Somebody slowed the motor down and the motor got hot. That is the whole problem, and it surprises people because it runs against the instinct that a machine working less should run cooler.
The instinct is wrong for a specific mechanical reason, and once you see it you cannot unsee it.
The fan is on the shaft you just slowed down
A TEFC motor is totally enclosed, fan cooled. The fan is not a separate piece of equipment. It is bolted to the back of the motor shaft, inside a shroud, blowing air across the finned frame. It is the same shaft the rotor is on, which means it is the same shaft the drive is controlling.
Cut the motor to half speed and the fan turns at half speed. Take it to 10 Hz, one sixth of base speed, and the fan turns at one sixth of its design speed.
Airflow tracks speed. At one sixth speed you have roughly one sixth the cooling air. Nothing is broken, nothing is misconfigured, and no fault code will tell you about it. The motor is simply being cooled by a fan that you slowed down along with everything else.
The heat does not leave with the air
Here is the part that turns a curiosity into a failure.
A motor never stops making heat. Even with nothing coupled to the shaft, it draws magnetizing current to build and hold the rotating magnetic field, and on a typical induction motor that is somewhere around 30% of full load amps. That current flows through winding resistance and produces heat whether the motor is doing useful work or not.
Now put a load on it. On a constant torque load, the current at 20 Hz is the same current you see at 60 Hz, because the load needs the same torque either way. A loaded belt does not get lighter when it moves slower. A mixer blade does not meet less resistance in cold adhesive because you turned the speed down.
So you have full current making full heat, and a third of the air to carry it away.
This is why fans and pumps get a pass
The reason low-speed cooling is not a universal crisis is that most drives in the world are on centrifugal loads, and centrifugal loads unload themselves.
A centrifugal fan or pump follows the affinity laws. Torque falls with the square of speed and power falls with the cube. At 80% speed you are drawing about 51% of the power. At 50% speed you are drawing about 12.5%.
That is the whole trick. The heat generation falls off faster than the cooling does, so the motor stays in balance on the way down. A pump at half speed is producing a fraction of the heat it made at full speed, and the reduced airflow is enough to handle it.
Constant torque is the opposite case. Conveyors, mixers, extruders, hoists, feeders, positive displacement pumps, reciprocating compressors. The current stays flat and the cooling collapses, and the two curves separate as you go down.
If you want one sentence to carry out of this: the load type decides whether low speed is a cooling problem at all.
Why the number is 20 Hz and not 15
The general guideline for a TEFC motor on a constant torque load is to avoid sustained operation below about 20 Hz, which is a third of base speed, without independent cooling.
That number is not a physical constant. It is where the industry has settled after enough motors came back burned, and it moves with the application:
Some sources use 30 Hz, particularly on motors at or above 50 HP, where there is more heat to move and more mass to move it through.
Reciprocating compressors should not run below about 30 Hz regardless of the cooling question, because of lubrication and mechanical stress. That threshold has nothing to do with airflow and everything to do with oil.
Intermittent is not the same as sustained. A conveyor that jogs at 8 Hz for ten seconds during a changeover is not the case this rule is written for. Continuous running is.
The word doing the work in the guideline is sustained. The failure mode is thermal, and thermal failures need time.
Check the nameplate before you assume
Before you conclude a motor cannot run slow, look at what it says it can do.
Inverter-rated motors carry a constant torque speed range rating, written as a ratio. 20:1, 100:1, 1000:1. It tells you how far down that motor can run at full rated torque without a cooling problem.
On an 1800 RPM motor, 20:1 means 90 RPM, which is about 3 Hz. A 1000:1 rating means 1.8 RPM, which is essentially stopped and still making full torque. Motors are built to do this. They are just not the motor that came out of stores.
Two other things on that plate matter here. Enclosure type: TEFC is shaft-cooled and has this problem, TEBC is blower cooled and does not, TENV has no fan at all and lives on convection. And insulation class: B is good to 130 degrees Celsius, F to 155, H to 180. Every step up buys 25 degrees of margin. Class F is standard on modern inverter-rated motors, and it is a meaningful cushion over the Class B motor somebody pulled off the shelf.
A TEFC motor with Class B insulation, on an application that needs sustained running below 20 Hz, is a motor that will fail and a decision that was made before anyone touched a parameter.
What actually fixes it
In the order you should consider them.
Read the speed range rating first. If the motor is already rated for the turndown the application needs, there is nothing to fix and the answer took thirty seconds.
Blower cooling. A TEBC motor carries its own cooling fan driven by a separate small motor that runs at full speed no matter what the main motor is doing. This is the standard answer for sustained low-speed constant torque work, and it solves the problem at its source rather than working around it.
Oversize the frame. More thermal mass and more surface area means the motor runs cooler at every operating point. A 30 HP motor in a 40 HP frame generates less heat relative to its ability to shed it. This works when low-speed running is intermittent rather than the normal condition.
Forced ventilation. Ducted air to the motor from an independent source. Common on retrofits where changing the motor is not practical.
And one that does not fix it, which is worth stating plainly because it gets tried.
Sizing the drive bigger does not cool the motor. There is a low-speed factor in drive sizing, and applying it is correct, but understand what it buys. It protects the drive by giving it current margin during sustained low-speed operation. The motor’s problem is that no air is moving across it. More available current does not move air.
Two more things that get tried and should not be. Raising the motor overload trip setting does not cool anything, it just removes the alarm that was telling you the truth. And counting on the service factor does not work either: NEMA MG1 Part 31 says service factor capability is reduced or eliminated on PWM power. On a drive, a 1.15 service factor is not 15% of headroom. It is closer to zero.
The short version
A TEFC motor cools itself with a fan on its own shaft, so slowing the motor slows the cooling roughly in proportion. On a centrifugal load that is fine, because the heat falls off faster than the air does. On a constant torque load it is not, because the current never drops.
Below roughly 20 Hz on constant torque, plan for cooling that does not depend on shaft speed. Above it, and on any centrifugal load, you are inside what the motor was built to handle.
The decision is made when the motor is specified, not when the drive is programmed. By the time somebody is asking why the windings are hot, the expensive part of the answer is already installed.
Frequently Asked Questions
What is the minimum speed for a TEFC motor on a VFD?
For a constant torque load, the general guideline is to avoid sustained operation below about 20 Hz, which is a third of base speed, without independent cooling. Some sources use a more conservative 30 Hz, particularly on motors at or above 50 HP. Variable torque loads such as centrifugal fans and pumps are far less affected, because the heat they generate falls off faster than the cooling does.
Why does a motor overheat at low speed when it is not working hard?
Two reasons. The cooling fan is mounted on the motor shaft, so when the motor slows down the fan slows with it and airflow drops roughly in proportion to speed. And the motor never stops generating heat. Even with no mechanical load it draws magnetizing current, roughly 30% of full load amps, to maintain the rotating field. On a constant torque load the current does not drop at all while the cooling does.
Do fans and pumps have the same low-speed cooling problem?
No, and this is the part people miss. A centrifugal fan or pump follows the affinity laws: torque falls with the square of speed and power falls with the cube. At half speed a centrifugal pump needs about one eighth the power. Heat generation drops roughly in step with the lost cooling, so the motor stays in balance. A conveyor at half speed still needs the same torque, so the current stays up while the air goes away.
What is a constant torque speed range rating?
It is a ratio printed on inverter-rated motor nameplates, written as something like 20:1, 100:1, or 1000:1. It tells you how slowly that motor can run at full rated torque without overheating. On an 1800 RPM motor, a 20:1 rating means 90 RPM, roughly 3 Hz. A 1000:1 rating means 1.8 RPM. Check this before you assume a motor cannot run slow, because the answer may already be on the plate.
Will oversizing the drive fix a motor cooling problem?
No. Sizing a drive with extra margin for sustained low-speed operation protects the drive. It does nothing for the motor, because the motor's problem is airflow, not available current. If the motor runs slow under load continuously, the fix has to be at the motor: a blower-cooled frame, a larger frame, or a motor rated for the speed range the application actually needs.
Can I use the service factor as headroom for low-speed running?
No. NEMA MG1 Part 31 states plainly that service factor capability is reduced or eliminated when a motor runs on PWM power. A motor operating into its service factor on a drive has no margin left for PWM harmonic losses, reduced cooling, ambient temperature, or altitude. Treat service factor as reserve for short overloads, never as part of the continuous rating.