Electric Motor to Simulate Mass?

If you are interested in using electric motors to create a variable load, I would suggest using a stepping motor. They are "digital" meaning that you can easily control the input, rather than varying the volts or current to control the load factor. In some cases stepper motors can be electricly braked to hold the load. This avoids the stalling of a normal motor which is a hazzard especially if the load is 200kg. Size wise they range from a size to fit in a hard drive to ones that operate elevators.

If this becomes complicated, try using springs. They to can be accurate and if properly calibrated, they can act as the feed back source like a fish scale.

Good luck.

Check out http://www.anaheimautomation.com/intro.htm

or http://en.wikipedia.org/wiki/Stepper_motor

Thanks for the speedy reply :)

Springs is definately out, we need to electronically control the 'loadweight'.

Won't stepper motors have 'steps' in the load ?, we need smooth transitions.

Brakes are also out, the force needs to simulate weight or mass.

The effect we want is that of a stalled motor applying controllable torque to a shaft I guess, which can be cyclically dragged against the direction it's trying to run in, and do this all day long.

Why not try a hydraulic coupler attached between the motor and a drum. This will give you the dynamic load that you request and the motor does not need to be stalled. A stalled electric motor can be very dangerous in terms of overheating and fire. Plus they will not last long given that you plan to run it all day. A hydraulic coupler works exactly like an automatic transmission torque converter. It is smooth in operation and the amount of torque is determined by the RPM of the motor. It can be stalled and even reversed if needed.

Thanks for the hydraulic coupler idea but is getting very complicated :(

I guess we are trying to make a motor do all the things it doesnt like to do...

Does anyone know of a motor / motion control manufacturer that has a really good helpful application / design support department that we could approach ? If we can get it right there would be several units to build.

cheers :)

Guest wrote 'Does anyone know of a motor / motion control manufacturer that has a really good helpful application / design support department'

I have had good experiences with Rockwell Automation, they can be found online at www.rockwellautomation.com

Okay, here's my two cents' worth: The winch drum is coupled to a permanent-magnet D.C. motor through a gearbox. The gear ratio between the winch drum and the motor shaft is fixed at some value such that the motor spins at some "reasonable" RPM as the cable is pulled at whatever linear rate you're pulling it. It is presumed that that the motor will spin at some higher (perhaps significantly higher) rate than the drum. If it's not practical to use a fixed-ratio gearbox, you might use some kind of transmission or even one of those fancy infinitely-variable trannies like the kind in my Civic Hybrid. For now, let's just assume a fixed-ratio gearbox.

During these tests we're not using the motor as a motor, but as a D.C. generator. This is the key. Later, when we want to wind the cable back onto the winch, we'll use the motor in the conventional sense - as a motor. D.C. motors have this nice bit of reciprocity, as you probably well know. From this point on I'm calling the D.C. motor a generator, because that is how I'm using it. During these tests no power is applied to its terminals. It's a generator. Period.

To set the cable tension at some value, we put a load (possibly resistive, for proof-of-concept perhaps) across the generator terminals to make the generator "fight back" with some given torque. The torque is a well-behaved function of the generator current (which we control via the load), not voltage, and by varying the load we can control the torque - and thereby the cable tension - with reasonable accuracy. The generator torque, in combination with the gear ratio and drum radius, sets the cable tension at some given value - provided the pull rate is constant.

If the pull rate is highly variable during a typical run, or tries to be steady but still contains a lot "tension noise," or whatever, then the generator RPM and generator current will vary accordingly. The objective is not to keep the generator RPM constant, by the way, but rather keep the generator current as constant as we need to maintain the tension at the setpoint. As the generator current largely determines the torque, and we thereby keep the current constant, we should expect a reasonably constant tension on the cable - in an ideal world. Consequently, the load will need to be continuously variable (or in the case of a digital system, 'continous' insofar as having a large number of discrete values thanks to the resolution of the system's DAC), and should be part of the control loop that aims to maintain the tension at some setpoint.

You get the general idea.

-E

One more thing: concerning the DAC's resolution - if you take this approach - you will need a multiplier on the DAC's output to scale the DAC's resolution so that your feedback loop will be able to compensate for variations in the cable tension at various setpoints. Let's say you're using an 8-bit DAC to control the value of the dynamic load. This will give you 256 discrete levels of load value. Let's say you also have 200 tension setpoints, 1 kg per step, together with the DAC's 256 levels, corresponds to a tension step size of slightly under 1 kg. At a 200 kg setpoint, one step amounts to a little under a +/-0.5% variation in the tension. But at the other end of the tension range, ie 1 kg, one step up corresponds to a +200% increase in tension, and a step in the other direction a change to no tension at all.

Without following the DAC's output with a multiplier, the feedback loop is useless at the lower end of the scale. In the DAC + multiplier scheme, you'll set things up to use the DAC to establish the load baseline for a given tension, with the multiplier assumed to be set at the middle of its range. No matter what the DAC's output, the multiplier scales the output so that the ratio of the load step size to the baseline setpoint is reasonably constant. Another way to say it is that the multiplier linearizes the load step size from the standpoint of the baseline setpoint such that the load step size is always remains a fixed percentage of tension setpoint itself. By following the DAC with a multiplier, the feedback loop sees the control problem as being fairly independent of any particular tension setpoint.

I trust I've made this as clear as mud.

-E

One quick, dirty, and simple approach when using a permanent-magnet D.C. motor/generator to fake your "mass" is this: Build a switch-selectable bank of load resistors whose values have been hand-calibrated to correspond to different desired cable tensions at some fixed generator RPM.

The drawback is that the generator RPM must be constant during the test, and it must be the same generator RPM against which the load resistors were calibrated to produce the correct tension. If you don't keep the pull rate constant and at the correct value, you might as well forget it because your tensions will be all over the place. If you can keep the pull rate fairly constant and if you have a reasonable number of discrete setpoints, this approach might be the quickest route. If the rate cannot be guaranteed, or if you have some huge number of tension setpoints (a 200kg range, but to three decimal places, for example), your system will necessarily be more complex and may involve some form of a digital control system and computer.

Btw, avoid stepper motors like the plague. Heck, avoid stepper motors and the plague! Steppers, insofar as they're useful for this application, ain't. Their torque curves highly non-linear - unless lots of torque ripple is what you're looking for. (Maybe that's why they're called steppers, yes?)

-E

As I often tend to circumnavigate the earth before realizing that the corner store is just down the street, let me finally post a simpler alternative: Using the same winch-drum, gearbox, and D.C. motor setup as before, drive the motor, as a motor, in the opposite direction as the actual shaft rotation due to the cable motion. But drive the motor with a programmable current source; not a voltage source. Same effect, fewer gizmos.

--E

Perhaps a clearer statement would be: "...drive the motor, as a motor, in the opposite sense as the actual shaft rotation..." ie, The motor shaft will still rotate in the direction that pays out the cable, but the polarity of the current is such that the motor shaft would rotate in the opposite direction if it were free to do so.

--E

Hydraulic couplers would work but require maintenance. What about a magnetic coupling like this.

The principal it that the rotating magnet induces a current in the coupling plate and that results in torque being induced on the coupling plate and hence the rope. It doesn't matter if the plate is rotating or not and the plate doesn't need to be magnet either just conductive so something like aluminium is a good choice because it's light.

Hi Masu,

Although these kinds of magneto-mechanical "brakes" are perfect for some applications, it is highly unlikely they can produce enough braking torque to simulate a 200 kg mass. The effect is rather weak, as brakes go. You could increase the torque by using a very large diameter disk and more magnets around its periphery, but the disk would need to be very large.

Btw, it does matter if the plate is rotating. Not only does it matter, but movement is absolutely critical to the whole effect!

No movement => no induced current in plate => no counter magnetic fields from induced current => no braking action.

-E

The current induced in the plate is proportion to the difference in rotational velocities. If the magnet is rotating and the plate is stationary then there would be induced currents in the plate and hence a torque induced on the plate. If both the plate and magnet were rotating at the same speed than there would be no current induced in the plate and no torque so no further rotational acceleration. The induced current and subsequent magnetic field is the driving force so it only important that the plate be conductive and not necessarily ferrous.

I havn't bothered to work out how much torque you could generate so you are probably correct about it not being strong enough, it was just an idea.

I did however remember working on some old tape drives, the ones with the vacuum columns. The idea was that the two reel motors were set up to supply a constant torque in but in opposite directions so the tape was kept at a constant tension. The capstan was the thing that did all the driving. As soon as the capstan moved the tape then the tension on the take up reel would drop and so the motor would turn the take up reel taking up the slack. The opposite would happen at the supply reel.

It dose however imply that this initial concept of a stationary DC motor would work. These were specially designed motors but if you over sized the motor you may be able to get away with it.

Thanks for all those good thoughts, & sorry you went 'round the world to the shop next door' :) But I get your point, tnx.

Trouble is.... (collective sigh)

Pull rate is not constant. Variable current source drive is interesting thought. Can a motor provide the torque required at no rpm though?

Seems to me I'm working against physics anyway, motors got torque cos of rotational inertia huh.

Which makes a motor running good, then we use some 'electromagnetic variable torque hysteresis slip clutch thingy' that I'm looking at on the web. This is simply controllable they say. Never heard of them before, & getting a headache trying to remember highschool physics - how many inch-lbs equates to 10, 20 or 100 kg ? Cable will spool on shaft, so dia will vary, so we need an external strain guage to maintain constant 'mass', or allow control of variable 'mass' during cycles.

interesting though, might even be getting somewhere :)

Umpteenth thought... If there IS a motor type that provides torque at 0 rpm & can do so all day then it would be nice and simple to use.

Driving it in some controllable way in the opposite direction of cable pull was always the intention, but surely big currents & big motors are involved in 100 kg simulation ?

Does the controllable slip clutch seem a better option ?

cheers :)

You wrote: "but surely big currents & big motors are involved in 100 kg simulation ?"

If you're thinking of a direct-drive setup, this is true. But you can also use a smaller motor connected to the winch-drum through a gearbox, as the gearbox will multiply the motor torque by the gear ratio.

-E

You wrote: "Does the controllable slip clutch seem a better option?"

For my part, I tend to think a slip-clutch may not give as fine a control over the tension as you'd like, and may have a longer response time (ie, larger built-in 'time constant') to boot. A slip-clutch may work for all I know, but if these either or both of these problems arise in practice, the control loop may be given a real run for its money - not to mention a nasty headache. As the only slip-clutch experience I have is the result of owning a Ford (failure-by-design) Fiesta years ago, my knowledge of intentional slip-clutches is somewhat limited.

Have you posed this question to the GearHeads? (CR4's Mechanical Engineering Section?)

-E

Look up "torque motor" on the internet. Any DC motor can be used as a variable torque motor with a current-mode amplifier. The relationship between current and torque is linear for the most part (things like thermal resistance coefficients of windings (copper?) and brush voltage drop (if brush-type DC motor is used) changes relationship slightly from linear).

Things to look out for are winding temperature, if motor is fan-cooled (not running doesn't provide air cooling), and HP rating of motor, which can be exceeded for short periods if winding insulation temperature limits are kept within limits. Also be aware that instantaneous measurements of temperature some point on the stationary part of the motor may not be the temperature of another, and that temperature measurement of the rotating part of the motor must be done "non-contact" and the same-as-above thing about temperature gradient applies to it.

A couple of thoughts...

1) A good commercial programmable current source will 'servo' to maintain the current at the setpoint. That it is "programmable" simply means you can conveniently set it to whatever value you need to produce the desired torque. Once set, the source will do everything in its power (pun intended) to maintain that current regardless of what the load (the motor) is doing - even if the motor is forced to rotate backward, as it does in this application.

2) Unlike internal combustion engines, for example, electric motors have torque at zero RPM. That's why IC engines need a starter motor and electric motors don't. A D.C. electric motor has maximum torque at zero RPM, in fact.

3) Winches have rotational inertia, as well, and the cable itself has linear inertia. The magnitude of the rotational inertia of a small motor connected through a gearbox will be multiplied by the gear ratio, but still, this may not actually be the most significant contributor to the overall tension induced in the cable by variable pull rates.

4) An 'electromagnetic variable torque hysteresis slip clutch' or any other kind of slip clutch has the same effect as the motor/gearbox/winch setup described earlier. Nor do any of these solutions compensate for the variable radus of the winch as the cable unwinds.

5) You've said nothing about what is actually doing the pulling. A better solution might involve both the puller and the pullee. Can you describe this? It might be helpful here.

6) What pull rates are you considering?

7) What is the cable mass per unit length? Alternatively, what is the cable material and diameter (from which we can calculate mass per unit length, and thereby cable inertia, which will also be a function of overall cable length)?

8) How many inch-lbs of torque equate to x kg? Think of the effective radius of the winch drum as a lever arm with the torque applied to the pivot point and the cable pulling the other end of the arm and in a direction perpendicular to it. That's all the cable sees; a lever arm. It doesn't actually 'know' about drums.

-E

Motors can provide their maximum torque while stalled. Some, with adequate size and cooling, can do this all year long (the stuff in a motor is no different than the stuff in a tranformer). You might also read up about eddy current dynamometers.

A simple way to go from a particular force to torque is to imagine a wheel with a radius of one unit. Then the torque and force are equal in magnitude. Use ratios from that to get to the real world sizes.

I notice that my first post provided "answers" that had already been provided by my comrades. Odd – in my stupor, I thought there were only a few prior posts. In any event, I am unclear re the need for motion of the cable vs the need for a simple static tension test. If the idea is to roll through 1000 feet of cable, testing it continuously (as you might with good quality rope), then that requires more equipment than to simply pull on the end of a 1 meter hunk of rope.

Static test:

To simply pull on the end of a piece of cable, a pneumatic cylinder would work well. A 100mm (4") cylinder could easily pull 2000 newtons without breathing hard. One could control the regulator electronically if one were so inclined (in a typical lab you just adjust the regulator by hand) and the simplest weigh (ho ho ho) to measure the force would be with a strain gauge at the rope attachment point.

In the same scenario, applying a static load with a DC motor would work just fine too – it just seems a bit more costly.

Dynamic test:

Here, having two dc motors (one on a supply reel, one on a takeup reel) would work well, with them pulling in opposition through suitable gearing. You could have any combination of speed and force either direction. In industries where materials are removed from reels under a desired tension, air motors are often used for tension control (and rewind, if necessary).

HOWEVER: I wonder if the "mass" specification is there for a reason. Perhaps in the actual application of this cable, a mass is accelerated by the cable perhaps at a rate equal to 2 or 3 G. Then it is yet another type of test.

As shown, the diagram implies that only the drum is moving. Sure, as long as there is differential rotation between the plate and magnets, you will have induced currents in the plate and, thereby, braking torque.

-E

Sorry Europium the diagram isn't that good. The idea was to rotate the magnet in the direction of the arrow that's on the shaft that is attached to the motor. Since the cable is attached to the drum and is stopping it from rotation then the tension on the rope would be in some way proportional to the speed the magnet was rotating. My maths isn't currently up to calculating the magnet strength, size of plate, gap, speed of rotation etc. but I remember doing it back in university as an exercise. When they use this principal in reverse as a break they can get some fairly big torque values so my guess is that it may be possible. The advantage of this it that there is no mechanical coupling and hence nothing that will wear out rapidly like in the winch, hydraulic and clutch systems. If you could buy an off the shelf magnetic breaking system and attach an appropriate electric motor it might work.

I believe the simplest way however would be to use a DC or universal motor with a current source. The torque needed would be roughly 2.0Kn x Drum Ǿ. The difficult bit would be finding a motor that could provide the torque and cope with the continuous locked rotor situation. Hopefully with a bit of searching both items would be off the shelf devices.

It would depend on which parts were the easiest to source.

Thanks.

Reading Guest's post carefully, my impression is that the drum is, in fact, moving during the tests. It is not stalled, per se, just moving slowly. In the solution I proposed the gearbox serves two purposes: Firstly, it serves as a 'torque transformer' that allows a smaller motor to be used than what would be required for a direct-drive setup; with it's attendant high cost, high currents, expensive high-current electronics, and the potential for lots of smoke. A smaller gearhead motor would very likely be cheaper than a large, gearless direct-drive motor, and because of the motor's comparatively modest current requirements, the motor-control electronics will be cheaper (and more likely to be available off-the-shelf by multiple vendors).

Secondly: although the winch-drum is moving slowly, the motor shaft itself would rotate at a higher RPM, thanks to the gear ratio. The advantage here has to do with the motor's electrical commutation: In a direct-drive setup, the motor torque will vary widely as the brushes bridge the gaps between commutator segments. Since the motor shaft is turning quite slowly in the direct-drive setup, there will be long dwell times (relatively speaking) where the motor torque might be quite unstable (also read highly non-linear) and difficult to compensate by the control loop. This "torque ripple" will still be present even if a gearbox is used, of course, but the duration of these torque instabilities will be far shorter, making life a lot easier on the control loop.

One really nice thing about your approach is the total absence of "torque ripple." If the effect were more pronounced and could be had in a smaller space, this would be the "torque transducer" of choice IMO.

-E

You are absolutely correct Europium about the torque ripple on the electric motors and it's something I hadn't thought of myself so good thinking. The magnetic coupling is starting to look more viable though. Is your understanding of the maths behind it better than mine? If so maybe you can see if it will actually work or do I need to go a search for my old university notes? That is if they still exist.

Yes, I can still do the analysis, and I might do this at some point when I have the time.

Btw, there's nothing that says only one large disc (plate) be used here. You could use a bunch of smaller discs, actually, all mounted on the same spindle. The magnets might consist of annular rings spaced at regular intervals along the interior length of a cylinder, which can itself rotate, such that the discs and the rings are interdigitated. The magnets could actually be mounted on (or intrinsic to) either the discs or the rings. Both poles of the magnetic field might exit only on one side of the ring or disk and, by moving the shaft along its axis (thereby moving the disks closer to or farther from the rings on the field side) an additional level of control could be made available. Just an idea, for what it's worth, in making a more compact version of your device...

-E

The nice thing about those tape drives (I've seen 'em too; the ones with the vacuum columns), is that both the puller and the pullee participate in the solution, not to mention that silly capstan!

-E

Just as an aside I remember inheriting a problematic tape drive from fellow engineers I worked with. It would work fine until it either went into high speed rewind or fast forward at which time it would pull the tape out of the columns and shut down. Five of us tried to fix it and in the end we decided to scraping it and use it for spare parts. Strange thing is that everything we took from it worked fine when used in other tape drives. It's the only thing I have worked on that I couldn't fix and it still bugs me.

First thing you need to keep in mind is that 200Kg is a mass not a force so you need to convert this to a force. On Earth gravity accelerates things at 9.8 ms^-1 so a 200Kg mass will exert a force of 1,962 Newtons. If you multiply this by the radius of the drum the cable is wound on it will give you the maximum torque you will need.

I doubt there will be anything that you can buy off the shelf that can what you wish. Probably the most practical way is to use a DC motor with some sort of current control unit that can drive the motor and give you the appropriate torque. The motor is however going to need to be capable of coping with that sort of current with a locked rotor and that sort of information needs to come from the manufacturer.

I do however believe its possible because as I mentioned earlier I have seen them used in this mode before

Sounds like a job for a hydraulic cylinder used as a linear motor. You can dail up any Mass (pressure) you need and hold it as long as you need. A strain guage will still be required as a feedback device. The cable could wrap around a capstan winch like device. Maybe two capstans drums, to be pulled apart for the testing sequence. The cable could then be advanced to the next section for testing without cutting or damaging it.

I agree completely to this solution. You can tune smoothly de hidraulic pressure, including differents functions. You will be able to regulate the test by force (with pressure) or measuring the strain.

Gabriel, Buenos Aires, Argentina

I agree with the hydraulic solution as well. I did some quick calculation to see if a pneumatic actuator could do the job but the pressures would be too high for it to be feasible.

The best choice using DC motors is an four quadrant controller. Machine can switching between generator or motor on the fly. Motor must be air or water cooled and may be manufactured by continuous torque at 0 RPM without burning. I prefer use and hydraulic motor with servos and little hydraulic central using electronic controlled pressure relief valves.

My 2 cents worth,

Why not use a conventional load cell? you get linear and continues reading which is easy to implement and interpret. Here's what we did some years ago while measuring cable tension in variable wind conditions. (bridge structure) What we did was very simple but worked flawlessly: The cable was routed through a pulley which was mounted on a steel bar of about 4' long, at the opposite side of the bar we had several resistive strain gauges and what we were actually measuring, was the bar flexing under the variable load. The biggest problem was to calibrate it which is not rocket science as you well know. Feel free to develop the idea if you like it,, or file it in the nearest garbage can.

Wangito

Guest, is the 'mass' being hoisted vertically? You also called it a 'weight', so I'm guessing you're wanting to simulate the overall cable tension resulting from lifting a weight through some vertical distance? This seems to be implied in light of your need to place a "constant' as well as 'variable' strain" on the cable section.

On the other hand, it is entirely possible you meant to say that 'constant' and 'variable' do not, in fact, occur at the same time. Is it possible, perhaps, that you want a fixed tension during one test, and a variable tension during another, different test, for example? Do you see the ambiguity in "as well as" (at least to me)? Sorry for seeming like a bonehead.

-E

Thanks to everybody tackling this :)

Europium - you are getting the gist of it all the best, tq.

Yes, the mass is being hoisted through a vertical distance of about a metre.

The hoister is a slowly rotating eccentric (stuck for words) cam ? Cable section under test is placed between eccentric pin and mass carrier. Guys have to run it for a while, change mass etc etc.

It would be really good if we could:

use a DC motor

trying to run

but either stalled or being pulled against its attempted run direction

except when cable relaxed; whereupon DC motor pulls it back at x kg

(x kg is programable to be either fixed (e.g 99kg) in test A

or vary controllably during cycles (from f.e.g. 87 to 99 kg or whatever at any points during a cycle)

I can see that a contollable current driven DC motor (max torque at orpm = bonus, heat is a problem i guess) with a strain guage under the motor mount to close a feedback loop would be cool. Hope to avoid gearboxes cos of drag, inertia.

I'll try to come up with a clear diagram, hope I can insert one here. I will try to register again, something wouldnt let me before :( (probably my stupidity)

Guest, you need to keep in mind that the force on the rope with a 200Kg mass at the end of it is roughly 1960 Newtons. Since you are using the metric system any motor you assess will have its torque measured in Newton Meters or multiples of them. Keep in mind that when you use SI units force is measured in Newtons. Kg is a measure of mass which is the ability of an object to resist a change in motion not force.

Stating that a rope has a breaking strain of 200Kg is actually incorrect even though most people understand what you mean. You should state that the breaking strain of the rope is roughly 1.96KN but I can't give you an exact figure. I know that I am being pedantic but the force that a 200Kg mass will exert on a rope will vary slightly depending where you are on earth as well as your altitude and the time it is. For example it will be slightly greater at the North and South poles than it is at the equator. While these variations are negligible and would make no difference to your application they are measurable.

It's important to understand the correct units to use because if you don't then confusion can creep in. As I have said in another thread that's how a billion dollar Mars explorer probe crashed. It turned out that when a variable was passed from one subsystem to another one programmer was working in imperial units while the other was using metric. The result was the probe crashed and a few thousand man years of work went down the toilet. Just imagine what that sort of mistake could do to your career prospects.

Here are some links:

http://www.machinedesign.com/ASP/viewSelectedArticle.asp?strArticleId=55643

Application discussion...

http://pergatory.mit.edu/2.007/Resources/calculations/motorcalc/motorcalc.pdf

More basics...

http://www.electricmotors.machinedesign.com/guiEdits/Content/bdeee3/bdeee3_4.aspx

A specific type of torque motor with applications...

http://www.orientalmotor.com/products/pdfs/A_OM/AcTrqAll.pdf

Kollmorgen torque motors (applications link @ bottom of page)...

http://www.motionvillage.com/products/motors/torquers/

Of course, you can always search:

www.globalspec.com

-E

Dunno if this helps or makes things worse !

Cheers :)

Hi Max2tall,

The diagram helps a lot. Thanks!

May I ask you a few more questions (this question itself being one of them? )?

1) What is the radius of the puller?

2) How fast does the puller turn?

3) How tightly must you maintain the cable tension? How much variation in tension (expressed, perhaps, as a percentage of 'mass') can you tolerate? (As an example, when simulating a 1 kg mass, is +/-10g of mass error is acceptable, ie, +/-1%?)

4) Must the solution take the form of an 'active' system (with a motor, controller, and other hardware)? Can it have a 'passive' solution? A purely mechanical one, perhaps, not involving a motor, slip-clutch, and so forth? If so, I have a much simpler (and far less expensive) solution in mind. And no, it doesn't involve springs!

If the answer to #4 is 'yes', the specific answers to questions 1-3, are critical to this solution being feasible.

And finally,

5) Will this be a commercial product; possibly one which must have some degree of "shelf appeal?"

Thanks!

-E

Tnx for your questions Europium,

1) puller varies with different cables, but r = 0.2 to 0.6 m say

2) puller rpm v slow, varies between 10 & 20 rpm

3) 1% should be ok, less better if not too hard to do :)

4) This 'active' solution (if it exists) is to replace an existing pasive one :(

5) Has to be pretty & not leak bloody H fluid everywhere...

Mevel123: Wouldn't the drive controller actually be trying to drive the motor at 1g in the opposite direction of cable pull in order to simulate a weight being hoisted against gravity? My little brain can appreciate this, but I get lost when I think the current has to be limited to simulate a particular weight- it's like I'm trying to get the motor to do two opposing things at once :( Maybe I am, it just seemed like a simple idea when I started ! Now its turning into a brain teaser :p

Those torque motors look really promising, seem to have all the right factors built in to them.

The diameters would allow a nice drum connection also.

let me know if I'm on the right track at all with 'weight' vs torque please:

if say a 4" (100mm) dia drum is used, with 100 kg dangling from it, does this mean (r = 50mm), that 5kg-m of torque is applied to the shaft, which = 49Nm ?

If so, a torque motor providing 49Nm to a 100mm dia drum would pull a cable the same as if it was a 100kg lump of steel or bag of marbles ?

(i'm too scared to call it a mass or a weight anymore )

That's correct. The 100Kg mass and yes its OK to call it a mass, exerts a force of 980N or pretty close to it. and 980N x 0.05m = 49Nm. I can however remember back in my laboratory days we carried out a series of experiments to find out what the local gravitational acceleration was and surprisingly found it to be 9.786ms^-2 which is not what we were taught. For engineering purposed though 10ms^-2 is close enough and give you a bit of lee way.

You wrote: "(i'm too scared to call it a mass or a weight anymore.)"

Mass is what's important when you throw your weight around.

-E

Max2tall wrote: "(max torque at orpm = bonus, heat is a problem i guess)"

There is no 'rule' which says 'a stalled motor must overheat'. If the motor must operate in a stalled condition, then it must be appropriately sized to account for this condtion. The popular impression that stalled motors overheat may in large part derive from seeing stalled motors overheat because a stall condition was not part of the motor's normal operation and therefore wasn't accounted for. During stall the motor is producing maximum torque (ie, consuming maximum current which, in conjunction with the winding resistance, dissipates maximum power). If the application requires the motor to be stalled, then the motor's power dissipation in under these condtions must be taken into account when sizing the motor.

Larger motors may incorporate some kind internal fan which is attached to the shaft and which helps to keep the motor cool during normal operation. If the motor depends on this fan turning in order to perform within its rated temperature range, then in a stalled condition it will be necessary to apply external forced-air cooling. But if the motor does not incorporate an internal fan, there are other ways to cool the motor.

One way is to heat-sink the motor and cool it using forced air or liquid cooling. I've seen compact, ultra-high-performance servomotors which had to be liquid-cooled or else they would incinerate (I kid you not!). One such motor powered the base drive in a 'pick-and-place' robot arm which could lift a 2000 kg load, rotate 180 degrees about the vertical axis, put down the load, and move back to its original position in just under three seconds. Neither was any gearing involved; the motor was driving the robot base joint directly!

It's all about choosing the appropriately-sized motor for the job; one which can comfortably handle the power dissipation in a (possibly-indefinite) stall condition.

-E

Problem is that you are hung up on "elecrical" motors. A pneumatic cylinder should do this really easily. Better than hydraulic because the air will act as a "spring". This would replace your mass with air, pressurized air is easy to get and cheap, a pneumatic cylinder would be cheaper than an electric motor, and the results can be read right off a simple gauge. You can fancy this system up with digital valves and readouts if you want.

If you absolutely NEED to use electricity, then maybe some sort of electromagnet might work. Just tug on magnetized plates until they separate. You can calibrate the "breaking" tension of an electromagnet pretty accurately by controlling the current. Set it up so that you minimize the effects of the lash. You want to wind it right so that the back emf will slow down the applied current enough to keep it from heating up. (That would mean ac...)

Not really "hung up" on electrical motors, per se. For instance, I've privately offered max2tall the use of my horse.

-E

I have read over most of the posts, I hope I am not over simplifying this, however a standard DC Shunt wound motor combined with a resonably performing DC 4 quadrant drive is capable of producing 0 through rated torque, in either direction(aiding or opposing shaft rotation) throughout motor's speed range including 0 speed. (Add blower to motor)
The drive could be controlled as a speed regulator (Requesting 0 speed) Then the torque limits (of the drive) need to be actively controlled to produce desired tension. If cable snaps- drive will imeadiately come out of torque limit and begin regulating 0 speed.

Typical drive / motor should easily be able to produce torque in 1% increments. Even finer resolution is achievable using better than standard drive hardware. I recommend Rockwell Automation or Reliance Electric.

It is true that ripple is evident in DC motor current, however resulting torque deviations are minimal compared to rated.

I just noticed how this forum slots posts in amongst earlier ones if you reply to it, which means you can't just go to the end & see how things are going. So i'm repeating myself here to make life easier I think...

Tnx for your questions Europium,

1) puller varies with different cables, but r = 0.2 to 0.6 m say

2) puller rpm v slow, varies between 10 & 20 rpm

3) 1% should be ok, less better if not too hard to do :)

4) This 'active' solution (if it exists) is to replace an existing pasive one :(

5) Has to be pretty & not leak bloody H fluid everywhere...

Mevel123: Wouldn't the drive controller actually be trying to drive the motor at 1g in the opposite direction of cable pull in order to simulate a weight being hoisted against gravity? My little brain can appreciate this, but I get lost when I think the current has to be limited to simulate a particular weight- it's like I'm trying to get the motor to do two opposing things at once :( Maybe I am, it just seemed like a simple idea when I started ! Now its turning into a brain teaser :p

DC motor torque is proportional to armature current on fixed field or permenant magnet motors.

The drive will not actually be trying to do 2 things, instead it will simply be providing current (torque) in a direction.

If the motor is uncouplued the motor would accelerate (same as g force) at a constant rate.

If the motor is locked it will not accelerate, but the force can still be measured.

If a load of greater force is applied to the motor shaft (in opposite direction) the motor will accelerate in the opposite direction

Under all 3 cases the drive (properly tuned) will simply be putting torque current through the motor in the same direction-ignoring motor shaft movement/direction.

This application is very common for machines that unwind large rolls of material, the motor is being used as a brake to hold constant tension.

Tnx Mevel123, you make it sound a lot simpler !

So where can I find something to do it ?

Am I looking to control one of these things:

http://www.motionvillage.com/products/product_literature/product_brochure/pdfs/KOL1154.pdf

or some other type of motor for my simulations.

Is it now a matter of a controller that limits current to a setable value to set the 'mass' and control the motor speed to mimic an accelleration of 1g. And then attatch our cable to it & pull away against the run direction ?

Would like to control the controller via a laptop... doable ?

I do not believe you need a special purpose torque motor, unless I am underestimating your resolution requirements.

I am picturing a standard digital DC drive, DC motor. A Logix controller would be the 'middle man', between the human operator and the drive. This allows the operator-using Laptop PC- to enter desired setpoints (in your desired units-Kg) The controller will calculate the current to produce the torque, and command it to the drive.

The entire system would also require a speed feedback device on motor shaft (encoder/resolver) to prevent 'run aways' if cable snaps. All of this is a very standard setup for many applications.

I have recommended Rockwell Automation in the past (there is a link posted above). There are a few other vendors that I have not personally used, but in the name of fair play, I believe Eurotherm, and Seimens have similar DC drive offerings. All though they both use Rockwell Logix controllers as the 'middle man'

mevel123 is right on about the motor. max2tall, to get a good 'feel' of the 'motor doing two things at once' consider taking a small DC motor - such as might be found in a motorized toy - and apply battery power. Stop the shaft with your fingers and you can feel it still trying. Turn the shaft backward, and it still fights you. Same idea, except that the motor's torque, its 'fighting back', as it were, is maintained at a precise value by the control loop.

-E

Eur- Your test using DC motor and battery is accurate, however you left out 1 point.

If someone attempts the test, the motor may overheat very quickly. Since the battery can not actively regulate motor current, the current will approach short circuit amps as the motor is stalled- they will also notice that the torque is increasing as they stall the motor.

This is not the case with DC drives since they control current (torque) regardless of rotation.

Sorry for the tweak, but I just pictured burnt fingers

From childhood onward I've played with more motors of this type than I care to count. And yes, leaving the motor connected to a fresh battery will cause the motor to heat up. Maybe not fry the fingers, but it will heat up. To date I've never burned my fingers on one of these toy motors. Perhaps this was my motive in emphasizing the "toy" aspect?

Of course the motor is not in a control loop but is driven by a battery - a voltage source - meaning the torque is not regulated. And of course the torque will increase as the motor approaches stall. Yeah? So? (Max2tall has been reading my posts, too, by the way. Like the part where the torque is a function of current, not voltage, and where I explicitly point out that a programmable constant-current source, not a voltage source be used to produce a constant torque?)

Back to the experiment: You hook one of these two-dollar motors to a AA, C, or a D battery (all 1.5v), noticing 1) that you feel the stalled shaft wanting to rotate and 2) that you can rotate the shaft against its will whilst noticing that it still fights you with pretty much the same strength. After you've fiddled with the thing for maybe ten seconds or so, you're satisified and even notice that your fingers are still warmer than the motor itself. You also notice that the motor torque does not vary appreciably between stall and when you momentary twirl the shaft backward at perhaps 50 RPM, max (if you're really good). Nor do you much notice that the torque is a function of speed in the normal direction, as the armature's angular momentum dominates when you pinch the shaft to a stop. What really grabs your attention is the feel of the motor's cogging action as you turn the shaft manually between your index finger and thumb. This effect alone swamps variations in torque reduction (due to back-EMF) by at least ten to one. More like 50 to 1 at the speeds discussed here.

And even if you should notice the torque vary, you don't care because it is immaterial for the purposes of this experiment! The point of this experiment is not to gauge how much torque there is at different speeds, but that there is a torque at all. Not only that, but the fact that you can actually rotate the shaft in a direction opposite that torque (an unusual concept for those having little or no motor experience, btw. Like the concept that an airplane can actually fly backward in a sufficiently strong headwind. The onlookers don't care one whit about how fast the plane is flying backward. Only that it is).

So, the experiment is a success - all without your spending one penny on a motor controller that regulates speed, torque, acceleration, position, or any other parameter. Nor did you burn your fingers. During your experiment you felt the motor fight you when it was stalled, and you noticed it still fighting you when you rotated the shaft backward. You walk away from the experiment with a solid, qualitative feel for what's going in the Real Thing.

Which was the whole point.

-e

You are completely right-I only pointed out the difference so others would not be mis-lead into thinking that DC motors over heat when operated at 0 speed. It appears from a few posts here, and my life experiences that it is not widely known that DC motors work fine at 0 speed and full rated torque. Your experiment creates excessive (when compared to rated) torque at 0 speed, and this could damage the motor.

I do appreciate you helping me clarify my original point. I did not mean to insinuate that your experiment was dangerous or invalid, rather I only wanted to point out where it was 'different' from the DC drive system I was proposing.

Yes, my tweak was not required if people read all of your previous posts, I apologize.

I decided to have a look at the problem in a little more detail as shown in your diagram. In an attempt to make it a little clearer I have redrawn it here.

Now Ø (see first diagram) is continuously changing according to the rotations such that

Ø = θt = RPM x 2π ÷ 60 radians per second and therefore

θ = RPM x π/30 Equation 1

so we can now say that the vertical displacement d can be expressed as

d = r sin Ø and by substituting we get

d = r sin θt Equation 2

The next part refers back to these equations

and the instantaneous vertical speed v is the rate of change of vertical displacement and the derivation of this is shown in Equation3

The instantaneous acceleration a is now derived in Equation4

Now since the weight w is being pulled by gravity and the rope is being accelerated by the cam we can say the tension on the rope K is the sum of a and g as in Equation5

To find when the rope has the minimum and minimum tension in the rope we need to know what the maximum and minimum accelerations that the cam generates. This will happen when the rate of change of acceleration q is zero so we need to differentiate the acceleration again, as in Equation6

Therefore q will be 0 when θt is equal to π/2 and 3π/2 and using Equation1 t for our 20RPM example then this would occur when t was ¾ or 0.75 seconds and 9/4 or 2.25 seconds respectively.

Now in summary we know

t = 0.75 & 1.25 seconds

θ = 20 x π/30 ≈ 2.094

m = 2,000Kg

r = 0.6m = 600mm

g = 9.81ms^-2

and substituting into our tension equation

K = m(g – rθ2sinθt)

we find the two rope tensions K_min and K_max as follows

K_max = 2,000Kg(9.81 - 0.6 x 2.094²sin(2.094 x 2.25))

K_max = 2,000(9.81- 2.631 sin2.712

K_max = 2,000Kg(9.81 + 2.631)

K_max = 2,000Kg x 12.441ms^-2

K_max = 24.881Kn

K_min = 2,000Kg(9.81 - 0.6 x 2.094²sin(2.094 x 0.75))

K_min = 2,000Kg(9.81 – 2.631 sin1.571)

K_min = 2,000Kg(9.81 – 2.631)

K_{min =} 14.358Kn

Now I know all this is convoluted but I wanted to make sure you understood the derivation of all the equations and use them yourself. Something to note is that when the maximum and minimum accelerations are experienced the value of sinθt is 1 and -1 which make solving the force equation for maximum and minimum tension easy.

Now comes the bad part. If you replace the mass with a motor or device of some kind that supplies a constant torque you will loose the load cycling. The reason is that the motor will reel in or out the appropriate amount of cable to keep the tension on the cable constant. Remember the vacuum column tape drives I mentioned earlier. That's exactly how they work and when you pull on the tape they just reel out more tape.

This means if you want this system to work you will need to replace the cam with a fixed mounting for the rope and then build a control system that is capable of supplying a varying amount of torque and is capable of ramping it up and down over a period of between 3 and 6 seconds. This makes the whole thing a hell of a lot more complicated and consequently more expensive because you are no longer talking of a constant current source.

From what I have worked out here I think the least expensive, easiest to build, maintain and operate system would be the pneumatic cylinder. It would need some sort of pressure regulator that could be cycled so you still have the varying loads required. Sorry to put a dampener on the original idea but that's what engineering is all about. Work out if it will work on paper before you build it.

All is not lost Masu.

max2tall, your existing mechanism looks to be putting the cable through continuously-varying tension above and below a baseline established by the selected weight. In your current setup, at no time will the cable experience a constant tension and so the constant-current approach won't work. Instead of carefully considering your diagram and evaluating what your mechanism actually does, I stuck to answering your original question. Consequently, a constant-current (ie, constant-torque) solution is the wrong one. You can simulate the mass PLUS the sinusoidally-varying tension by using either a pneumatic cylinder or a motor-driven drum. Each has its advantages.

One very important modification I'd make to the current design is to eliminate the rotating crank altogether. You want varying tension, true, but you don't actually need motion to get it. What is the objective? Seems to me the objective is to apply continuously-varying amounts of tension on a cable. You don't need motion for this, so toss the crank entirely. Instead, anchor that end of the cable in a stationary position. It doesn't have to move at all. As no motion is involved (other than due to cable stretching), you don't have to be concerned about inertia, momentum, frictional losses, and so forth. How long does 1 metre of cable stretch, anyway?

If you go with a motorized drum, one thing to consider when driving the drum with an "intelligent" controller is that you can program an arbitrary number of different torque profiles. You may wish to apply a constant tension in one case, and a varying tension in another where the variations are about a baseline, just like you're doing now. But you also have the option of changing the variation from a sinusoid to a triangular tension profile, or even of gradually changing the amplitude of the the waveshape as the test progresses, or you might even wish to 'shock' the cable repeatedly by applying a sudden torque, then relaxing, then torque, etc. Within the capabilities of the mechanism, what you can actually do with the setup is pretty much up to your imagination.

As your final solution will very likely involve a computer, you will have the option of saving test setups in a file, designing test profiles, saving test data, auto-calibrating the system, and so forth.

-E

Also...

To continously measure the tension on the cable, mount a strain gauge at the fixed end and attach the cable to that. Incorporate the strain gauge output into the control loop as a direct measurement of the tension, rather than trying to derive the tension value from other system parameters such as applied torque and drum radius. The computer can also log the tension reading as part of the test results.

If the cable snaps, you don't want the thing to run away. This was raised as a real possibility earlier. A simple approach might be to eliminate the drum, as well, and substitue a lever arm having one end attached to the motor shaft and the other to the cable under test. Mount a limit switch behind the lever arm (in the direction it would rotate should the cable snap) to shut down the motor amplifier should the arm attempt to run away.

I'd include a diagram, but the only graphics software I have currently produces photorealistic 3-dimensional raytraced stills and animations, but nothing to draw even a simple block diagram! Let me see if I can find something to my liking on the web.

Meanwhile, these might constitute the major components of a system:

motor

lever arm

strain gauge

data-acquisition board (installed in computer) with strain-gauge-suitable inputs

motion controller (Rockwell, et al)

servo amplifier (output goes to motor)

limit switch

computer

software

-E

Europium I use DeltaCad^® to do the drawings I post. You can down load a demo version from here

http://www.deltacad.com/

It's a drafting package so it can't do fancy things like 3D modeling but what do you expect for $40. It can't save an image directly but you can cut an image from it and paste it into something like Microsoft Paint then save it as an image. It takes about 2 hours to do the tutorial and become fairly proficient, even less if you know your way round a drafting board.

Thanks! I'll also see what's available for linux. (I have a dual-boot machine with Windows and linux, and I use both but prefer linux).

This looks to be something I might have a lot of use for.

Thanks again!

-e

Tnx again gentlemen, am emailing some of those manufacturers now to see what they can provide, nice to know its doable, cheers.

Europium: you may be interested in:

http://www.ribbonsoft.com/qcad.html

seems a good cheap (free trial is time limited) linux etc cad package.

Very nice, max! Thanks!

-e

This is really a copy of the mechanism in the tape drives that I mentioned earlier. It would be reasonably easy to construct using a couple of constant torque motors and a stepper motor all controlled by a Programmable Logic Controller PLC.

The idea is that the two constant torque motors have the rope spooled on them and the constant torque (shown in RED) controls the tension on the rope. The stepper motor could then move in either direction (shown in GREEN) to move the tape from the supply reel to the take up reel. By reversing the direction of the stepper motor at regular intervals and controlling the acceleration you could mimic the test with the weight and the cam at regular intervals along the entire length of the rope. It wouldn't be too difficult to achieve and most of the technology would be off the shelf equipment. The other advantage is that the motors would not remain stalled indefinitely so could be reduced in size.