View Full Version : VFD,air clutch and enertia?

09-18-2006, 07:27 PM
I am building a machine to pull rubberized fabric from the out feed side of a commercial sewing machine.The drive for this machine is a 1/2hp 1750 rpm 3~ motor and a 60:1 right angle gear reducer seperated by an air clutch.I'm using a VFD to vary the speed of the motor and the clutch allows for the feed to be dropped in and out without the VFD being shutdown and restarted each time.

The VFD shows 1.1amps with the clutch out and 1.4amps with the clutch engaged.Even at 2x's full load on the puller the amp reading never climbs above 1.7.Motor and drive are rated at 2.5amps@230vac.

Here is the problem,above 45hz when the clutch drops in the drive ,looses rpm and lags,at 60hz every second or third time it trips the over current limit.Setting the display to readout current draw when this event happens the current spikes at 4.3amps @ the 60hz setting.

Some of the DC drives I have seen used a flywheel on the motor shaft for just that reason to compensate for the effect.

So what I am thinking is I have three options-
1 Experiment with a flywheel
2 Get a larger drive/motor
3 Change out the wormgear reducer for a planetary reducer and eliminate friction.

I have tried resetting some of the drives parameters with no real improvement.I was also thinking of putting an orifice in the air line to the clutch to slow the engagement,but I am thinking that might make things worse.

Any clues?

09-18-2006, 09:11 PM
Big Capacitor?

Opps, AC motor. Never mind!

09-18-2006, 09:26 PM
Use a bigger reduction. A dinky 3ph motor like that can probably run over 100 hz no problem.

09-19-2006, 04:44 PM
Since the motor is well below full load steady state it sounds like the HP is adequate. Since the current goes well above FL when the clutch engages this implies that the problem is in accelerating a large inertia so the immediate question is: "Do you really need that fast an acceleration?" If not then perhaps your idea of an orifice in the air line might work although a lot of the slowdown would probably come before the clutch engagement so more time would be wasted. Another approach in this line would be to try reducing the air pressure feed to the clutch. This way the clutch would slip more after engaging while it accelerates the inertia at a lower acceleration rate and this would present a lower torque to the motor thus decreasing the current required.
If you actually need the fast acceleration then a flywheel would probably be useful. If the flywheel inertia is large compared to the motor inertia you might have to lengthen the speed ramp in the motor drive. On any I've set up this is simply a software setting.
Another approach would be to look through the software settings and see if you can add a speed droop function. This allows the speed to fall off with current so the acceleration would be lower. Somewhat in line with this is to look for a current limit setting and if possible set it to down around FL current or less for the motor. This would prevent tripping and would allow the speed to sag for a bit after clutch engagement.

I don't see that setting the hertz to 100 would be much help. Doing this simply runs the speed up above base speed and the rules here are pretty much the same as for a dc drive in the field weakened range-- the hp cannot go above the rated value so the allowable torque must be decreased. Normally this is factored into the software and there's not much you can do about it.

My experience was mostly on larger drives so this little one may be less flexible in settings


09-19-2006, 05:59 PM
By using a higher RPM you can use greater torque multiplication even if the horsepower doesn't increase. Since it isn't near using full hp the increased torque should help.

09-19-2006, 09:32 PM
As is: motor speed=1750 rpm hp=.5
speed after 60:1 gear reducer = 1750/60=29.167 rpm

Tavailable=(5250*.5)/1750=1.5 lb ft
after 60:1 gear reducer T=90 lbft

Now run the motor at 100hz giving n=(100/60)*1750=2916.67 rpm

Tavailable=(5250*.5)/2916.67=0.9 lbft
Now put in a new gear reducer to give 29.167 rpm from 2916.67 motor speed
GR=2916.167/29.167 = 100:1
Now the torque available at the gear output = .9*100=90lbft

so we are back where we started.

Are you depending on some transient effect which happens when the clutch closes? If so I can't see it at the moment.


09-19-2006, 09:57 PM
Off the top of my head if you double the rpm of the motor there should be about 4 times the stored energy in the rotor. The rotor is a pretty significant flywheel.

A.K. Boomer
09-19-2006, 10:13 PM
I like your bigger flywheel idea wierd, very dependable, and minimum weight if you have the room for it, just a light flywheel that is much larger than the rotor, with all the weight on the outer parimeter of the flywheel.

09-19-2006, 10:46 PM
Well I played with it some more today.I took the fan and shroud off the motor and popped on a heavy cast iron pulley I had on hand to act as a flywheel.It was roughly I think the same mass as the rotor and as large a diameter as the motor case.It did nothing,the drive fell flat on it's face just like before with no improvement.

The next experiment was redcuing the air flow to the clutch,this worked to a degree,but still no dice.I was able to pull off a start at 50hz with no problem,but even at 51hz it reverted to it's bad habits.Now,I also tried engaging the clutch at low pressure,this worked flawlessly,but the delay between stop and engage was a bit long.

The next test was to remove the drive from the system and cycle the clutch with the motor running full speed directly off shop power with an amp meter in place.I got the same readings as with the drive in circuit,less the spike.The amp spike was no more than 2.0amps during engagement.This tells me the drive is stumbling and can't compensate fast enough to the clutch being engaged.

I tried going over the parameters once again.It can be set either for constant speed or constant torque,niether made much difference.I reset the accel. limits lower from the default 30seconds,down to 5 seconds trying different settings in 5 second intervals.Performance did improve,but cycling the clutch produced the same result of an over current.

I am begining to suspect Marathon motors.The motor they sold me is an all-in-one mount 56C face motor that is supposed to be inverter rated.I wonder if the motor is actually designed for inverter service,or if the insulation is just rated for inverter use.

At any rate I have a meeting with the customer tommorow and I'll see what he thinks.Even if I can get this setup to work,I don't want to deliver something that is marginal at best.He may also need to up the feedrate from it's current max in the future,which I don't think this motor/drive combination will be capable of.I may be looking for a larger drive and motor.It will be over kill,but it will work.

09-20-2006, 09:52 AM
It seems to me your experiments have pointed the finger straight at the drive so I would look at this or even consider changing to a better drive before blaming the motor. You've gone through the parameter settings and I've seen no mention of either current limit settings nor of any sort of "response" setting. On a well designed drive these should be available to set during setup and if you replace the drive make sure they are. The current limit setting sets the maximum current allowed to the motor and if it's working properly you should not get overcurrents. The "response" concerns regulator gain and lead or, in another language, proportional and integral gains. If these are not set right you will get current overshoot on a step change in load which is precisely what's happening--and they can be very difficult to get right. You really need the help of an application engineer from the drive supplier for this. You mentioned adjusting the accel and decel rates in the 5 to 30 second range but these affect only the rate of change of speed when going from one speed setpoint to another so I wouldn't expect much effect on this problem.
I've designed and started stop and go drives like this (usually dc)in larger sizes and usually there is no clutch; you simply accelerate and decelerate the drive. A good drive might allow you to do this.


09-20-2006, 05:58 PM
I too would tend to think that there may be a firmware setting tweak that may give you enough fudge factor to get by.

However, along the line of your orifice idea to limit clutch engagement, I would think that a small reservoir would have the desired effect. The reservoir (empty prior to the air being turned on) would take just a bit to fill and the pressure at the clutch would therefore rise a bit more slowly. I don't know if the clutch would then actuate smoothly and more slowly or not, but it might be worth a try. All it would take is a piece of iron pipe for the reservoir and a few fittings. You could vary the size of the iron pipe (either diameter or length) to tune it.


09-20-2006, 08:57 PM
I'm tending to think that slowing the engagement of the clutch would make things better. Obviously the clutch will be slipping more over its lifetime, but maybe that won't be an issue. A bigger issue will be the engagement characteristics of the clutch. It might be smooth, or maybe grabby. A test would be in order. I go with the orifice idea, maybe something with a needle valve adjustment.

I like Evan's idea of a higher speed motor with a larger gear reduction for the reasons he mentioned. I think there would be less variation reflected back to the VFD as the clutch is engaged.

Something else I wonder about is where the clutch is placed- between motor and gearbox, or after the gearbox-

09-20-2006, 11:25 PM
I've been into the over-current settings and the recovery settings.Over current allows for 5.5 amps max and recovery seems to be a repeat of accel/decel,niether helped.I did talk to the mfg today and the "plugging" load current limit is 3.0amps max,it is not recomended or warrantied beyond that.

I'm going to keep the drive and motor for a normal constant speed/load application.The other requirement here is the input voltage must be 115vac,that eliminates a VFD,since 1/2hp is the max I can find in the 1~115vac in 3~220vac out.

Talked to the customer today and he wants to checkout the puller to see where the speed band falls and see if we need to change the final reduction ratio before setting things in stone.

Darryl,the clutch is between the motor and gear reduction.It's an 1800 rpm nominal motor and a 60:1 box yielding an output of 30rpm more or less.From there it goes into a 5:1 split chain ratio.

I would have prefered to use a clutch directly on the traction roller shaft for obvious reasons,but at the torque requirement I needed there the clutch would have cost $850-1,000,bit steep.

Given the voltage reqirement a DC drive is the obviuos choice at this point.A 1hp drive can be had for $150,the motor though will probibly run $200 or better.I know for a fact the 1hp DC controller and motor will do the trick.

Plus I have the option of mounting the speed pot on the sewing machines clutch arm,so if the speed of the walking foot increases the traction roll follows suit.

J Tiers
09-21-2006, 01:06 PM
It surely seems as if the flywheel should have done SOMETHING..... might not have been good...

The problem with the flywheel is that it makes speed changes pull more current, since a fast speed change requires more energy input.... and you HAVE speed changes.

When the motor is slightly slowed by engaging the drive, it has to speed back up. The flywheel makes that harder, so more current is pulled from the drive. if that goes over a current limit, you have a stumbling problem.

At higher speeds, you have more energy, but you also have to return it to the flywheel faster. With a faster rate of energy return, it is easier to over-current the drive.

I don't think this is really related to a "plugging" current, you are not throwing it directly across the line for reverse, nor apparently are you reversing it at all. Seems like the actual max current limit is your real issue. The only way that the plugging current limit would affect it is if your start-stop duty cycle ends up with as much heating as plugging the motor would... I don't see that, since plug -reversing is a much longer process than a speed recovery.

Most drives like that try to MINIMIZE the mass of any rotating parts that are started and stopped, etc.

Do you have any idea of the mass involved in the input of the reducer? That would have to include the "reflected" mass looking into the drive. All the gears, rollers etc reflect back according to the ratio. The higher the ratio the less "effective mass" is reflected (not rigorous engineer-speak, but you see the point).

How come this has to be slammed into gear so fast that a slower clutch engagement is not good enough?

Clearly the VFD is your limiting factor, based on what you seem to have found. So the DC drive may be best for now.

It doesn't seem reasonable that the AC drive would be that bad.... except that you have a limit since the largest size VFD with 115V input that you can get evidently does not have sufficient peak current.

09-21-2006, 01:35 PM
Since this is a take up roll for fabric what's wrong with using a visco-elastic coupling along with the clutch? So what if it slips? That should reduce the momentary torque load.


From what I understand here you don't need a set rpm, you need a set torque sufficient to reel in the fabric. A slip mechanism that will transmit enough torque while permitting the motor to overrun should work. It can be something like the coupling that I mentioned that works like a auto xmission torque converter or it could be a magnetic eddy current coupling.

See here:

http://www.motionsystemdesign.com/Zone/Article/17784/Transferring_torque_with_permanentmagnet_couplings .aspx

J Tiers
09-21-2006, 04:44 PM
That's a very good idea....

The whole deal is a RATE issue.... you have plenty of AVERAGE power, but you "sometimes" need energy at an instantaneous rate (peak power) that exceeds the power you have.

A slick solution might be the viscoelastic coupling (or similar) between the motor and a flywheel. The flywheel would be connected to the load through the clutch on the gearbox input shaft.

That could limit the peak power required from the motor, while retaining the energy storage of the flywheel and the ability to start the load fast.

The motor would not "see" the peak load, the coupling would prevent it, yet it would speed up the flywheel (restoring the expended kinetic energy) pretty fast.

I am assuming this is not a rapid-fire application, the clutch is operated no more than a few times per minute.

09-21-2006, 08:03 PM
J and Evan,the driven half of the clutch(friction disc and spindle)has a fair amount of mass that must accelerate when it engages.It is probibly equal to the weight of the rotor which would explain the lack of results from adding a flywheel.The rest of the system is moving so slow the major draw would be due to friction.
That brings me back to another point,the wormgear reducer.60:1 ratio on a worm reducer means and effeciency of 60% or less,most being gobbled up in friction.I had the choice between an inline helical reducer and the worm gear,but I chose the wormgear,bad move looking back since they are better than 80% effecient even at that high a ratio.If I had that 20% back the motor and drive would most likely work as planned.

The reason for the fast engagement is this unit idealy does two things.It keeps the material from piling up and it keeps slight tension on the seam being sewn.The first three feet of material is sewn and fed manually,then the puller is supposed to take over and feed the next 100 or so feet.This material BTW is similar to a very heavy truck tarp,it's about .065" thick,Kevlar/Nitrile.

The visco coupling will probibly come into play shortly on the traction roller.I need a coupling there that will apply a constant tension,but still slip in the event of a stall.

09-21-2006, 08:54 PM
This has me wondering if the VFD is needed at all. It seems to me that the cloth exiting the sewing machine will be going at a rate determined by that machine. If your motor and gearbox assy will drive the cloth faster than that at normal 60 hz, then what you need is to allow for a speed reduction only. Slipping a clutch or viscous coupling, or eddy current coupling, will do this. I' would now lean towards an eddy current coupling, since there's no actual contact between surfaces, and it can keep 'slipping' forever without wear as long as the heat buildup isn't too much.

I'm envisioning an aluminum disc attached to the motor shaft, and another aluminum disc attached to the input shaft on the gear reducer. This second disc is free to be moved toward or away from the motor's disc. It does this by sliding on splines or by any other means that will allow easy motion while torgue is being transferred. It gets actuated by a roller that moves in response to a sag or slack in the cloth as it flows out of the sewing machine.

One of the discs will have a series of round neodymium magnets embedded in it. When the discs are close together, there will no doubt be a significant torque transfer, and it will be greater the faster the motor runs. The higher speed motor would probably be better for this, but I don't know. There could very well be more than enough torque transmitted through this coupling than is required, even with the 1725 rpm motor.

This is a guess now, but I will suggest that if 1 inch round magnets are used (these are typically 1/8 thick, or maybe .1 inch) then the disc spacing for minimal torque transfer, which would be just enough to overcome friction in the drive, should be about an inch. If this is right, then the moveable disc needs to have a range of motion of just about an inch. The two discs never need to touch, and in fact by not touching will allow for some misalignment between the motor and the gearbox.

I'm assuming that both discs would be about the diameter of the motor housing, making the mechanical arrangement easier to implement. If I assume this to be 8 inches diameter, then you would need about 10 or so disc magnets embedded into it in a ring about 6 to 6.5 inches in diameter. This is just a series of equally spaced holes that you epoxy the magnets into.

Of course, I'm making the assumption here that you might want to make this up yourself. Maybe you don't.

A.K. Boomer
09-22-2006, 10:39 AM
That brings me back to another point,the wormgear reducer.60:1 ratio on a worm reducer means and effeciency of 60% or less,most being gobbled up in friction.I had the choice between an inline helical reducer and the worm gear,but I chose the wormgear,bad move looking back since they are better than 80% effecient even at that high a ratio.If I had that 20% back the motor and drive would most likely work as planned.

Very surprised at those specs, that is one horsepower gobblin drivetrain, most prob. means that the more load the worse it gets also, that sucks a sour lime for effeciency... Does the worm run in a fluid box or is it just packed with grease? If its something like a 90W you can reduce much load by going with a thinner synthetic but this sounds like it would be just a fraction of what you need to come up with...

09-22-2006, 11:44 AM
Changing lubrication isn't going to make much difference. The efficiency of a worm drive is fixed by the principle of operation just like the poor efficiency of an acme lead screw.

A.K. Boomer
09-22-2006, 01:40 PM
Thats why i stated that it would be just a fraction of what he needed, if he found other area's that could add to this then the entire sum may be enough to make the difference, dont underestimate thinner fluids and power savings, there is a reason that many manual transmissions now run automatic trans. fluid, the main one being that it pushes thier auto's to meet the newer fuel economy standards, the thinner serpentine belt drives also have added to this factor --- every little bit adds up and if your lubrication is superior AND thinner its a win-win...

A.K. Boomer
09-22-2006, 01:50 PM
Changing lubrication isn't going to make much difference. The efficiency of a worm drive is fixed by the principle of operation just like the poor efficiency of an acme lead screw.

Also, there are roller worms that are the same principle of operation but the efficiency is increased drastically, The friction is due to a shearing action or "scuffing" action, it is the same reason we have to run Hypoid lube in the rear ends of vehicles, because the ring and pinon are offset (on most) they have to constantly shear across each other during operation, hence the biggest advantage of the transversely mounted engine front wheel drive cars, not only no hypoid but also no bevel (another H.P. robbing system of power transmission)
So ring and pinon run in same line --- plus its better to pull than push.

09-22-2006, 03:33 PM
I think we (or I at least) need to get the requirements pinned down a bit better. It looks from foregoing statements like you have a process line (the sewing machine) which runs at a fixed speed and the problem drive is a take up drive which is only required to keep tension on the material, and run at whatever speed the process runs at, including standstill. If this is so it is a problem which has been solved many times in many industries such as steel, cloth, paper etc.
The basic control algorithm is that you run the motor in torque mode, not in speed mode. Of course there must be a speed over ride set slightly higher than line speed so that the motor will not run away when there is no strip.
If this is done the drive simply keeps a constant tension on the strip and runs at whatever speed is set by the strip, right down to zero speed and a clutch is not required unless your process has a special need for it. Digital drives, ac or dc normally have provision for this type of operation built in and it requires only the right programming to use it. The torque, which sets the strip tension is normally progammable either via software settings or a separate analog input into the drive controller. In the case of a dc drive there is a limit on the time permitted at zero speed to prevent overheating of commutator bars.

Is there any chance we are making a big problem out of a little one??


09-22-2006, 08:46 PM
AK,it's another one of life's trade offs.Wormgear reducers are fairly compact in size and have few moving parts,the trade off is the friction.As the ratio increases the effeciency decreases just like in other systems,but much it's more drastic.
It is an off the self reducer,56c face mount on the input and a 7/8" keywed shaft output,oil bath,bronze wormwheel and hardened steel worm.I'm already running 75w full synthetic gear oil in it,so the lube isn't the problem,if it were it would be throughout the speed range.

Ken,this system consists of a feeder which pulls the material up off a table and feeds it over to the sewing machine operator.One section over each shoulder,he feeds the two together overlapped and controls the seaming and stiching manually.The reason for the feeder is the material comes in 40" wide strips,two strips 100' long are joined edge to edge,then two more are joined,then those two assemblies are joined and the process repeated until there are two sections 100'x35',then those two sections are joined and the result is a 70'x100' sheet.Once the first two strips are joined they are double folded so they can be feed through the throat of the sewing machine.

This part of the process becomes quite heavy after the first two strips are sewn together not to mention once both half panels are being joined.That feeder operates at a fixed speed and uses the same 1/2hp 60:1 gearbox as the one I am building now,it works great even though it is pulling many more times the load this new one will.
This new feeder,puller actually simply takes the out feed material from the sewing machine and mantains a small amount of tension as it leaves the machine.At the moment the seamer has a helper pulling tension manually as the seams are sewn,it's working,but 8hrs of this is too much on the helpers back.
This is one reason why I need the clutch,I will also need the visco coupling to prevent over running and snapping the sewing needle.

All of this is being done as low cost as possible.The current contract I gather is for around 600 of these mats,after that maybe more,but then to maybe none.There isn't a large enough production number to allow for a fully automated system.Even if there was the material is expensive to which the operation benifits from a hands on system,and it would be vastly complicated by the constantly changing workloads and envelopes.

I just had an idea that I may try.Since the first puller is a mindless brut and does many more times the work this new one will do and it does it with the same 1/2hp.How about I switch the 1/2hp for a 1/3 or even a 1/4hp?

I'm wondering if the drive could hold it's own at 3amps(max)while the motor sees a momentary overcurrent?The motor could certainly handle a few seconds at 3xs the FLA,they do it all the time while plugging so why not?It's also not going to be stopped and started very often.Once at the start of each seam and maybe once or twice on occasion to change bobbins and what not.

09-23-2006, 03:53 PM
It worked!

Swapped the 1/2hp with a 1/3hp Baldor "Invertech" motor I had laying around for years.At max rpm I dropped the clutch in and the current spiked at 2.2 for less than 1 second,then dropped to 1.3 running.No stumbling,overloading or smoking:D

That should do the trick,now all I need to do is build the slip clutch for the drive roll.Thanks for the ideas/help,good to have a brainstorm without hail for once:D

09-23-2006, 04:21 PM

The key statement in your post is:
"This new feeder,puller actually simply takes the out feed material from the sewing machine and maintains a small amount of tension as it leaves the machine."
As is, the existing feeder is a "master" setting speed, I presume at an operator's whim. When you have a human pulling the strip out of the process, he will be holding more or less constant tension and will move at the strip speed, as long as his back holds out. Therefore the new puller drive is a replacement for the human and must hold tension, not speed on the strip. If you run the new drive in speed mode, you have two problems:
1/ You can never set its speed to exactly match the process speed and even if you could you can be sure the process speed will wander a bit with time, even if only 1/1000 of 1%. As long as there is any mismatch in speed the tension will slowly increase or decrease until something drastic happens. Use of a variable frequency drive will help a bit as the rotor slip can change with tension, however motors made for vf operation, as I remember it have very low rotor resistance so it will take a lot of torque change to get a small speed change.
2/ On startup, you are faced with the problem that you have to run a bit fast to take up any loose material and establish tension then you must quickly reset the speed to exactly match the line speed. I can see the operator chasing his tail here and having choice words about the machine.

If the drive is set to torque control the torque is translated to tension by the mechanics and these problems disappear. In torque control, the motor holds a constant torque independent of speed, running at whatever speed is necessary to get the torque. Of course the software puts in a limit on the speed so it will not damage itself. Now, on startup assuming a slack strip the motor runs at the preset top speed until the slack is taken up and torque is established then runs at whatever speed is needed to hold torque thus tension. This will give much less chance of problems than trying to use speed. I doubt that it will increase cost to implement it as it is very likely that provision is made for it in the drive software since this scheme is used almost universally for follower drives.

As for the clutch, I can only assume that you know the requirement better than I so I can't comment on it.

You mentioned early on that you had tried running in torque mode but it made little difference. If you had no load on the motor it would only run at the default max speed and sure enough you would see no difference. If you had put a load on it, it would probably have stalled and sat there holding torque.


Edit: Glad to see in you last post you got rid of one problem, however I would keep the comments above in mind during startup.


09-23-2006, 07:16 PM
KF,will take that under advisement.

The clutch is there basicaly for two reasons,at the start of a run the clutch can be dropped in with the motor and drive running at a pre-set speed hopefully eliminating some of the delay in take-up.The operator can vary the feed of his sewing machine feed simply by easing up on it's clutch briefly until some tension is applied then resume his original feed speed.Actually it could be setup to work in reverse,the operator could slip the sewing feed while the puller catches up to it.

I left out one detail about the existing feeder.It's speed is set slightly faster than the sewing machine's max feed.This provides a loop of material behind and slightly above the operator.If it feeds out too much the loop contacts a limit bar and switches the feed off until the loop shortens and then switches the motor back on to begin feeding again.

I intend to use an over-run slip clutch on the feedroll drive of this new puller so the speed setting needs not be exact,just close.Once the first 3' or so of a run is sewn,the helper will position it on the puller traction roll,drop the press rollers down and then be free to tend the puller.I made sure I had a drive with a speed pot so if more or less speed is needed it can be quickly adjusted by the helper,if not there is the clutch to drop the feeder out.

The over run clutch I think I will build so it has simply a knob and indicator that can be quickly set.I figure I can get the helper to pull on a strain gauge attached to a strip of fabric,then use that same strip and reading to set the slip-clutch.His tension won't break a needle,and the fabric really won't tear anyway.

The setup they are using now is like riding a bicycle,it takes a couple seconds to get started,but once they get started they carry through to the end.The panels are trimmed to size after they are sewn,so that works in our advantage,if the first few inches of a seam isn't perfect,it doesn't matter since it hits the cutting floor.