# mechanical load sharing of coupled motors

Discussion in 'Physics' started by strantor, Sep 19, 2011.

1. ### strantor Thread Starter AAC Fanatic!

Oct 3, 2010
4,302
1,988
This post is for the electric go cart project I'm brewing up (bought a rolling frame today, it's UGLY 50\$!)
I think I want about 20hp worth of motor on it. I can get 2 10hp motors for much cheaper (and lighter weight) than I can get one 20hp motor.

I've seen already here that mechanically coupling and electrically paralleling brushed DC motors results in one motor (the one with the least resistance) taking the bulk of the electrical power, to the point of melting. I am wondering, Assuming 2 identical (not perfect world "identical", but same model bought from the same place at the same time) motors are coupled to a shaft and provided the same PWM duty cycle; if by using 2 seperate controllers (or 2 seperate banks of FETs in the same controller) for 2 motors, and mechanically coupling them, would they share a load evenly? Well, this makes me think that if they are both receiving identical PWM, then functionally they are still in parallel, just with seperate switches between the motors and ground and would encounter the same problem.

Another idea: what if on one motor (call it master motor) the PWM duty cycle is controlled by the accelerator, and the the duty cycle for the second motor (call it slave motor) is controlled to maintain a match of the current measured on the first motor?
Anybody have any wisdom to share about load sharing with DC brushed motors? I have a feeling it's doable but probably a lot trickier than first thought.

one question, on the theory of it; which of the following 2 statements are the closest to being true?
1. If the 2 motors aren't putting out the exact same torque at the same time, then the lower torque motor is actually acting as a load on the higher torque motor and slowing it down, not helping it. (I.E. motor 1 is putting out 100 ft*lbs of torque, motor 2 is putting out 75ft*lbs; total torque is 75ft*lbs) Chain is only as strong as it's weakest link.
2. If the 2 motors aren't putting out the exact same torque at the same time, then the lower torque motor is adding to total torque, but to a lesser degree than it could be if they were matched.(I.E. motor 1 is putting out 100 ft*lbs of torque, motor 2 is putting out 75ft*lbs; total torque is 175ft*lbs)

2. ### shortbus AAC Fanatic!

Sep 30, 2009
4,013
1,531
Are you still talking about starter motors? A starter motor is no where near 10 HP. What motors are you using?

Oct 3, 2010
4,302
1,988
4. ### THE_RB AAC Fanatic!

Feb 11, 2008
5,435
1,305
2 is correct, The two motors on the same shaft are both making positive torque so both torques are added to total shaft torque.

Paralleling motors is not that hard and is done in industry. Two identical motors of the same speed:volts curve will parallel fine, and some motor types even include 2 or more "stacks" inside the same motor, on the same shaft making a longer motor.

Anyway if you have 2 motors of the same brand and type on the same shaft they should parallel very well driven from the same PSU.

5. ### strantor Thread Starter AAC Fanatic!

Oct 3, 2010
4,302
1,988
Thanks for that; that's what I was hoping

Did you see the link in my first post? That guy was using 2 "identical" motors (same brand/model) but they were not identical, so one took all the load. Or maybe his motor leads were not equal length/resistance. In industry I have read (for example, old DC conveyor systems) where many DC motors are paralleled and motor leads are unequal length, they would use variable field resistors and adjust the resistance for each individual motor until it took out the slack between motors. Since the motors I'm talking about are permanent magnet motors, that can't be done. In the e-bike/e-go cart world that I'm stepping into, they usually wire 2 motors in series and double the applied voltage. This solution is not ideal, because that means you double the battery weight. I'm trying to come up with a win/win solution.

6. ### t_n_k AAC Fanatic!

Mar 6, 2009
5,448
782
Did some simulations (using PSIM) of two identical separately excited DC motors on a common shaft driving a constant torque load at about 10HP. With a 5% difference in field excitation, I was surprised to obtain a load share ratio of the order of 18:1. The lower excitation motor takes the bulk of shaft power. This confirms your notions of small differences leading to large disparities in load sharing.

strantor likes this.
7. ### shortbus AAC Fanatic!

Sep 30, 2009
4,013
1,531
Have you looked into getting a motor from an electric forklift? Around here the industrial surplus places have them fairly cheap.

8. ### THE_RB AAC Fanatic!

Feb 11, 2008
5,435
1,305
Not with DC permanent magnet motors. The motors will do about 1200 RPM on 12v with no load, and sag to maybe 800 RPM at 12v with a decent load. That represents a fairly "soft knee" that allows the motors to be paralleled quite effectively.

You should be able to get the rpm/voltage/load curves for your desired motors and do some math if you like.

Imagine;
motor A; 12v 1A 1200 RPM, 12v 10A 900 RPM
motor B; 12v 1A 1250 RPM, 12v 10A 950 RPM
(what will the 2 motor currents be at shaft speed 1100 RPM?)

If you are worried about motor sharing and this is a high dynamic load situation you can always parallel the two motors and use two current sense resistors, and control the two PWMs to give equal current.

Interesting result, but I don't think you will get that problem with perm magnet motors in the real world.

9. ### strantor Thread Starter AAC Fanatic!

Oct 3, 2010
4,302
1,988
Man, this is getting more and more antagonizing. I've asked this question in multiple forums and got a whole host of conflicting answers (even within the same forum, I have people arguing about it in more than one place). All the application notes I find on the web which address this issue are DC drive application notes and they focus on field would motors, not permanent magnet motors. I have found industial application notes that seem to agree about the "speed controlled master, with current copying slave motors"; that's how they used to do conveyor lines. But none of these notes go into great detail about how well the load is actually shared (which is critical to me) and again, they focus on field wound motors. Guys in the E-bike racing circles with practical experience doing exactly what I'm describing with permanent magnet motors don't even agree; some say they've done it, it's easy, they share the load as long as the current is the same; others say they've done it, and it's a recipie for disaster, even with the same current in the armature, they won't share a load.

yes, briefly. Went a lot further looking into after reading your recommendation. Looks like an economical solution, but these forklift motors look heavy; and they should be, being in a forklift, they try to make forklifts as heavy as possible. Also, I cannot find any info on the motors; they don't publish data sheets (that I can find) because they're purpose-built (I assume that's why) and people just buy them as replacements, going off a part#.

your motor A & motor B numbers are what's confusing me (the idea behind it). Let's say, using your numbers, motor A is the Master motor. The controller is configured to make the current in motor B (slave) match the current in motor A. So right now, using your numbers, We're drawing 10A so theoretically on the electrical side, we should be delivering equal torque, but in reality, mechanically, the slave motor is overrunning the master motor, causing it to draw less amps, so in turn the controller would tell the slave motor to draw less amps, so what would happen? would they get locked into a loop of drawing less and less amps until holding the accelerator all the way down does nothing? And what if we reverse the numbers, so now motor B is master and A is slave. So motor B is drawing 10A and trying to turn 950rpm and motor A is drawing 10A and trying to turn 900rpm *now this is where I get confused* are the coupled motors running at 950rpm, or are they running <950 rpm? if they are running <950, then motor A is acting as a load on motor B, decreasing efficiency and output power. if they are still running 950rpm then I guess that means that motor A is helping out, and there's the answer to the overruning problem, just make the weaker motor the slave and all problems solved (but that could change during operation as motors may heat differently i think).

for the sake of arguement, lets say motor B is master and motor A is slave. you send motor B 10A of current, but send motor a only 1A. Is motor A a load, or is it helping? at what point does the slave motor become a load?

wait, I think I figured it out; as long as you send the slave motor enough voltage to spin the same RPM of the master motor, then it is helping. any less and it is a load. is that right?

10. ### t_n_k AAC Fanatic!

Mar 6, 2009
5,448
782
Again some thoughts on the simulation results.

If I run the separately excited motors in parallel (armatures) from the same supply vs running them in series from the same supply then the outcome is quite different.

The previous result I quoted was for identical motors with the only difference being field voltage excitation. This was only a 5% difference.

Running the motors (armatures) in series from the same supply gives a situation in which the load sharing is far more robust. A 2:1 ratio in field excitation voltage gives a 2:1 ratio in load sharing - as one would probably expect.

Just a thought: The issue about whether the motor fields are derived from permanent magnets or from fixed electromagnetic winding field seems a secondary matter to me. The field is there to provide the armature emf - if it (the field) is a constant value, then for a pair of connected machines at steady-state condition any knowledge of where the field comes from might be indistinguishable in relation to the combined motor performance.

I think the crucial factors for motors powered in parallel from the same source & running at identical speed are the individual motor armature emf and the armature resistance values. A low armature resistance and small difference in emf can lead to substantial difference in armature currents & therefore the resultant torques.

Last edited: Sep 21, 2011
11. ### strantor Thread Starter AAC Fanatic!

Oct 3, 2010
4,302
1,988
I was only specific about the fact that they were permanent magnet motors, is because playing with field current is the common old school way of getting DC motors to load share. I wanted to save anybody the time of suggesting I put variable resistors in series with the fields to adjust the ratio. does your simulator have the ability to change the brush resistance of one of the motors? I would be curious to see the result of that.

12. ### t_n_k AAC Fanatic!

Mar 6, 2009
5,448
782
Not specifically brush resistance but the model does include the lumped armature resistance which would include all the series resistances.

The load sharing isn't as sensitive to differences in the armature resistance compared with differences in the armature emf. For the somewhat arbitrary motor parameter values I'm using in the motor model, a 10% change in Ra (with everything else the same - including field excitation) equated to about a 7% difference in motor powers. So the armature emf looks like the prime "suspect".

13. ### THE_RB AAC Fanatic!

Feb 11, 2008
5,435
1,305
Yes the armature EMF is the most closely related factor to the speed, and "identical" permanent magnet DC motors will have quite closely matched rpm:volts ratios. Two DC perm mag motors in parallel will have the same armature volts (obviously).

No, nothing "over-runs" anything, the answer is simpler than that. Devices with a "soft knee" for load can be paralleled reasonably well. I happened to have some nice 24v DC perm mag servomotors on my testbench as I was considering adapting them for CNC machine control. I just did some tests using my regulated PSU and an optical tacho;

16v 0.4A 1350 RPM (no load)
16v 1.0A 1250 RPM (load applied)
16v 2.0A 1100 RPM "
16v 3.0A 950 RPM "

The second motor was within 1% or 2% of identical speed as the first motor, so they are "identical" I guess, they do have the same part numbers.

The soft knee is pretty apparent, and is quite common to DC perm mag motors.

So if 2 motors were in parallel (at 16v) and on the same shaft, when you loaded the shaft with enough load so RPM goes down to 950 RPM both motors would be drawing about 3 amps each.

As an electrical analogy it's like 2 LEDs in parallel, and each LED has a resistor in series. They may be slightly mismatched but one LED will never draw ALL the power, because of the soft knee. The softer the knee the better the current sharing. And White LEDs, with their higher Vf and relatively soft knee built in, are paralleled by many manufacturers WITHOUT needing a series resistor for each LED.

14. ### strantor Thread Starter AAC Fanatic!

Oct 3, 2010
4,302
1,988
I got that idea of overruning from the attached document pg5
Thanks for the real world example, that makes things clearer for me. But, just to throw another curve ball, I'm now entertaining the idea of making a BLDC motor controller instead of a regular brushed motor controller. I suspect that, being synchronous motors, the brushless types will have a lot less of a "soft knee" and harder to synchronize. is that true?

File size:
113.6 KB
Views:
93
15. ### GetDeviceInfo Senior Member

Jun 7, 2009
1,571
230
consider if on the end of a shaft you place a one foot bar, upon which you place a hundred pound wieght. You would have a 100ft/lbs of torque applied to the shaft. Now add an extra 75 pounds. The result would be 175ft/lbs of torque. This would hold true if 75 lbs where added via a second bar at a different point of the shaft.

In a parallel motor situation, as long as the motors are not at full unloaded speed, they will be adding some torque to the shaft. This is true any time that Emf is less than supply.

In regards to the findings of the seperately excited fields, field strength and it's ability to generate Emf change substantially with small changes in field current, so the results are not surprising.

where you could run into problems is when the paralleled motors are generating different Emfs at the same speed, and you are using that Emf as speed regulation feedback. In industry, this is typically overcome by using shaft mounted encoders/resolvers.

The BLDC should be no different, unless you are using frequency speed control where you'd need to phase time the controls to make up for mechanical seperation.

16. ### THE_RB AAC Fanatic!

Feb 11, 2008
5,435
1,305
If they are synschonous 3phase motors on the same shaft, and you align them in phase (and parallel or series connect the 3 windings) you have created a single 3phase motor in every sense of the word, and can drive it from a single controller.

But I think you will find a BLDC controller MUCH harder to build and much more expensive when you blow it up!

There's no reason not to use a brushed DC motor the brushes will last many times longer than your battery system, and the brushes will cost very little to replace (unlike the battery).

Yep that's true, but in DC motors the rpm:volts as determined by magnet type and the number of turns of wire on the armature. In most cases "identical" motors (with the same part number) will have the same magnets and the number of armature turns wound by machine, so they will be very close to a match in their rpm:volts ratio. And even a few percent mismatch should only cause a few percent difference in the current sharing under load.

strantor likes this.
17. ### strantor Thread Starter AAC Fanatic!

Oct 3, 2010
4,302
1,988
Am I that transparent? Right now I'm in the middle of reading microchip AN857 "Brushless DC Motor Control Made Easy". You may have noticed last week that I was asking a flurry of questions about brushed DC motor control. Now questions about motors. That's because I had just taken for granted the motor. now I've taken a step back, seeing what's out there, and where to go from here. good thing I hadn't ordered any parts yet, as I nearly had my whole controller designed. I am weighing all my options... so why, other than the reasons which are right now apparent to me (need 6 FETs/FET banks, need to incorporate commutating logic), would it be much harder and more expensive to make a brushless controller? I'm only on page 11 of the AN857 and have other docs to read so really don't have a good idea what I'm getting into yet. Assuming I don't blow up my controller first try, I can actually find a brushless motor of comparable (actually slightly higher) HP than the brushed ones I was looking, at for about 50\$ cheaper, so I would save money.

18. ### Wendy Moderator

Mar 24, 2008
20,766
2,536
Combat robotics has done quite a bit of research on the subject. I remember reading some of the heavy weights (500 pounds and bigger) drawing over 500A stall current before moving. There are a bunch of online stores to support the hobby too.

strantor likes this.
19. ### shortbus AAC Fanatic!

Sep 30, 2009
4,013
1,531
This goes back to what I've said a few times in your threads A VFD using a DC input is exactly what your describing. It also has the advantage of not needing the Hall sensors, if you use a "sensor-less vector" VFD.

strantor likes this.
20. ### strantor Thread Starter AAC Fanatic!

Oct 3, 2010
4,302
1,988
Yes, the circuit is functionally the same; I would be building my own VFD. the main problem is the voltage (as I've said a few times in response) If I wanted to use an off the shelf motor, I'd have to have it rewound for a lower voltage, or increase my battery pack voltage. having a battery pack on the order of hundreds of volts isn't really cheap or practical from my hobbyist perspective (I'll be using deep cycle marine batteries, at 12V each). I am finding 48V brushless motors a lot easier to come by than 48VAC induction motors. for example I can get this motor (just the motor) for 390\$. Where can I get a 48V 15hp induction motor that only weighs 22lbs?

or if I somehow developed deep pockets, there's this one:
http://www.ebay.com/itm/ME1012-Brus...844?pt=LH_DefaultDomain_0&hash=item3f0b34de74
72V, 32hp, 92% efficiency, 35lbs for 82\$.
A 32hp induction motor is going to weigh hundreds of pounds.