VFD vs Servo Drive: Which Does Your Application Need?

A VFD controls the speed of an induction motor and gets no feedback while it does it. A servo drive controls the position of a permanent magnet motor and never stops measuring where that motor is. That one architectural difference sets everything downstream, including the price.

Here is the part that surprises people. Most machines specified with servos did not need them. Measured against what the process actually tolerates, a large share of positioning and indexing jobs land around plus or minus one millimetre, and a closed-loop VFD with an encoder already holds that. The servo gets specified anyway, because the drawing from the last machine said servo.

You already know both technologies work. What you need is the number that separates them, and a clear picture of what each answer costs once installed.

This article gives you the tolerance thresholds that decide the question, the real cost of each option, the middle tier almost nobody prices, and the cases where servo is genuinely the only defensible answer. We build VFDs and we do not build servos, so read the servo sections knowing that. We have argued that side as hard as it deserves, because a comparison you cannot trust is worth nothing to you. If you would rather talk through your own duty point, talk to our engineers and bring the tolerance figure with you.

Key Takeaways

• A VFD controls speed open loop. A servo drive closes current, velocity, and position loops at once. That architecture, not the brand, sets the accuracy ceiling.
• Position tolerance decides the choice. Below plus or minus 0.1 mm you need servo. At plus or minus 1 mm a closed-loop VFD with an encoder holds the job.
• Between those sits the tier almost nobody prices: a closed-loop vector VFD with an encoder and a PLC position loop, covering roughly plus or minus 0.5 to 20 mm.
• A servo axis costs roughly six times a VFD axis on a documented 2.2 kW example, and servo wins outright on only three things: accuracy under 0.1 mm, torque at zero speed, and coordinated multi-axis motion.
• Servo drives run to 400 kW and beyond. The cost gap widens with power; servo does not stop scaling.

VFD vs Servo Drive: The Short Answer

A VFD varies the frequency and voltage fed to an induction motor, which changes its speed. It runs open loop by default, so it commands a speed without measuring whether it achieved it. A servo drive runs a permanent magnet motor with continuous feedback and closes three control loops at once, so it commands a position and holds it against load.

What you are comparing	VFD	Servo drive
What it controls	Speed	Position, velocity, and torque
Motor type	Three-phase induction motor	Permanent magnet synchronous motor
Feedback	None by default; optional encoder	Always. Encoder or resolver
Positioning accuracy	About plus or minus 0.5 to 1 mm with encoder	About plus or minus 0.01 mm on a 2.2 kW axis
Response time	100 to 500 ms	1 to 20 ms
Torque at zero speed	Poor without a brake	100 percent rated torque at 0 rpm
Overload capacity	About 150 percent	About 300 percent
Speed range	20:1 open loop, 1,000:1 vector	5,000:1 or greater
Relative cost	Baseline	Roughly 3 to 6 times, drive and motor
Typical power range	0.1 kW to tens of megawatts	40 W to about 5 kW, specialists past 400 kW

Read that table as a set of ceilings rather than a scorecard. The VFD column is not worse at its job, it is a different job. If your machine needs a fan at 70 percent speed, the servo column’s advantages are ones you would pay for and never use.

One clarification saves confusion later. A VFD vs servo motor question is not the same as a servo drive vs VFD question, because the drive and the motor are separate purchases. The accuracy figures above describe the two working together.

If you already know a drive is the answer and are choosing between drives, our guide to choosing a VFD covers the sizing side.

What Actually Separates Them

The gap is not the motor and it is not the power electronics. It is the control architecture, and specifically how many loops are closed.

A standard VFD is open loop. You give it a frequency reference, it produces that frequency, and it assumes the motor follows. That is adequate on a fan or pump, where the load is predictable and nobody cares whether the shaft sits at 1,482 rpm or 1,487. Adding an encoder closes the speed loop, taking speed regulation from roughly plus or minus 0.5 percent to about plus or minus 0.01 percent, and it is available on drives you may already own.

A servo drive closes three loops at once, each running inside the next. The current loop runs at 1 to 10 kHz and controls torque directly. The velocity loop runs at 100 to 1,000 Hz, converting position error into a speed command and damping the approach so the axis does not overshoot.

The position loop runs at 10 to 100 Hz, comparing commanded position against encoder feedback. A VFD on a fan closes none of them.

The reason a VFD cannot match that at standstill is that an induction motor has no magnets. Its rotor current is induced by the stator field, so the drive estimates rotor position rather than measuring it. Sensorless algorithms handle speed control well, but with no encoder and no magnets there is no reference to hold at zero speed.

That is a physics limit, not a tuning problem, and it is why every serious VFD positioning scheme starts with an encoder. We cover the control modes in more depth in our article on what a VFD does for motor control.

The Tolerance Number That Decides It

Find the position tolerance your process actually requires. Not the tolerance on the drawing from the last machine, the tolerance your product needs. That single number resolves the VFD vs servo drive question in most cases before cost enters the conversation.

There are four accuracy bands in practice, and the third is the one buyers miss.

• Open-loop VFD. Speed regulation of about plus or minus 0.5 to 5 percent, and no position control at all. Right for pumps, fans, and mixers.
• Closed-loop vector VFD. About plus or minus 0.01 percent speed regulation, and with built-in position firmware, positioning to roughly plus or minus 0.5 to 1 mm. A standard VFD with an encoder card and the right parameters.
• VFD with encoder and PLC position loop. Roughly plus or minus 0.5 to 20 mm. The VFD handles speed, the PLC handles position, so the loop runs at PLC scan rate. A field-proven example is a ten-ton overhead transporter running a 100 metre trajectory and holding plus or minus 5 to 20 mm.
• Servo. About plus or minus 0.01 mm on a 2.2 kW axis, holding to within one encoder count, which on a 20,000 pulse-per-revolution encoder is under 0.02 of a degree. The only band that holds position at zero speed against a load.

Tolerance your process requires	What holds it
No position requirement	Open-loop VFD
Plus or minus 5 to 20 mm	Closed-loop VFD plus PLC position loop
Plus or minus 0.5 to 1 mm	Closed-loop vector VFD plus encoder
Plus or minus 0.1 to 0.5 mm	Servo, or VFD with careful mechanical design
Below plus or minus 0.1 mm	Servo
Position held against load at zero speed	Servo

The closed loop vector vs servo decision is where most of the money sits, because that is the widest band and the one buyers most often skip past on the way to a servo quote.

The cutoff most engineers settle on is this. If your process tolerates plus or minus 1 mm, a closed-loop VFD with encoder feedback will do the job. If it demands plus or minus 0.1 mm, you need servo, and no amount of drive tuning will change that.

When a packaging line in Portugal came up for a controls refresh, the specification called for servo indexing on the sealing head. Beatriz, the controls engineer handling the project, measured what the line actually needed before ordering anything. The tolerance on seal position turned out to be plus or minus 2 mm, carried forward from a machine built a decade earlier when the sealing geometry was different.

She replaced four servo axes with closed-loop vector VFDs and a PLC position loop, and spent part of the savings on a higher-resolution encoder for the one axis that genuinely needed accuracy. The line has run since without a positioning complaint.

What Each Option Actually Costs

This is where the published numbers contradict each other, and it is worth understanding why before you trust any of them.

The servo drive cost vs VFD question has no single answer, because the published numbers contradict each other. Across the sources, the comparison ranges from “30 to 50 percent cheaper” all the way to “4 to 8 times more expensive” for a single 5 kW servo axis. That spread is not noise, it is a definitional failure.

Some sources compare drives only, some compare drives and motors, and some compare an installed system with cabling, encoders, and commissioning labour folded in. A servo amplifier alone runs 2 to 4 times a VFD of equivalent power, and a servo motor runs 3 to 6 times an induction motor. Reorder the same factors and you get the 4 to 8 times figure.

The Cheapest Way to Compare

The cleanest framework comes from a Rockwell-sourced rule of thumb reported by DPA Magazine: each tier costs roughly two to three times the one below it.

Open-loop VFD is the baseline, closed-loop vector is two to three times that, and servo is two to three times the closed-loop vector figure again. That stacking gives you a comparable multiple at any power level, which the flat hardware ratios cannot.

The Ownership Costs Nobody Quotes

Hardware is the visible half. The rest of the bill arrives over the following five years.

• Cable. A servo axis needs shielded power cable and a separate feedback cable, both often manufacturer-specified and both sold at a premium. A VFD needs one power cable.
• Tuning. Gains, filters, and settling behaviour all take commissioning time from someone who knows what they are doing.
• Spares. A servo motor at 3 to 6 times the cost of an induction motor changes what you keep on the shelf and what you are willing to write off.
• Technician dependency. Servo systems narrow the pool of people who can service your machine. That is an operational risk, not a training line item.

That last point is where a vendor’s own figure becomes useful. KEBA estimates about 1.5 days of basic training per person to move a team onto servo technology. Multiply that across a maintenance department, add the productivity lost while they learn, and allow for attrition, and you have a number that never appears on a quotation.

A Worked Example

The clearest published comparison covers a 2.2 kW CNC feed axis, priced in India.

Path	Components	Cost	Accuracy
VFD	Induction motor about 4,000 rupees, VFD about 6,000, encoder about 2,000	About 12,000 rupees	Plus or minus 0.5 to 1 mm
Servo	Servo motor with encoder about 35,000, servo drive about 40,000	About 75,000 rupees	Plus or minus 0.01 mm

That is roughly six to one on a small axis. Two caveats, stated plainly: prices vary enormously by region, and a single example does not set policy for your project.

The 75 kW Claim Is Wrong

You will read that servo stops being practical around 75 kW. That claim is incorrect. Lenze rates its 9400 servo range from above 75 kW to 420 kW for low torque-peak duties like extruders, and Baumüller’s b maXX 6500 family reaches 400 kW in a single unit, more by paralleling.

The accurate statement is that the cost per kilowatt gap widens sharply as power rises, which makes servo economically unattractive on large drives rather than technically impossible. If you want a configured drive priced against a servo alternative on your own duty point, send us your duty point.

Can a VFD Do Positioning, and Can You Run a Servo Motor on One?

Two questions come up constantly, and the honest answers are more conditional than the short versions suggest.

Yes, with a PLC Running the Loop

A closed-loop vector VFD with an encoder can reach a commanded position, and many drives include position control firmware for simple indexing. Accuracy of roughly plus or minus 0.5 to 1 mm is realistic, and it depends as much on the mechanics as on the drive. Backlash, belt stretch, and gearbox play will eat your tolerance budget before the drive does.

Most VFDs do have a limit. They have no internal position register and no motion profile generator, so they can follow a speed reference but cannot plan a trapezoidal move or coordinate motion across two axes. That work moves to a PLC, which calculates the profile and issues speed commands to the drive. It works well, and it is how the ten-ton transporter above achieves plus or minus 5 to 20 mm over 100 metres.

The catch is loop speed. A PLC position loop runs at the scan rate, typically 5 to 20 ms, against a servo’s internal position loop at 10 to 100 Hz. That gap of two to three orders of magnitude is why PLC-based VFD positioning suits a transporter and not a high-speed pick-and-place head.

Running a Servo Motor on a VFD

This comes up because a servo motor is often already on the shelf. It can work, but only under strict conditions, and most generic drives fail at it. A permanent magnet servo motor is not an induction motor, and a generic VFD expects to induce rotor current.

Connect it to a rotor that already has magnets and the behaviour degrades fast: jerky motion, vibration, overheating, repeated overcurrent faults, and in the worst case a loss of synchronism where the motor slips a pole and never recovers. Speed range collapses to roughly 10:1 against the 5,000:1 a real servo drive delivers.

To have any chance of working you need a drive explicitly rated for permanent magnet motors with field-oriented control, such as SEW Movidrive B, Control Techniques Unidrive, Yaskawa G7, Invertek P2, or INVT GD35 and GD350. You also need correct feedback wiring, voltage matching, since servo motors often run at 240 V rather than line voltage, and flux weakening above rated speed.

Above all, the magnetization routine must be disabled. A VFD’s commissioning sequence injects current to identify an induction motor’s magnetizing inductance, and applied to a permanent magnet rotor that current can permanently demagnetize the magnets. This is the step people skip.

Threads on Eng-Tips disagree about whether the practice is worth attempting at all, and many manufacturers simply answer no. If you are considering this path, request an application review before you buy anything.

Where Servo Is the Only Answer, and How Real Machines Use Both

If this section reads as a strong case for servo, that is deliberate. There are jobs where it is not a preference, it is the only thing that works.

Three Cases Where Servo Wins Outright

Accuracy below plus or minus 0.1 mm. No mechanical arrangement makes a VFD hold this reliably. If your process needs it, the discussion is over and the only remaining question is which servo.

Holding position at zero speed against a load. An induction motor without an encoder has no way to resist a force applied to a stopped shaft. A servo motor delivers 100 percent of rated torque at zero rpm. This matters on vertical axes, on anything holding against product weight, and wherever the machine stops mid-cycle under load.

Coordinated multi-axis motion. Electronic gearing, electronic camming, and interpolated paths across two or more axes require loops running in the same controller at the same rate. A PLC running VFDs cannot do this convincingly.

Why Real Machines Use Both

The binary framing is where most comparisons go wrong. Real machines are usually hybrid, and the decision is made per axis rather than per machine.

A CNC machining centre runs servos on X, Y, and Z because those determine part geometry, and a VFD on the spindle because the spindle needs speed control across a wide range. A packaging machine runs servos on film indexing and the sealing head, and VFDs on the infeed and outfeed conveyors. Same cabinet, same PLC, different answer per axis. Our overview of VFD applications in manufacturing covers more of these architectures.

What Overbuying Costs You

When a beverage plant in Ghana could not hit fill accuracy on a filler head, the controls team spent two years tuning. They changed control modes and acceleration ramps, and eventually replaced the drive.

None of it worked, because the problem was architectural. The filler head had to hold position at zero speed against the weight of product already in the container, and no VFD holds torque at standstill without a brake. A servo handled it in a week.

That is the failure in one direction. The failure in the other is quieter but more expensive across a fleet. Buying servo capacity you did not need costs you the hardware premium, the doubled cable, the tuning time, and a permanent dependency on a narrower skill set. Our products page covers the drive side of that calculation.

How to Decide on Your Own Machine

The question of when to use servo vs VFD resolves itself if you work through these in order. The first step settles most cases on its own.

1. Measure the actual tolerance. Not the drawing, the process. Ask what happens to the product if the axis is 1 mm off. If the answer is nothing, you have your answer.
2. Decide whether position must be held at zero speed. If yes, you need servo or a mechanical brake. This is binary.
3. Check whether axes must move together. Electronic gearing or camming across multiple axes points to servo. Independent axes do not.
4. Work out the response time the cycle needs. A VFD responds in 100 to 500 ms, a servo in 1 to 20 ms. If your cycle has slack, the VFD’s slower response costs you nothing.
5. Price the configured system, not the drive. Include the encoder, the shielded cables, the feedback wiring, the tuning time, and the spares holding cost.
6. Check who can service it. Ask your maintenance team honestly whether they can tune and repair the option you are considering.

A German machine builder inherited a machine with servos on eight axes. Before the next build, Anja worked through that list. Six of the eight ran at a single fixed speed with a tolerance of plus or minus 3 mm, comfortably inside what a closed-loop VFD holds.

She left the existing machine alone, because retrofitting working hardware is rarely the cheaper answer, and specified VFDs on those six axes for the next build. The drive cost of that build fell substantially and nothing about the machine’s performance changed.

If you would rather not work the list alone, our engineers will go through it against your actual duty point. Talk to our engineers with your tolerance figure and cycle time, or read our full VFD selection guide for the sizing side.

Frequently Asked Questions

What is the difference between a VFD and a servo drive?

A VFD varies the frequency and voltage supplied to an induction motor to control its speed, and it runs open loop by default. A servo drive controls a permanent magnet motor with continuous position feedback, closing current, velocity, and position loops simultaneously. The VFD controls how fast a shaft turns. The servo controls where the shaft is and holds it there against load.

Can a VFD do positioning?

Yes, within limits. A closed-loop vector VFD with an encoder can position to roughly plus or minus 0.5 to 1 mm, and many drives include firmware for simple indexing. For more complex motion a PLC runs the position loop. The constraint is speed: a PLC loop runs at 5 to 20 ms, against a servo’s internal position loop at 10 to 100 Hz.

Can you run a servo motor on a VFD?

Sometimes, but the conditions are strict. The drive must be rated for permanent magnet motors with field-oriented control, correct feedback must be wired, the magnetization routine must be disabled or the magnets can be permanently demagnetized, and the drive must match the servo’s voltage. Generic VFDs driving PM motors typically produce jerky motion, overcurrent faults, and a speed range of about 10:1 against a servo drive’s 5,000:1.

How accurate is a VFD with encoder feedback?

Roughly plus or minus 0.5 to 1 mm for positioning, and about plus or minus 0.01 percent for speed regulation, against closer to plus or minus 0.5 percent open loop. Positioning accuracy depends as heavily on the mechanics as on the drive, because backlash, belt stretch, and gearbox play consume tolerance before the drive does.

Is a servo drive worth the extra cost?

It depends entirely on the tolerance. If your process needs below plus or minus 0.1 mm, must hold position against load at zero speed, or requires coordinated multi-axis motion, servo is the only option and the cost is simply the cost. If your process tolerates plus or minus 1 mm or more, a closed-loop VFD with encoder feedback does the job for a fraction of the price.

Do servo drives work above 75 kW?

Yes. The claim that servo stops being practical above 75 kW is incorrect. Lenze rates its 9400 range to 420 kW, Baumüller’s b maXX 6500 reaches 400 kW in a single unit, and Rockwell’s Kinetix 5700 spans 1.6 kW to 112 kW. What is true is that the cost per kilowatt gap widens as power rises, which makes servo economically unattractive on large drives rather than technically impossible.

Conclusion

The choice between a VFD and a servo drive comes down to one number, and it is not the price. It is the position tolerance your process actually requires.

Below plus or minus 0.1 mm, servo is the only answer. At plus or minus 1 mm, a closed-loop VFD with encoder feedback holds the job. In between, a closed-loop vector VFD with a PLC position loop covers more ground than the market gives it credit for.

Two things are worth carrying away. The cost multiples you will read range from 30 percent to 8 times because nobody defines what they are comparing, so define yours before you quote it. And the hardware multiple is the optimistic case, because shielded cabling, tuning, spares, and technician dependency never appear on a quotation.

Start with the tolerance. If it points to a drive we build, our products span 0.1 kW to 53,000 kW with vector control and up to 150 percent overload capacity. If it points to servo, we would rather tell you now than sell you the wrong drive. Either way, talk to our engineers and we will work the numbers against your real duty point.