lifts/tilts

Late 2024: Logitech's '8/88/888 update' gave HERO 2 -> 8kHz polling, 88G, 888 IPS.

6. Razer: naming + exclusive deals

How to read a Razer name, e.g. Focus Pro 50K Gen-3:

Focus	sensor family
X / + / Pro	tier: X = budget, + = older flagship, Pro = flagship
50K	max DPI (like Logitech)
Gen-3	generation

Before 'Focus', Razer used names like the 5G sensor (= PMW3389, used in DeathAdder Elite).

Razer name -> PixArt twin:

5G	→	PMW3389	widely reported
Focus+ 20K	→	PAW3399	exclusive version
Focus Pro 30K	→	PAW3950	confirmed (Basilisk V3 Pro)
Focus Pro 35K	→	no public twin	Viper V3 Pro, DeathAdder V3 Pro
Focus Pro 45K	→	not confirmed	newer Razer mice
Focus Pro 50K	→	newest	Gen-3, Frame Sync (Viper V4 Pro)

The exclusive deal pattern:

Focus+ (2019)	→	PAW3395 (2022)	Razer first, then everyone
Focus Pro 30K (2022)	→	PAW3950 (2024)	Razer first, then everyone
Focus Pro 35K/45K/50K	→	not public yet	watch this space

7. Ninjutso: AimNinja

AimNinja works like Razer Focus: Ninjutso's own name on a sensor developed WITH PixArt.

How to read the name, e.g. AimNinja 2 Pro 45K:

AimNinja	sensor family (Ninjutso)
2	generation
Pro	tier
45K	max DPI

AimNinja ladder + mice:

1. AimNinja 2 Pro 45K - Sora V3 (40g)
2. AimNinja (upgraded) - Sora V2 8K
3. AimNinja (gen 1, based on PAW3950) - Ten, Ten Air

• AimNinja (gen 1): based on the PAW3950, 30,000 DPI. Ten Air pairs it with 8000Hz polling.
• AimNinja 2 Pro 45K: Ninjutso calls it an exclusive dual-optical engine with gen-2 motion sync, and claims up to 52% more power efficient than PAW3950.

(52% = brand claim -> test battery life yourself before repeating it)

8. Which sensor = which? (rough tiers)

HERO 2	↔	PAW3950 / 3955	today's top class
HERO 25K	↔	PAW3370	25K was a software bump
HERO 16K	↔	PMW3389	both ~16K DPI, ~400 IPS
PMW3366	↔	PMW3360	same classic era
AimNinja gen 1	↔	PAW3950	Ninjutso Ten / Ten Air
AimNinja 2 Pro	↔	PAW3950 Ultra	specs match, not confirmed

AimNinja 2 Pro also shares 'FrameSync' with Razer's newest Focus Pro 50K -> same newest generation.

PAW3395 sits between HERO 25K and HERO 2. Razer Focus Pro 30K = PAW3950 (top row).

9. Logitech vs PixArt (companies)

• Who uses it: Logitech -> only Logitech mice. PixArt -> almost every other brand.
• Naming: Logitech number = DPI. PixArt number = model code.
• Focus: Logitech -> battery efficiency. PixArt -> performance at every price.
• Upgrades: Logitech sometimes via software. PixArt -> new chip each generation.

~ like: Logitech = Apple (own chip, own products). PixArt = Qualcomm (sells chips to everyone).

10. Two sensors in one mouse?

People say 'dual sensor' for 3 different ideas:

• Type 1: 2 full sensors, front + back. Compare how much each moved -> the computer knows if you ROTATED. Used in research (some built with Arduino + 3D printer).

  ~ like: 2 people at the front + back of a boat. Same move = straight. Different move = the boat turned.

• Type 2: main sensor + a 2nd sensor that measures lift height (down to 0.5mm). Stops the cursor skipping when you lift + land.

  e.g. SteelSeries Rival 600. Razer tried dual sensors back in the early 2010s too.

  ~ like: one eye watches the road, the other watches how high you jumped.

• Type 3: ONE sensor with two sensing areas inside. Better tracking when you lift or tilt.

  e.g. Logitech HERO 2 ('Dynamic Spot Edge Detection', dual-array) + Ninjutso AimNinja 2 Pro ('dual optical engine').

How is it possible? The MCU (brain) gets data from both parts and combines it. Brands don't publish the inside details - just look for the BENEFIT it claims.

11. Sensor position

• Where the sensor sits changes how the cursor reacts when you rotate or flick with fingers/wrist.
• No 'correct' spot - it's personal preference, but worth checking before you buy.

e.g. G-Wolves Lycan advertises a forward sensor position. PAW3955 mice can also move a 'virtual' sensor centre in software.

12. How to test a sensor in your first week

• CPI deviation: set 800 DPI, move exactly 10 cm along a ruler, compare cursor distance with a 2nd trusted mouse. Big difference = deviation.
• LOD test: stack sheets of paper under the mouse, add one at a time, note when tracking stops.
• Glass test: try it on a glasspad - jitter? spin-outs? (AimNinja 2 Pro + PAW3950 claim better glass tracking).
• Polling test: use a polling rate checker + MouseTester to see if 1K/4K/8K is stable.
• Battery test: if wireless: test battery at 1K AND 8K.

-> Same test setup every time (same pad, same skates, same DPI) = you'll actually know if it's good.

(specs of older sensors are approximate - double check on eloshapes.com or sensor.fyi before trusting the numbers)