Holding Torque: Standard Servos vs Micro Servos

If you’ve ever stripped a gearset on a 9-gram plastic servo during a routine flap deflection test, you know the exact moment when “good enough” becomes “not even close.” That sickening click-click-click of stripped nylon teeth is the sound of a micro servo motor failing to hold its commanded position under load. And yet, in the same bench test, a standard-size servo—same voltage, same PWM signal, same theoretical angle—barely breaks a sweat. The difference isn’t magic. It’s holding torque, and it’s the single most misunderstood spec in the entire RC, robotics, and automation hobbyist ecosystem.

Let’s get one thing straight: holding torque isn’t just “how hard the servo pulls.” It’s the static torque a servo can maintain when the output shaft is stationary and an external force tries to rotate it back. For a micro servo motor, that number is often quoted as 1.8 kg·cm at 4.8V or 2.2 kg·cm at 6.0V. For a standard servo, you’re looking at 6 to 15 kg·cm. But raw numbers don’t tell the story. The real story is about torque density, stall behavior, gear train geometry, and the brutal physics of tiny motors spinning at 10,000 RPM.

The Micro Servo Motor: A Physics Problem in a Tiny Box

A typical micro servo (like the ubiquitous SG90 or MG90S) houses a coreless or pager-style DC motor that’s barely 7mm in diameter. That motor spins at a ridiculous no-load speed—often 8,000 to 12,000 RPM—but produces almost zero torque at the shaft. A standard servo’s motor is maybe 15mm across, spins slower, but produces proportionally more raw torque. Then comes the gear reduction.

Here’s where the magic (and the weakness) lives:

• Standard servo gear reduction: Typically 200:1 to 300:1 using metal or high-durability nylon planetary gears. The final stage has a large output gear with wide teeth.
• Micro servo gear reduction: Often 300:1 to 500:1 just to get usable torque, but the gears are tiny. A micro servo’s output gear might be 6mm in diameter with 0.4mm module teeth.

That high reduction ratio is a double-edged sword. It multiplies the motor’s weak torque into something usable, but it also multiplies backlash and increases the mechanical stress on every single tooth. When you apply an external load to a micro servo’s horn, that load is transmitted through the output gear, into the intermediate gears, and finally onto the motor pinion. Every gear tooth carries the full reflected load. If the gear train has any slop—and most micro servos have plenty—the load is concentrated on one or two teeth at a time. That’s how you strip gears.

Stall Torque vs Holding Torque: The Dirty Secret

Manufacturers love to quote stall torque—the torque at which the motor can no longer rotate and just sits there burning current. But holding torque is different. Holding torque is the maximum external torque the servo can resist without being back-driven, while the motor is actively trying to maintain position. For a brushed DC motor, holding torque is essentially equal to stall torque at that specific PWM duty cycle. But here’s the catch: a micro servo motor’s stall torque is measured at a specific current limit. When you command a micro servo to hold position against a heavy load, the motor draws maximum current, heats up, and the magnets (if they’re cheap ferrite) start to lose strength as temperature rises. Within 30 seconds, that 2.2 kg·cm holding torque can drop to 1.4 kg·cm. A standard servo with a larger motor, better magnets, and a metal heat sink doesn’t suffer that dramatic a drop.

Why Holding Torque Matters More Than Speed for Micro Servos

You might think: “I’m building a tiny robot arm. It only needs to lift 100 grams. Why do I care about holding torque?” Because dynamic loads are liars. A 100-gram load at the end of a 5cm horn creates 0.5 kg·cm of static torque. But when the robot arm swings, accelerates, or hits an obstacle, that load can spike to 3 or 4 times the static value. And if you’re using a micro servo motor without a metal gearbox, the gear teeth flex under that dynamic spike. Even if the teeth don’t strip, the flex introduces positional hysteresis—the output shaft settles at a slightly different angle than commanded. For a camera gimbal or a pan-tilt mechanism, that means jitter. For a robotic gripper, that means dropped objects.

Real-World Example: The 3D-Printed Robot Hand

Let’s say you’re building a five-fingered robotic hand using micro servos for each finger. Each finger joint uses an MG90S (metal gear micro servo, 2.0 kg·cm holding torque at 6V). You think that’s plenty. But then you realize the tendon routing uses a 10mm pulley. The fingertip needs to exert 200 grams of pinch force. That’s 0.2 kg·cm if the tendon is perfectly perpendicular. But at full flex, the tendon angle changes, the effective lever arm shortens, and the required torque jumps to 0.45 kg·cm. Still within spec, right? Now add the weight of the finger segment itself, plus the friction of the tendon through the guide tubes. You’re now at 0.6 kg·cm. Fine. But then you power the hand from a 4.8V BEC instead of 6V, and the holding torque drops to 1.6 kg·cm. You’ve got 1.0 kg·cm of headroom. That seems safe.

Until the finger hits an unexpected object. The motor stalls. The current spikes. The BEC voltage sags to 4.5V. The holding torque drops to 1.3 kg·cm. And the dynamic load from the impact is 1.5 kg·cm. The finger back-drives. The gear train slips. You hear that click. Now you’re replacing a gearset. This is the reality of micro servos: they are marginal by design. They work beautifully—until they don’t.

The Metallurgy of Holding Torque: Brass vs Steel vs Titanium

If you’re serious about micro servo motor holding torque, you need to look past the plastic vs metal gear debate. Here’s the hierarchy:

1. Nylon / POM gears: Lowest holding torque capability. They flex under load. They strip easily. They are fine for toy applications and zero-load test benches. The SG90’s nylon gears have a practical sustained holding torque of about 60% of the quoted stall torque.

2. Brass gears (MG90S, etc.): Better. Brass has higher yield strength than nylon, but it’s soft. Under repeated high-torque holds, brass teeth deform—they mushroom out. This increases backlash and eventually causes the gear train to bind. A brass-gear micro servo motor can hold its rated torque for about 10,000 cycles before you notice slop.

3. Hardened steel gears (rare in micro servos, common in mini servos like the DS3218): Steel gears in a micro form factor are hard to manufacture, but they exist in premium models (e.g., the Hitec HS-5086MH or the KST X08). These hold their rated torque indefinitely under normal conditions, but they are heavy. A steel-gear micro servo weighs 15-20 grams, which defeats the “micro” purpose for FPV drones or lightweight gliders.

4. Titanium gears (exotic, expensive, mostly marketing): Titanium has a high strength-to-weight ratio, but its fatigue life under cyclic holding loads is worse than steel. You’ll find titanium gears in high-end sailboat winch servos, but for micro servos, it’s mostly a gimmick.

Key insight: The gear material doesn’t increase the motor’s torque. It increases the transmission’s ability to survive that torque. A micro servo motor with plastic gears will still produce 2.0 kg·cm of holding torque—for about 3 seconds. Then the teeth strip. A metal gear version will produce the same 2.0 kg·cm for 10 minutes. But neither will match a standard servo’s ability to hold 8 kg·cm for hours without thermal derating.

Thermal Derating: The Hidden Killer of Micro Servo Holding Torque

Let’s talk about heat. A micro servo motor is essentially a tiny DC motor with a permanent magnet stator. When you command a hold, the motor doesn’t spin—it just pushes against the stall torque. That means the motor is drawing stall current continuously. For an SG90, that’s about 650mA at 6V. The motor’s internal resistance is maybe 6 ohms. Power dissipation = I²R = 0.65² × 6 = 2.5 watts. That’s a lot of heat for a package that weighs 9 grams. The motor case will hit 70°C in under 60 seconds of continuous holding.

At 70°C, the ferrite magnets in a cheap micro servo motor lose about 20% of their magnetic flux. Lower flux = lower torque constant (Kt) = lower holding torque. So that 2.2 kg·cm holding torque at 25°C becomes 1.76 kg·cm at 70°C. And if you’re using the servo in a confined space (like a robot arm joint), the heat has nowhere to go. The temperature climbs to 90°C. Now you’re at 1.5 kg·cm. And the nylon gears (if you’re using plastic) soften at 80°C. They deform. The gear mesh tightens. The motor draws even more current. It’s a death spiral.

A standard servo, by comparison, has a larger motor with more surface area, better cooling, and often a metal case. A standard servo can hold its rated torque for minutes without significant derating. A micro servo motor cannot. This is the fundamental reason why micro servos are not suitable for continuous-duty holding applications—like a pan-tilt camera that must keep a fixed position against wind, or a robotic leg that must support a static load while waiting for the next command.

The “One-Second Hold” Fallacy

Most hobbyists test holding torque by applying a force for one second and seeing if the servo holds. That’s a terrible test. Real applications require holding for 30 seconds, 2 minutes, or indefinitely. Here’s a simple rule of thumb:

• Micro servos: Safe for intermittent holding (less than 5 seconds) at 70% of rated torque.
• Micro servos: Safe for continuous holding (more than 30 seconds) at only 40% of rated torque.
• Standard servos: Safe for continuous holding at 80% of rated torque.

If you need a 2.0 kg·cm holding torque for a continuous application, you don’t buy a 2.0 kg·cm micro servo. You buy a 5.0 kg·cm standard servo and run it at 2.0 kg·cm. That’s called derating, and it’s the mark of an engineer who has learned the hard way.

Torque Density: The Micro Servo’s One True Advantage

So why do micro servos exist at all? Because of torque density—torque per unit weight. A typical micro servo motor with metal gears weighs 12 grams and holds 2.0 kg·cm. That’s 0.166 kg·cm per gram. A standard servo (like the MG996R) weighs 55 grams and holds 9.4 kg·cm. That’s 0.171 kg·cm per gram. They’re actually very close! But here’s the catch: the micro servo’s torque density is peak—it only holds that for a few seconds. The standard servo’s torque density is sustained—it holds that for minutes.

For applications where weight is the absolute priority and holding time is short—like a drone’s camera gimbal that is constantly moving, or a throwable robot that only needs to grip for 2 seconds—micro servos win. For anything that requires static stability, standard servos win. And there’s a middle ground: mini servos (like the 25-gram class, e.g., the HS-225MG). These offer 3.5 kg·cm holding torque at 25 grams, giving a torque density of 0.14 kg·cm/g, but with much better thermal behavior than micro servos. If you’re on the fence, go mini.

The Gear Train Ratio Trade-Off

Here’s a subtle point that many designers miss. A micro servo motor’s gear train is optimized for speed reduction, not strength. The ratio is high (say 400:1), which means the output shaft has a lot of mechanical advantage. But high ratio also means high reflected inertia. When you try to back-drive a micro servo, you’re not just fighting the motor’s holding torque—you’re also fighting the inertia of the entire gear train and the motor rotor. For a standard servo with a lower ratio (say 250:1), the reflected inertia is lower, so the servo is easier to back-drive when unpowered. But when powered, the standard servo’s larger motor produces a stronger magnetic detent torque.

What does this mean practically? A micro servo with a 500:1 gear train will feel stiffer when unpowered than a standard servo with a 200:1 gear train. But that stiffness is a false friend. The micro servo’s gears are smaller and have more backlash. Under a sustained load, the gears will wind up, and the output shaft will slowly creep (this is called gear train wind-up). A standard servo’s larger gears have less deflection, so the position hold is rock-solid.

How to Measure Holding Torque Without Fancy Equipment

If you want to evaluate a micro servo motor for your project, don’t trust the datasheet. Do this:

1. Rig a torque arm: Use a 1cm or 2cm arm on the servo horn. Attach a string and a hanging weight bucket.
2. Power at your intended voltage (not the max voltage—the voltage you’ll actually use).
3. Command a 90-degree position and wait 5 seconds for the servo to settle.
4. Gradually add weight to the bucket until the servo back-drives by 2 degrees (use a small pointer and a protractor).
5. Note the weight. That’s your practical holding torque.
6. Wait 30 seconds. If the servo starts to creep, note the new weight. That’s your sustained holding torque.

You’ll often find that a micro servo’s practical holding torque is 50-70% of the rated stall torque. And its sustained holding torque (after 30 seconds) is 30-40% of rated. This is the single most important data point you’ll ever collect about a micro servo motor.

The Future: Coreless Motors and Closed-Loop Micro Servos

The micro servo motor market is evolving. Newer micro servos (like the T-Motor TMM-1412 or the Bluebird BMS-127WV) use coreless motors with neodymium magnets and ironless rotors. These have lower inertia, faster response, and—crucially—better thermal performance because there’s no iron core to saturate. They also have a higher torque constant for the same size. A coreless micro servo can sustain 80% of its rated holding torque for 2 minutes without significant derating. That’s a game-changer.

Additionally, closed-loop micro servos (with magnetic encoders on the output shaft) are becoming affordable. They don’t increase holding torque, but they compensate for gear train wind-up. If the servo detects a 2-degree position error due to load, it applies more current to correct it. This effectively increases the apparent holding torque by 20-30% because the servo actively fights back. However, this comes at the cost of higher current draw and more heat. So the thermal problem remains.

Final Practical Guidance: Choosing Between Standard and Micro

Here’s a decision matrix for your next project:

• Choose a micro servo motor if: Your total weight budget is under 15 grams per joint, your holding time is under 5 seconds, your load is less than 50% of rated stall torque, and you can tolerate some backlash.
• Choose a standard servo if: Your application requires holding position for more than 10 seconds, your load is above 50% of the micro servo’s rated torque, you need precise position under variable load, or you’re working with a power supply that can’t handle repeated stall currents (micro servos will draw stall current continuously during a hold, which can brown out your BEC).
• Choose a mini servo if: You need the weight savings of a micro but the sustained holding torque of a standard. This is the sweet spot for most robotic arms and camera gimbals.

And one last rule: never use a micro servo motor as a direct replacement for a standard servo in an existing design. The mounting holes, spline size, and voltage range are different. But more importantly, the holding torque behavior is fundamentally different. A standard servo’s 10 kg·cm rating means it will hold 10 kg·cm all day. A micro servo’s 2 kg·cm rating means it will hold 2 kg·cm for about 3 seconds, then start to fade. Plan accordingly.

The next time you hear that little click of stripped gears, remember: it’s not the servo’s fault. It’s the physics of tiny motors, tiny gears, and the unforgiving math of holding torque. Choose your servo like you choose your tools—based on the sustained load, not the peak spec. Your robot arm (and your sanity) will thank you.