Can 2 "Identical" 3-ph. AC Motors Be Made to Run in Synch?

Optical encoders with some type of syncro control?

Hook the shafts together with gears?

How exact is exact? 3450RPM vs 3451RPM?

I don't know.

One single motor with two sets of gear systems in opposition possible ?

Oops. Lyn already said this. Can't cancel the post unfortunately.. consider this unsaid.

Can't think that even two motors from the same batch from the same manufacturer will be identical...

And even if they were, the loads on each might differ, hence the slip and the speed.

By the way have you seen Double chuck glass blowing machines? it can be a ready made solution. Glass tubes don't brakes in rotation when clamped in both sides jaw.

Only goes to 400 RPM and who knows how much HP. Need around 3500 RPM - 15 HP per 2 motors, or around 25-30 HP for a single motor driving two outputs @ 3500 RPM.

A 1:1 gearbox connected to both shafts would satisfy the specificatiion so far, as would turning both output shafts through this gearbox from just one motor.

_{What does ASAP mean in this context? Things are always done as soon as possible.}

You have to firsat define "exact". A pair of good quality Closed Loop Vector Drives running in Electronic Drive Shaft mode will give you +-1% speed accuracy, maybe better. But NO TWO THINGS can be absolutely perfectly exact.

Here's a web site I found that talks about this. I hope it helps.

http://www.controldesign.com/articles/2002/235.html

Connecting multiple motors together is not a problem provided you meet a few important criteria first.

If they are standard non synchronous induction motors all you need it have all identical motors connected to run at equal speed. Typically thats done by either direct shaft coupling or a fairly close matched drive belt system.

If they are synchronous motors then you have to also have all identical motors and the shafts first have to be indexed to within a few degrees of each other. If you know how to use a low voltage DC excitement of both the stators and the rotors you can get the motors to line themselves up exactly before the shaft couplers are locked in place.

Using multiple motors tied together is not a big deal if you understand the operating principles and basic physics behind how each type of motor system works.

Why not redesign to use just one motor for both positions, always in sync no matter what the speed. Really simple. Just use toothed belts or similar to take off the drive.....

Anything else with two motors will need electronic or mechanical (gears/toothed belts) connection, the electronic method may still not be as accurate as you wish, within 1% would be normal I feel......but exact, I think not, or better said, within a reasonable price.....pay top dollar and you might achieve it.....

Can you define for us EXACTLY how accurate it must be and for how long.....

The only other way might be two synchronous AC motors with electronic speed control, but as I am no expert in this area, maybe someone else here can speak better on the subject.......The motors might need to be even larger so that changing loads on one do not drop it out of sync......

Really interesting problem......many thanks for bringing it to our attention.

How about a dual-shaft motor (output shaft at either end)? I've used small (0.37kw) IEC versions of these, but larger NEMA motors should also be available as dual-shaft.

"... the attached mechanisms will operate 180 degrees out of phase at all times."

A single motor shaft(1) with a single gear to a parallel shaft(2) with the same gear would easily counter-rotate shaft(2). Repeat with Shaft(3). 3 shafts with the middle counter-rotating. All 3 rotating at the same RPM (speed). Connect mechanism(s) 180 degrees out of phase to either shaft 2 or 3 as applicable. Increase specs on Motor. Perhaps that is a simpler and better solution if the "AT ALL TIMES" is accurate.

If there is any risk of damage or other "failure" if the motors go out of sync. then I suggest you add a pair of meshing gears to the 2 motor outputs, and just use 1 motor (appropriately sized). Assuming this arrangement is possible in your application.

This might help:

http://www.controldesign.com/articles/2002/235.html

That was my tip.....in post #10.

Sorry, I missed it. My industry has used dual motors locked in phase for many years, so this is not a new idea. We even run one motor separately, then lock the two together as needed to run 1 to 1, and the timing must be perfect. The motors aren't even the same HP. One is a 15hp, the other is a 10hp.

OK, How?

We use wound rotor induction motors.

For an explaination of how it works, you can call the manufacturer of the FL Smithe Envelope Machine at 1-814-695-5521, ask for an electrical engineer, (which I am not) and ask about the dual motor drive on a RA 600 envelope machine. He may take the time to talk with you about it.

These have been in use for over 40 years.

Thanks.

It was STL Engineer who wanted to know how, now he can find out for sure.

What was really cool about this set-up was when the machine wasn't running, you could manually turn one driven side of the machine, and the other motor would turn and follow. Almost spooky in a way. When they were miss-wired, like after a rebuild, they could run in sync, but one would run the opposite direction, as the OP wanted them to do. I have the schematics for these, but they're not in very good shape and I don't have the capacity to scan them in. Best bet for anyone REALLY interested is to call the number I posted before.

Do you know who (what company) built the envelope machine, or was it an in-house design-build job? I am really no expert on electrical motors or motor controls, so I am not sure what type of motor you are referring to, or the drive system.

The apparent lack of interest from the OP astounds me. Interesting problem, lots of people wanting to help, no response.

Why am I required to respond? I have followed the response and while interesting, have not seen any definitive solution. I did not see any need to comment, I am simply seeking ideas. Besides, other projects have been made a priority for me since I originally posted.

At least I am gratified you consider it an interesting problem. I will address the suggestions as soon as I can review the project requirements with my boss, who has "bigger fish to fry" at the moment.

It is a CR4 courtesy to drop in and make a short post thanking or criticising the recent posts, just to show that all this help is not just going to waste!!!

Its a 2 way street......

From my experience, manufacturers' claims, especially from those in the Far East, are often exagerated or, at best, calculated for ideal or theoretical conditions. In any event, "exact" means zero variation, of which, as it appears now, only a non-slip mechanical linkage would be capable. If you re-read my reply you should understand why this is necessary.

A direct mechanical linkage/synchronization might be the simplest and cheapest method. But it is incorrect to say that zero variation cannot be achieved by non-mechanical means.

If you look at the SIHIdry vacuum pump SIHI , it is a twin-screw vacuum pump with electronic synchronization of the rotors, and no mechanical linkage. It has been around for a while. I don't know what type of motors they use, but it's quite clever.

I'm sure I have also seen, sometime in the past, a printing press where all the rollers had individual motors, electronically synchronized. And you also have the machine mentioned by Envelope Guy .

These are all applications where zero variation is critical, so it is certainly possible. It might be an idea to consult a motor specialist on this.

You are fully correct, there are electronic methods to achieve just that requirement, but most of these do not seem to "fit" in with the OP's requirements. for some reason or another.

I am not drawn personally in any particular direction, but it would appear that a stable and reliable mechanical solution might be best way for him to go. Also easily understood by any maintenance personnel involved.

It will be most interesting (for me personally at least) to see what the final solution might be!! Mechanical, electrical, electronic or a mixture of several different technologies.....

Great Blog!!

Don't get me wrong. I know that "zero variation" (within practical limits) is certainly possible, especially when closed loop electrical control is provided with positional feedback encoders. Now you are in the area of servo-motors, or at least servo control. I am only questioning whether frequency control would be adequate to guarantee 100.00000% accuracy. Then there is the cost issue. A simple mechanical solution is most likely the best application in this case. There is plenty of available area, it will be well guarded, and the price is right.

You will not get zero variation with any approach, mechanical linkage, gears, whatever, they all have slack of some sort.

Yes, but it is the type of variation that causes the problem. "Slack", "backlash", etc. usually only happens on start and stop, or reversal, and is NOT cumulative. If it is, you have a whole different problem! In fact, error in one direction occurs only briefly, is very limited, and corrects itself in the case of reversal (e.g. cyclic motion). Error in speed matching via open or closed loop control (except for positional, i.e. servo feedback) can be cumulative and build up to an unacceptable point very quickly. In my situation, a speed matching error would be disastrous as I explained in another post.

OK, I will try to summarize my response to some of the suggestions. Thanks to all.

1. Of course I thought of using a larger single motor, but my boss balked at the suggestions, without really giving me an explanation, other than to say that the high relative speed was necessary that two motors working oppositely could provide. In researching availability of motors from our normal source (Grainger) the largest HP motor I could find was 25 HP and none of them had a speed over 2000 RPM. Now, I know what some of you will say, just gear up to double the speed, but then that will reduce available torque, will it not?

2. Not sure what advantage a dual-shaft motor would have, unless it included a gearbox to run the second shaft in reverse. Otherwise, two right-angle gearboxes might be required. And if we went the gearbox/single motor route, then a single motor output, with a power take-off gear or sprocket, driving a reversing gearbox, would work just as well. However, those would be expensive, power-robbing solutions, I believe.

3. My boss also balked at the suggestion of servo-motors. Possibly a cost/speed/power issue. Besides, unless they could track at 100% accuracy, it would not work. The problem is that at top speed, say 3600 RPM, the speed would be 60 rev./sec, giving a cycle time of 1/60 second, or a frequency of 60 Hz. If there is a +/-1% (36 RPM) difference in speed, or say 59.4 and 60.6 Hz, then the resulting "beat" frequency would be 1.2 Hz, or roughly just over once per second. This means that the relative speeds of the two targets would become zero repetitively, and worse, would be continously speeding up and slowing down between zero and the top speed. To my simple mind, that means that the effective average relative velocity would be cut in half. There will already be some slowing/speeding during each cycle due to the nature of the sinusoidal speed profile developed by the slider/crank mechansm, but at least the point of high speed would be consistent at the mid-point of the stroke (twice per cycle) and wear results at that point could be analyzed if the driver motors were in perfect synch.

3. So far my thinking is that some kind of mechanical synchronization is needed, although I am intrigued about the claims for the "Envelope Machine" drive and would like to know more about that. I am thinking that a roller chain and sprocket set up in a figure-eight configuration could work. I do know that flexible steel roller chain is available that could allow the figure-eight crossover point between the two motors. Anyone have any experience here? What would work best, a low friction, slippery seperator material (channels?), or twin sprockets mounted on a common axle? Or is there another idea?

Thanks again for your suggestions. I look forward to hearing more, and will let you know when I get any more input from my boss on this.

STL

Why make life difficult with a figure of eight chain?

Either use gears to link them or place them at 180° degrees to each other so that as one motor is running reversed, theý will both be able to run a chain in a loop without being a figure of eight.....

The reasoning has mostly to do with the layout of the components on the machine base and a desire to follow the KISS principle. The simplest, and best, layout would have the motors turn the cranks (disks in this case) to drive the slider components, test piece fixtures mounted on linear shafts guided by linear ball bearings (pillow blocks). Since the shafts are parallel and counter-rotating, a timing belt would have to act through some sort of reversal. A figure-eight seemed to be simpler and less expensive than a reversing gearbox, since the chain would be necessary anyway due to the substantial separation, 80 inches (about 2M). All that should be needed would be sprockets on the motor shafts and an idler in the center with two parallel sprockets mounted on free-turning bearings on a common fixed shaft to keep the chain sections separated where it crosses itself, much like the overpass in a figure-eight racecourse.

This could also be done with a 2-sided timing belt and one or two "dummy" or idler pulleys (the one/two being determined by how much belt wrap is needed on the driven shafts).

Great idea! Thanks.

At the RPMs envisaged, I do not anticipate that a "figure of eight" chain will have a long life.

Especially if the motors "try" to have a different speed at any point in the cycle, possibly straining the chain, causing further excessive wear.

PLEASE CAREFULLY NOTE THAT chains and belts need to run in perfect alignment to have any life at all........a figure of eight precludes that perfect alignment.

Its simply completely mechanically unsound.

DO NOT DO THIS, you will be MOST unhappy with the result. It also tells me that you are most unlikely to be a mechanical engineer......or you would not propose such a scheme.....You must reverse one motor with a gearbox or similar and then use a belt or chain to link them....

As you do not show us exactly how the two motors are physically placed with regard to each other, its difficult for us to help you further. A simple sketch could improve that situation dramatically.

Also, do not forget the forces that are developed by spinning chains etc. can be very excessive, screens need to be placed around such parts to "catch" them in the event of a mechanical failure.

Also, some sort of failure warning needs to be made if the two motors go out of sync with each other......

"At the RPMs envisaged, I do not anticipate that a "figure of eight" chain will have a long life." Please define "long life". This tester is not for production, but for R&D. Actual test time would be relatively short (minutes to a few hours per day).

"Especially if the motors "try" to have a different speed at any point in the cycle, possibly straining the chain, causing further excessive wear." As long as they stay in synch, motors will be seeing identical loading at every point in the cycle.

"PLEASE CAREFULLY NOTE THAT chains and belts need to run in perfect alignment to have any life at all........a figure of eight precludes that perfect alignment." Not entirely true. There are "flexible" chains that have side play and are capable of transverse displacement, albeit with a fairly large radius. For examples see:

http://www.mcmaster.com/#flexible-roller-chain/=brg38o

The catalog describes "Flexible ANSI Roller Chain" as: "The perfect solution when sprockets can't be exactly aligned, this chain has extra clearance between chain parts so it can curve side-to-side. Chain is made of carbon steel and is single strand."

Its simply completely mechanically unsound. ..in your opinion. Using flexible chain, a side displacement of around 1 inch (about 25mm) over a distance of 40 inches (about 1M) does not seem "mechanically unsound".

DO NOT DO THIS, you will be MOST unhappy with the result. It also tells me that you are most unlikely to be a mechanical engineer......or you would not propose such a scheme.....You must reverse one motor with a gearbox or similar and then use a belt or chain to link them.... WOW, "unlikely to be a mechanical engineer"? I guess I was just wasting the last 30 years!! $*&@#$ Besides, a gearbox is NOT necessary to reverse a 3-phase AC Motor, they are designed to run in either direction. I would never consider a "belt" for this type of application. Belts typically slip or stretch, steel chain does not (within its design parameters).

As you do not show us exactly how the two motors are physically placed with regard to each other, its difficult for us to help you further. A simple sketch could improve that situation dramatically. As I said before, it is a confidential competitive situation, and I do not want sketches floating around the web. I have provided plenty of information, and others have been very helpful WITHOUT any sketches.

Also, do not forget the forces that are developed by spinning chains etc. can be very excessive, screens need to be placed around such parts to "catch" them in the event of a mechanical failure. DUH! Every machine with moving parts under power I ever designed included a safety guard.

Also, some sort of failure warning needs to be made if the two motors go out of sync with each other...... If the motors go out of synch when linked by a chain, I think we will know it! ROFLMAO! Chains and sprockets breaking and whipping around makes a helluva big noise! Still, not a bad idea to electronically monitor the situation and shut-off power to avoid a total beat down!

Sorry if my tone is a bit snippy, but it seems you were sniping at me!!

Sniping was not intended, some sound advice was intended....what is the importance of R&D over a short period? Especially when there are mechanically far better solutions around.

But hey, you carry the can for all failures, not me!! Why should I bother about it?

If you are talking "slider/crank mechanism" - why not think about synchronizing those?

The closer to the 'precision point' the better the accuracy. I.e. after all the slack and tolerances, not before.

Hmmm. Good suggestion. Not sure how I would implement it though as the two sliders move in parallel and oppositely.

1. Make a belt, or chain, or string, loop between two idler pulleys - then attach one thingy to one run, and the other thingy to the other run.

Like vertical blinds work?

2. Make a slot at 90⁰, with a block, or bar with liner bearing. From that as a pivot - run two arms out to the sliders in a 'V'.

As your pantograph or scissor lift work.

3. Attach two racks to the sliders and have them run top and bottom of a central gear (on a fixed shaft)

As your 'distance doubler' works

These are the 'common' methods.

Depending on your mass and environment and physicals - one of these should/could do it.

Thanks, 34.5! Much appreciated.

Your welcome. I'm leaving you to decide what fits your dynamics, but if you want to chat further, feel free to use the private mail facility (click on my ID, then 'send a message' top rh corner of my profile page).

Adding some detail to the earlier Post 30:

Imagine 4 pulleys in this configuration:

A
------B------C
D

Wrap a double-sided timing (toothed) belt from the top of A to the top of C, from the bottom of C to the bottom of D, from the top of D to the bottom of B, and finally from the top of B to the bottom of A. B and C are the counterrotating motor shafts; A and D are the idlers. The belt needs only to handle the small torque difference between the two motors, so a fairly light-weight belt should do. This setup gives good belt wrap around B and C, and shouldn't be too costly. Best of luck with your project.

I thought about that the first time you described it and believe it is a solid approach.

Two factors have been added since;

1. being the 2 m centers

2. the variable inertia of the 'mechanism'.

'2'. may well make my suggestions 'easy', or 'difficult'.

If 'difficult' - I would go for your approach over all others suggested.

But, given '1', I'd be inclined to use 2 belts and put A, B, D, midway as an idler shaft assembly. B then becomes a, pair of "B's", or a double width B.

Then belt from B₂ across to E [off the page left], the other motor shaft.

From what's been "said!" since, it does rather sound like a 30 hp 'collider' , so I would also be inclined to go for;

Σ. steel reinforced belt for least stretch

Ψ. pulley diameters and belt widths 'rated' to not break if one motor goes down, or something jambs, or it doesn't make full stroke, and/or breakage downstream of the synchronization occurs.

Thoughts?

I don't know how much further detail can be provided on load inertia and so forth without spilling too many beans. My guess, right or wrong, is that the load characteristics are (nearly) identical between the shafts/motors. The high rpm all but rules out chain drives (silent chain might work, but I don't know if two-sided versions are available). If the loads do differ, your further suggestions might need to be in play.

Yes - double sided silent exists - fairly sure Tsubaki still make it also - but still 15 hp at 2 m centers? cant be a tiny sprocket, so the velocity and mass 'bit exciting'.

Which is why, when I started to get a bit concerned with inertia in the reciprocating end of things, I thought at least your idea is a more 'constant velocity'.

I suspect this is a cyclic manner of power use. Say similar to a flywheel type power press. Big ask for VFD to hold twins of those true without surge problems.

Anyhow we may hear some thoughts from the OP

[p.s. GA 4 #30]

Tornado and 34.5: You are right on the money. Though not an actual collider, the motion is similar. With the proper tension and lubrication, why should a roller chain be especially noisy? Besides, chains are by nature double-sided, so implementing your ABCD model should be relatively easy. I am thinking there may be more noise from the slapping action of the slider/crank and the scraping action of the two friction "coupons" as they slide back-and-forth across each other under load. If noise is a problem, there is plenty of room to place sound insulation around the timing chain.

Not sure what you mean by "tiny" sprocket. Motor shaft is 1.625 in (41.3mm) and the smallest sprocket I am considering is 3.5 in. outside diameter (16 teeth, .625inch pitch). And I do prefer working with the constant velocity motors for synchronization rather than the reciprocating elements. Much cleaner that way, no worries about changing momentum.

Also, by my calculations, given 3450 RPM target speed of motors and 3.5 inch dia. sprocket, chain speed should be poking along at roughly 52.7 fps (about 16 M/s), or about 35.9 MPH (57.8 KPH). Is that a problem for steel timing chain?