Harmonic-Drive vs. Industrial Linear Actuator: A Buyer's Total Cost Comparison

The comparison that keeps showing up on my desk

I'm the office administrator for a 31-person automation company. I manage all motion-component ordering—roughly $750,000 annually across 9 vendors. My job is to take an engineer's sketch and turn it into a purchase order without breaking the budget or the relationship with finance.

The question I hear most often is not 'which motor has more torque?' It is 'why does this simple thing cost so much?' That question is what started me comparing motion systems seriously.

The comparison that comes up again and again is a rotary harmonic drive reducer with a servo motor and a screw versus an integrated industrial linear actuator, or a linear stepper motor stage. They are not interchangeable, but they compete for the same purchase order. If you are buying either one, you need a framework, not a price sheet.

Cheap to buy is not cheap to own. The $500 quote can turn into $800 after shipping, setup, and revision fees. I've watched that story repeat for too many years.

What I compare before opening vendor folders

I compare four things: mechanical complexity, efficiency and heat, precision and backlash, failure modes. Then I add the only number that matters in the end: total cost of ownership.

Option A is a custom build. A harmonic-drive reducer mounts to a rotary servo and couples to a ball screw or lead screw. It gives a lot of torque in a small package, but it is a system, not a single component.

Option B is an off-the-shelf industrial linear actuator. That can mean an electric cylinder with a screw and a thrust bearing, or it can mean a direct-drive linear stepper motor with no screw at all. Both are sold as ready-to-mount motion modules.

The mistake I see most often is comparing the unit price of a gearbox alone to the unit price of a complete actuator. That is an apples-to-oranges PO.

1. Mechanical complexity: more parts means more invoices

Harmonic-drive technology wins on torque density. A 50:1 harmonic drive can be smaller and lighter than a planetary gearbox with the same ratio. But a harmonic drive is not a linear actuator. It needs support bearings, a coupling, a screw, a housing, and perhaps a thrust bearing. Each of those parts creates another line item and another possible mismatch.

An industrial linear actuator usually has the motor, the screw, and the thrust bearing in one sealed body. That simplifies ordering. A linear stepper motor goes even further: no screw, no thrust bearing, no coupling. The forcer rides on a magnetic platen. In a short-stroke machine, this is often the cleanest mechanical layout.

This means the simplest solution is usually the linear stepper motor, not the harmonic drive. That surprises people who read about harmonic drives first. But simplicity has a price, and the price shows up in installation skill.

2. Efficiency and heat: What is typical harmonic-drive efficiency?

At least once a month, someone at my desk searches for 'harmonic drive efficiency typical' values. The plain-language answer is this: typical harmonic-drive efficiency is about 70-85% at rated torque, depending on ratio, speed, and lubrication. For a 100:1 grease-lubricated unit, the published curves from Harmonic Drive SE usually show an efficiency closer to 75% than 85%.

Friction creates heat, and heat has to go somewhere. In a rotary harmonic drive, that heat sits in the gearbox housing, usually away from the encoder. It is manageable.

A linear stepper motor has no gear mesh, so there is no gear efficiency loss. But the coil sits close to the position it is trying to hold. If the motor runs at high holding current all day, the heat goes into the machine. Thermal expansion can move the zero position more than the motor's repeatability can control.

The surprising conclusion: the harmonic drive loses more energy as friction, but often loses it in a better place. For a linear stepper motor, heat is the hidden precision killer.

3. Backlash and repeatability: the gearbox is not the axis

An engineer once printed a page from Harmonic Drive SE and said, 'this gearbox has zero backlash.' He was not wrong. The flexspline and circular spline engage with near-zero clearance. That is the big selling point for a rotary load, especially when the load reverses direction.

On a linear axis, the question is whether the ball screw, coupling, and thrust bearing also have zero backlash. They usually do not. A preloaded ball screw helps, but the coupling can still wind up under torque. The assembled system is tighter than a normal gearbox, but not as tight as the gearbox brochure suggests.

A linear stepper motor does not have that problem. Without a screw or gear stage, there is no backlash in the force path. For short, stiff stages, a linear stepper is usually more repeatable in the direction of travel. That conclusion still makes some design engineers uncomfortable, but it matches what we see on our granite inspection table.

The same engineer and I were using the same words but meaning different things. I said, 'the coupling will add backlash.' He heard, 'you are trying to complicate my design.' The difference matters when the axis starts reversing direction.

4. What causes thrust bearing failure? Lessons from our service log

I'll admit, the first time a customer asked me what causes thrust bearing failure, I had to ask a technician. The short answer is axial overload, contamination, and misalignment.

• Axial overload: an actuator is rated for dynamic thrust, but a sudden stop can create shock loads several times higher than the catalog rating.
• Contamination: in a shop with dust or metal chips, a poorly sealed bearing race gets scratched and starts making noise.
• Misalignment: a motor mounting face that is not square to the screw puts a side load on the bearing. It can make the bearing fail long before the motor.

We did not have a formal load checklist at first. The third time a thrust bearing failed, I finally created one. That is the kind of process gap that only shows up after a service call.

This is why I do not automatically buy the smallest industrial linear actuator that fits the thrust spec. I ask what happens when the carriage hits an end stop. If the answer is 'the screw stops but the carriage keeps pushing,' the thrust bearing is about to die.

A linear stepper motor removes the thrust bearing entirely because there is no screw. That is a real advantage in a short-stroke machine. But the air gap becomes the next thing that can kill performance. You trade one mechanical failure mode for a setup tolerance.

Harmonic drive systems can also fail from torque spikes above the catalog rating. The flexspline is a fatigue part, not a maintenance-free black box. Plan for it.

5. Total cost of ownership: the part that changes decisions

Here is where my buyer hat goes on. In Q3 2024, we quoted two ways to make a 200 mm vertical axis with 1,500 N thrust and 0.01 mm repeatability.

The first quote was $1,380 for a harmonic-drive reducer and a servo coupling. The ball screw, thrust bearing, housing, and alignment labor were not included. When everything got added, the total was $2,760 before counting engineering hours.

The second quote was $2,450 for a complete industrial linear actuator with motor, screw, thrust bearing, and a sealed housing. The purchase order went to one vendor, the invoice was clean, and the lead time was shorter.

The linear stepper motor option came in at $2,150 for a short-coil table assembly with a magnetic platen. It would have worked for the stroke, but the operator needed access to the platen for cleaning, so we chose the actuator.

Which one is cheaper? The second one. The first one looked cheaper for about three weeks. Then the real cost showed up.

So which one should you order?

Use a harmonic-drive system when you already have a rotary servo in the design, when the axis is rotary, or when you need a very high reduction ratio in a compact envelope. The gearbox is excellent. Just budget for the rest of the system.

Use an industrial linear actuator when you need a full linear axis with screw and bearings in one assembly, and when you want to hold one vendor accountable for one device.

Use a linear stepper motor when the stroke is short, there is no heavy shock load, and repeatability matters more than force. The direct drive removes the thrust bearing, the coupling, and the screw from the failure list.

This advice comes from our world: low to medium duty cycle, optics and semiconductor tooling, controlled shop environment. If you are building a 24/7 packaging line or a high-speed press, the calculus changes. I can only speak to what appears on our purchase orders.

Compare purchase order totals, not component prices. Ask where the thrust bearing is, what happens at a hard stop, and how the manufacturer handles heat. The rest is just vendor preference.