VSD for squirrel cage motor retrofit


Thread Starter

Loren Schreiber

I have been following the VSD for wound-rotor induction motors with interest. What are the pitfalls of retro-fitting small squirrel cage
induction motors (2 HP , 460 volt, 3 phase) with VSDs? These are Baldor motors of 1990 vintage, currently controlled by simple reversing contactors.
Loren Schreiber
Systems Integrator, MCued http://www.mcued.com/
It just works with the following caveats:
Beware of the brake, if there is one. It must be externally controlled because as the drive ramps up, it changes both frequency and voltage and the brake won't release immediately, not something you want to have happen.

Don't run it too slowly so that the fan isn't cooling the motor. An inverter duty motor has enough mass and cooling fins so that it doesn't need a fan to stay within temperature specs. The small Baldors probably are TEFC class and need to spin at a minimum rate to stay cool.

Lastly, startup torque on a basic induction motor is essentially 6X the full load rating. When you put a VFD rated at 2HP you will get around 1.5X the rating to start it up, this can cause problems. The newer style vector drives do better (I don't remember the actual number) but they don't make 6X.
Loren, some possible problems which may result from operating a non-inverter duty rated induction motor from a VSD are as follows;

1/ Motor stator winding insulation may not be capable of withstanding the repetitive, fast-rising, peak-pulse voltages generated by VDS
transistor switching.
2/ Switching may generate shaft-to-ground currents which can damage the motor bearings.
3/ Switching can result in high frequency conducted and radiated RFI.
This may cause other systems operating on the same supply circuit or close by to misoperate.
4/ Operation a VSD without a properly rated input filter can result in the injection of power frequency harmonics back into the supply and
effect the supply transformer in specific and the supply in general.
5/ VSD operation may generate significant audio noise.

Best regards,

P. Talas, PEng.

I believe you have torque and current mixed-up concerning start-up. Depending on Nema class, line-operated start-up torque is in the 150-200% range - which is also why VSDs designed to same range. Start-up currents however are in the 6X range you state. Correct me if I'm wrong. Good point about any brake.
If long motor cable with IGBT drive, you may need protection (usually drive output filter) against motor insulation/bearing failure due to the high
motor voltage spikes phenomenon (reflected waves). If plan to run below 50% base speed and constant torque, may need extra motor cooling. Also, may need [extra] dynamic braking depending on inertia and decel time. Consult with VSD supplier.
Check the motors to see if they are rated for ISR (inverter spike resistant) duty. If so, no problem, if not, keep the motor leads between the drive and the motor short and twisted and set the drive up with the lowest switching frequency setting (typically 3kHz). You should not experience any problems.

Insulation breakdown and resulting instantaneous overcurrent trips are typical failure modes for this type of retrofit with substandard motors.
We have several hundred baldor motors 3/4HP - 100HP in operation with and without inverter duty ratings and have had no failures due to
insulation breakdown.

Ken Brown
Applied Motion Systems, Inc.
I hope this isn't duplicated. It didn't seem to go through the first time.

You and I are both right, the current of a stalled induction motor is 6 times its full load rating. But, current and torque are directly related (as long as voltage and frequency are held constant)...twice the current means twice the torque and six times the current means six times the torque. Therefore current and torque can essentially be used interchangably.

As and aside, note that torque and horsepower are related as long as the rotor is turning. If the motor is stalled, it is producing lots of torque (and heat - I-squared-R losses) but no horsepower.

The rating on a NEMA motor has to do with the maximum torque required to allow the motor to reach normal operating speed and not overheat trying to get there.

The nifty thing about and induction motor is that when it starts up, at time zero, it has essentially infinite (really 6X) torque (or current) which will break loose the static friction of most mechanical devices, especially inexpensive bearings. It also allows the motor to start up almost instantly. Once the motor starts to spin, torque does drop off and the ratings start to have meaning.

A VSD rated the same as a motor will take almost a second to come up to speed when driving a conveyor. Without the VSD, the conveyor will start almost instantly (it will probably start the conveyor faster than whatever is sitting on it will move). I fought this battle once trying to induct mail onto a tilt tray sorter using a VSD - so I always bring it up.

For larger inductions motors, a soft start is required so that at startup it doesn't bend shafts and break things.

Allen-Bradley used to have a great tutorial on motors in their Drives Catalog. It went through all the various types, starting with DC and went through synchronous, induction, servos, and I think it even covered steppers. I am not sure which Rockwell division catalog that would be in now. I think I have seen similar tutorials in Baldor and Siemens literature also.

If it would be helpful, I can paraphrase a tutorial but I need to go and look up all the equations and constants (I haven't used them in a while).

> Tom,


> I believe you have torque and current mixed-up concerning start-up. Depending on Nema class, line-operated start-up torque is in the 150-200% range - which is also why VSDs designed to same range. Start-up currents however are in the 6X range you state. Correct me if I'm wrong. Good point about any brake.

Daniel Boudreault

Hello Loren,

Your motor is relatively new.
Why are you thinking about the retrofit?
If it ain't broke don't fix it...

But if you insist, I do it all the time.
Just watch out for those long motor leads ( i.e. reflective waves )
Inverter duty motors have insulation breakdown of 50 to 100% of regular duty motors.

I would use a Baldor drive, since you already have a Baldor motor.
I never had any trouble with Baldor motors. They are of excellent quality. Their drives are great too.

What is the motor load now? The less load, the less current, the less heating...the less chance of motor failure when you switch over to the VFD. Other wise you will need external cooling fan.

I always put non-inverter duty motor on Variable Torque setting on the VFD if the motor is running beyond +/- 10% of its rated RPM, and clip the extremes at +/- 20% of motor rating.

Filter Filter Filter... use them.
Input Line filter...(protects other equipment from noise, and drive from transients)
Output Load filter...(protects motor from reflective wave)

Remember that if you put a contactor on the output side of the drive (between drive and motor) it can damage the drive outputs if opened
while the motor is running.

Good Luck, Dan B.

Bruce Durdle

If you have a look at the current/speed and torque/speed curves for a real induction motor, you will see that the current and torque are
NOT directly related. The current curve starts at about 6-8 x full load current at zero speed, decreases slightly to about 80% speed, then falls off rapidly to zero at synch sped. The torque curve will start at about 1.8-2 x full load torque at zero speed, rise to the "pull-out
torque" level of about 3-5 x full-load torque at 80-90% speed, then again fall to zero at synchronous speed.

For a full set of motor data sheets, and some good motor tutorial stuff, see www.westernelectric.com.au.

I think the explanation for the decreasing power factor as the number of poles is increased is that the effective amount of iron as a percentage of the total periphery has to fall, and the leakage flux therefore increases. This effectively increases the inductance of the
winding relative to the resistance.

Sorry, I have to disagree with many of your statements. Look at any T/S curve of a line operated induction motor. Torque is definitely not
proportional to current under high slip conditions (start-up). In fact, Torque proportional to current can only be possible at low slip. Torque proportional to current can be true for DC motors etc. [under set conditions] but not induction motors.

Also, your comment about torque and horsepower "related" in confusing. Typical industry terminology for T/S (power) curves is: constant
torque region (increasing power region); and field weakening region (constant power region).

BTW: I have been working in motor-drives engineering, for manufacturers, now for 16 years. 10 of those include asynchronous (induction) technology. I'd like to think I got some things right.

A good handbook covering VSD-motor basics is Facts Worth Knowing About Frequency Converters published by Danfoss (a VSD manufacturer).

Al Pawlowski, PE

My understanding is that item 1 is correct, but items 3-5 below would be a concern irregardless of whether the driven motor was inverter rated or

Also I have never heard about item 2 before. Can anybody elaborate? How does it occur? When is it likely? Who has seen it? If you get a bad bearing, how would you identify VSD caused currents as the cause?
It has been a while since I looked at the curves of a motor. You are correct about it being 200% torque at locked rotor. But...

Take a VSD that has a maximum source current 2X the full load specification of a motor. While the motor is starting, the current is limited to this 2X value, which means that the voltage supplied to the motor is a fraction of the 480 that it normally gets but the frequency is still 60Hz. The net result is you get about one third the starting torque that the motor is capable of delivering. This can be a problem depending on the application.

One approach is to oversize the VSD so that the full 6x current can be supplied.

The last time I bought a VSD, they had just started using a technique that attempts to run the motor (assuming it isn't at locked rotor) at the maximum torque possible for the current available by varying the frequency as well as the voltage. This approach provides impressive resilts compared to just commanding full speed and current limiting the motor. It still doesn't provide the same motor characteristics as a motor connected directly to 480v through a motor starter.

Running the motor as just described puts it into the linear range of current and torque which is how the VSD tries to run the motor. A good VSD can maintain this setting from about 1 RPM up to rated speed.

Sorry about getting confused.
I realize I said it wrong. I forgot a couple of things.

All I was really trying to say is if you take away the ability of the motor to get its 6x current, it won't start the same. The problem is that many motor experts won't tell you that.

Most of my experience with this is viewing the motor with the VSD attached and comparing it to a motor connected directly to the 480v.

I forgot about the the locked rotor torque only being approximately 200% full specified torque at locked rotor.

With a VSD sized the same as the motor with its 150-200% reserve current, only provides a third the starting torque at locked rotor.

Is this correct?

Also a vector type VSD figures out what the rotor is doing and runs the motor in the linear range of current and torque (the last 20% or so of the curve) by contolling both frequency and voltage. This works after the rotor is turning, while it is stationary, I think my first statement still applies.

Do you agree?

I learned about this the hard way. I started with an induction belt which had a 2hp induction motor running it. It had to launch a 5 to 50 pound parcel onto a tilt tray sorter (it had to hit a target in other words). It worked fine. Then someone got the great idea that we could vary the speed using a VSD to take into account different situations (the sorter was moving at different speeds and the packages were different lengths and I don't remember what else I was trying to optimize). I stuck a 2hp VSD on it and watched it consistently fail - the performance just wasn't there. I could get it to work with specific weights but it wasn't consistent and it couldn't launch anything over 30 pounds reliably. I put a 5HP VSD on it and it worked fine.

I just don't want someone to make a similar mistake. I was working with the reps from the VSD company and I did what they told me to do. When I learned more (I took a Siemens class on VFD fundamentals) and asked the question in a different way, I got the 5HP VSD answer.

Where I get into trouble now is that I view the induction motor characteristics through the intelegence of the VSD which confuses the issue considerably.

Michael Griffin

>Also I have never heard about item 2 before. Can anybody elaborate? How
>does it occur? When is it likely? Who has seen it? If you get a bad
>bearing, how would you identify VSD caused currents as the cause?
>2/ Switching may generate shaft-to-ground currents which can damage the
>motor bearings.
There have been articles on this subject in IEEE "Transactions on Industry Applications" (a magazine which is generally a good cure for
insomnia). You can search back issues of this publication for more information (I didn't keep a copy of this one).
If I recall correctly though, bearing currents are believed to be caused by capacitive coupling of the high frequency component to ground. This interacts with the bearing lubrication and various other factors in ways which are not yet fully understood (at least not at the last time I read about it).
The end effect to to produce pitting in the bearing which under causual examination appears to be due to physical contamination (sometimes even resembling corrosion). Microscopic study by people who know a lot more
about these things than I do reveals the damage is actually caused by a process akin to EDM.
If what I have read about it is correct, the frequency of this problem may be difficult to say, as it is often misdiagnosed.

Michael Griffin
London, Ont. Canada
[email protected]
You are correct. Today's IGBT drives generate even higher dV/dt wavefronts then last generation BJT types. In a motor, there is capacitance between phases and between phases and ground. Ground typically completes a path back to the drive input source. Any air space/material has capacitance and I = C dV/dt applies.

I also agree w/IEEE periodicals reading. My impression is these authors (typically University Ph.D.s) sit down and say - how can I explain my work in the MOST mathematically complicated/confusing ways. I guess this is part of the scientific publish-or-perish world's mindset?
You're getting close but missing some key fundamentals. NO. you don't need 6x current to generate the same starting torque! One beautiful thing about a VSD controlling the induction motor is that you can program an accel rate to match the motors off-line starting torque based on the inertia etc. In effect, the VSD generates waves
of V/Hz curves starting typically at 2.5-4 Hz. This means the slip is kept low and now the motor generates "same" starting torque with only 1.5-2x current instead of 6x current! This is why a VSD is inherently a soft-start as well. Taking it further, servo applications that require 3-4x rated torque can be achieved with high performance VSD (vector)sized to handle breakdown torque current of an optimized induction servomotor.

Regarding Vector Control, all types of VC - vector control, magnetizing current vector control, flux vector control, natural field
orientation, direct torque control, FAM field acceleration method etc. are all based on two key performance factors: in real-time (transient, steady-state conditions), keep the magnetizing current constant and equal to motor rating; and keep the slip low. High slip is bad. This causes high rotor impedance/losses and now the
magnetizing and torque producing current vector instead of "naturally" being about 90 degrees, for optimum torque generation is worse. Also, unlike a V/Hz drive, a closed-loop vector drive can produce torque at zero speed - a prerequisite for servo motion control.

You and I are talking about the same things, only I think we have slightly different perspectives.

Let me explain my perspective.

My first encounter with a VFD/VSD wasn't fun. What I needed was to approach the breakdown torque of the motor, but I didn't know that when I started and my expert motor advisors never hinted of the possibility. Also the small vector drives were really coming into their own a little over 5 years ago and the sales engineers were too caught up in their features I think.

My application was a loader onto a tray sorter. The way the loader works is a parcel (tray of mail in my case) would move to the head end of the loader and stop, waiting for the next available tray. When an empty tray would come along, the loader started back up and scooted the mail onto the sorter.

Using a 2hp motor driven directly off of a contactor, it worked but it didn't provide much control and the acceleration was too high (the typical clutch/brake approach made it worst). So I tried a VSD, and was told to size it based on the motor. It wouldn't work reliably, for light packages it worked pretty well, for heavier ones it consistently missed. It was also generally tempermental, medium weight parcels worked sometimes. What made matters worse was I was caught between the motor experts saying "it should just work" and my management asking "well, why haven't you made it work".

I was finally "sent" to a VFD class to learn the error of my ways. It became obvious about half way through the first day that I needed larger semiconductors (at that time I am much happier talking FETS than motors anyway).

The final solution was to oversize the VSD and everything got much better. Basically it allowed me to configure the system so that in about 1 sec the belt was up to speed, regardless of the weight of the parcel on it. In essence, I needed constant acceleration, not constant torque.

So now you know why, when someone asks the unqualified question: can I use a VSD with an induction motor, I will respond: probably, but there are other factors that may need to be taken into account. The motor behaves differently with a VSD than without it and there are situations where sizing the VSD is exactly cookbook.

Based on all this information, you could take the performance argument a couple of steps further. When you really have good control (flexible) over the commutation algorithm in an AC Induction
drive/motor combination, you can exploit the NEMA B motor design characteristics. (increasing rotor field strength without saturating the iron)

One method of achieving this is to actually boost magnetizing current on demand at low speeds (back EMF is not a concern as frequency is low . . . so bumping up V/Hz ratio has no ill effect) to
actually enhance torque performance at locked rotor and low speed operation. Using this technique, you can get much better torque response than starting the motor across the line.

It is also important to note that you are transferring power at a controlled lower applied voltage at the drive at a very high current. The converter bridge is still being fed by a 460V line and current at this feed is based on power consumed at the motor (less any losses due to efficiency effects). . . as such, it is possible to put 50Amps of current into a motor at low speed (say 6Hz -> 6 x 460/60 ~ 46 Volts) and only draw about 6Amps on the 460 Volt line. (I used 1/10th synchronous frequency to keep the math easy)

Hmmmm. . . . better torque performance . . . less slip . . . .less rotor heating . . . less current on the incoming line . . . . Must be magic!!!

PS . . . another neat trick is to apply a variation of this technique to suppress back EMF on PMBM's to go 200% faster than typically
possible . . . even for linear motors - your torque/force goes in the toilet . . . .but who cares if you are doing a rapid move without
regard to accuracy . . . just blend back into non-back EMF suppression mode as you drop below the synchronous speed of the motor. There is no end to the fun you can have with this stuff!

Ken Brown
Applied Motion Systems, Inc.
>My first encounter with a VFD/VSD wasn't fun. What I needed was
>to approach the breakdown torque of the motor, but I didn't know
>that when I started and my expert motor advisors never hinted of
>the possibility.

>Using a 2hp motor driven directly off of a contactor, it worked but it
>didn't provide much control and the acceleration was too high (the typical
>clutch/brake approach made it worst). So I tried a VSD, and was told to si
>ze it based on the motor. It wouldn't work reliably, for light
>packages it worked pretty well, for heavier ones it consistently

This is a good point that Tom makes about using VFD's for applications needing high acceleration rates. A line started motor will draw 600-800%
current to accelerate quickly. If you want to optimize acceleration with a VFD, you need to make sure the drive that you select can put out 600-800% current for short time periods. If the drive doesn't support these types of peak currents, you may need to oversize the drive. If you select an oversized drive, you be careful how you set continuous current limits so that you do not "smoke" the motor if it stalls or becomes overloaded.

Servo drives usually put out 200-400% current for a short time (< 1 sec.) This helps them accelerate quickly.

Bill Sturm
Reference my earlier e-mails re. this thread. With a VSD, you can achieve the motor's starting torque (e.g. 200% typical) with only approx. 200%
current because a VSD can control low slip unlike line starting. Likewise, you can achieve "breakdown" torque (e.g. 300% typical) with only approx. 300% current. During line operated starting (accel), the motor torque follows a trajectory from starting torque to breakdown torque to steady-state torque depending on slip. If the highest torque point [breakdown] can be achieved with 300% VSD current, why do you need 600-800% VSD current? If you do, it's not an apples-apples application conversion.