Inertial Addendum: Does Inertia Affect Torque? | ELS Part 25
ฝัง
- เผยแพร่เมื่อ 28 พ.ค. 2021
- You all jumped in the comments last week to tell me that inertia was affecting my torque measurements. Was it? Let's do some experiments and find out.
Tools used in this video:
*This site contains affiliate links for which I may be compensated
500N Push Pull Force Gauge (Amazon*): amzn.to/3hwzXjn
95 Paracord Type 1 (Amazon*): amzn.to/3eRJAat
Wilton 6-1/2" Tradesman Vise (Amazon*): amzn.to/33fqPXz
Rigol DS1054Z Oscilloscope (Amazon*): amzn.to/2JlhlhR
Rigol DP832 Triple-Output Power Supply (Amazon*): amzn.to/2FHtSfz
Neiko Digital Non-Contact Tachometer (Amazon*): amzn.to/3foeKpc
Shars Ultra Precision 1-2-3 Blocks (pair) (eBay*): ebay.to/2RxwqWc
Bondhus Metric Hex Key Set (Amazon*): amzn.to/3eqZzwb
Bondhus SAE Hex Key Set (Amazon*): amzn.to/3h9bJLZ
Mean Well 48VDC 350W Power Supply (Amazon*): amzn.to/3tUZpS4
Raw Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 3.0 License
creativecommons.org/licenses/b...
Some accusations in here of confirmation bias. Contrary to popular belief, I'm human, so that's a real possibility. There are some details that didn't make it into the video that might add some context. When I first tested at 500RPM with the plastic pulley, the values I got were all over the place, leading me to believe something weird was happening. So I moved on and tested at 1000 and 1500RPM. I ran multiple test runs at the higher speeds and they were consistent, and as expected, so I went back to the 500RPM test to investigate further. The values were still inconsistent, but some of the values were in the range I expected, and I identified a possible cause for the inconsistency (melting plastic). That's enough to say that the hypothesis is plausible, but probably not enough to call it proven. The next step would be to design a new experiment that doesn't suffer the same issues and repeat all of the tests. If you decide to do that, please post your results so we can all benefit.
machine a similar aluminum disk but bore out most of the center so the inertial mass changes.
And that is why I like to do things live on my channel rather than pretend everything just worked like some other channels. You don't always get the result you expect, and there is no point pretending that life always happens the way it should. In my mind it makes for better viewing. Sometimes the learning lesson is not in the success but rather in the failure.
@@DodgyBrothersEngineering Most times it is in failure
Yep, inertia only effects acceleration (both positive and negative accelerations) and has virtually no effect on steady state continuous conditions. This is the same thing with Automotive Dynamometers (Dyno's) in that there are Inertia type and Brake type and Inertia types can only be used in acceleration tests (unless they have an add-on braking module) where as brake type are used for steady state output (can do acceleration tests also).
As soon as I saw you pull out the printed pulley, the very first thought that came to mind was the friction heat melting issue that you ultimately had.
On a side note, since you are doing basically a steady state torque test, to really show the effects (or lack there of) of inertia, measuring motor current (amps) for a given force would probably be a better test since motor torque is proportional to its supply current. If it were me, I would make two pulleys of same material and dimensions just with one hollowed out to lower its mass / inertia and use the same number or wraps of the string and pull to a specific force on the scale and measure the current at the motor for each pulley. If inertia was giving any effect, you would see a noticeable difference in motor current for the same scale force and RPM between the two pulleys.
even if the ineria was an issue in your previous test you used the same pulley for every test.
Agreed. Normalizing data across the motors keeping variables to a minimum getting data that is consistent values relative to one another is fine as long as you know what that data represents.
This is really key. Nothing is perfect. The goal is to get what you need to accomplish what you're trying to accomplish.
It is correct if the inertia remains constant. However the moment of inertia changes with rpm.
Therefore the argument works when comparing different motors at the same rpm, but not for a single motor across diffent rpms.
Also agree. But I'm still not convinced about the inertia and the energy of the rotational force/energy.
Just imagine a flywheel, spinning at 1000 rpm.
By this method the metal one can be slowed nicely with almost no issue and the rotational energy can be released safely.
But the Platic one has it's own issue. Even if you pull very carefully, when the plastic melts to the wire you have a VERY abrupt stop. Even when there is not much rotational energy in the plastic wheel.
You will see a spike when the plasic wheel stops.
Additionally I can not think of a method that will confirm or deny this. Nor of a method that could measure this reliable since even the drive shaft is basically a little flywheel.
Ohh just thought of it... may be you can take the measurement when the speed is constantly like 10% of the target rpm.
This would be really hard to archive but when 1000 rpm is target and you could increase the resistance down to constant 900 rpm but not stopped. You could measure the work of the motor since the speed of the flywheel is constant.
No rotational energy is added or removed.
But I'm guessing this would be very hard to archive.
you should try warping the pulley in aluminum tape so its has about the same friction as an aluminum pulley
or, even better, wrap both pulleys in the same alu tape, so they are exactly the same.
I would have turned an aluminum band and pressed that on to a PETG hub.
I really appreciate you investing the time to put together this follow up video, thank you! It shows serious dedication to your viewers and to science.
First off, I wanted to say that I really enjoy your videos! Thanks for making them!
I think the inertia argument is because the force gauge is set to record the maximum force, which could be due to the deceleration of the system instead of the steady state friction you are adding with the cord. If the deceleration force is higher than the steady state force, then a high inertia pulley would impart a higher force than a lower inertia pulley (assuming the same deceleration time for both). An interesting experiment would be to add some ridiculous amount of inertia to the aluminum pulley and re-run the test. (Maybe couple a 25lb weight to it?)
Also I wanted to add a that this feels like an interactive version of Mythbusters, which is awesome! Keep up the good work!
Oh man, James, I really appreciate your dedication to your viewers and following up to the comments. This really feels like an argument that car enthusiasts have about comparing car dyno numbers. People really need to remember that this is a relative comparison between test subjects within the a controlled environment. Love your work, at least this series about comparing motors have been most informative for me. Looking forward to assembling the kit I have received from you!
James you should consider repeating the test with a Steel pulley of identical dimensions. This would reduce the number of variables down to just pulley inertia. The friction coefficient of a steel pulley system will be so indistinguishably similar to the aluminum pulley system that you’ll be able to use the same number of wraps per test per RPM. You also won’t have to worry about pulley deformation and melting. 👍🏼
Hollowing out the aluminum would reduce that variable purely to mass as well
I think hollowing out the pulley is probably the best apples-to-apples solution. However, judging from the viewership on these videos, there doesn't seem to be a ton of interest.
@@Clough42 yeah, and honestly I'm satisfied. The numbers are close enough.
@@Clough42 damn, i'd actually be interested in the results.
It's an interesting problem with two opposing intuitive solutions, could be great for a HS physics class or something.
I think you could avoid any friction coefficient differences in material and preserve the reference piece if you just add mass.
Super glue several round pieces of heavier metal to the face like when trying to mill tough to hold work items. Or even drill and tap a few holes and screw it. You could potentially have several different mass values just by adding and subtracting the numbers of aditional plates. Just keep the diameter smaller than the reference so it is obvious if a string end falls to the different material.
Great video build series, I'm learnimg a ton.
Great video James! This definitely takes TH-cam DIY to a whole new level. Thanks for sharing!
thank you for your careful work.
"Lets find out" is something I very much like to hear on this channel. The only other channel where that phrase rivets my attention is Applied Science. Shop projects AND shop science? Please and thank you!
Someone I looked up to would say "you're always correct if you don't verify your work". To me, engineering is figuring out what you don't know and finding a way to test it. Proving myself wrong is a huge victory ... Unless I have to wait for replacement parts to arrive. 😅
Your channel is a pleasure to watch! I'm thankful that I can absorb your teaching without being interrupted by midroll ads, it helps the learning a lot
Firefox + AdBlock Plus = no ads on any YT
@@Watchyn_Yarwood I usually watch on the xbox app plugged into my tv
What's for tea tonight, Oh I'm eating my hat. I was a commenter on the inertia. Great test and confirmation of the underlying physics.
Fellow enginerd here,
Just wanted to start by saying thank you! Your videos are always a pleasure and very well made. I've been doing various stepper motor projects lately and you've done the best job by far in showing the differences in the various components that I've seen.
I wanted to share a couple of thoughts regarding the friction and inertia stuff, adding my 2 cents to the long list of comments that you already have.
1) in your case of wrapping the cord around the drum like that, you have a condition that is evaluated by the expression T1/T2=e^μ*θ. T1 is the force gauge, you are T2, the constant e, and dynamic friction coef. μ. This expression shows that the amount of wraps matter and will influence T2 required to make the stall, so not exactly like the block on the inclined plane example. I don't want to get into a super long winded explanation, so I'll say there are many proofs on the internet for this equation relating to belt equations, winches, capstans, etc.
2) The equation for Torque is T=I*α where I is the mass moment of inertia (a function of both mass and radius) and α is angular acceleration. When you stall the motor, there is certainly an acceleration at present since you went from some speed "x" to 0. Therefore, the stored energy in all the rotating components including the rotor and the drum will generate a torque (force) in slowing them down proportional to the acceleration. Having said that, as long as it is consistent for all the motors, it's a fair test.
I really enjoyed the video. Thank you!
touche
Thanks for this retesting video to confirm test setup is valid. I think this video also demonstrates that your actions needs to be consistent to produce meaning results.
hey james! thanks for the awesome videos.
you should give post processing 3d printed parts in the lathe a shot!
i tired it on PLA with a sharp dcgt insert made for aluminium and got an astonishingly smooth and shiny surface finish
Conclusive results from a simple and effective experiment.
You could spin the wheel up and remove power, letting it free spin. If you stopped the free spinning wheel using the same method, your force gauge would read the opposing force imparted by friction against the rope. Based on that it seems intuitive that this would be added to the force being provided by the motor during your previous experiments.
I doubt I'll ever find myself applying this knowledge in the real world. But it's fun and fascinating to learn the science behind the best practices we use in our hobby/work.
When the video started, I was hoping you were going to use the aluminum pulley for both tests, altering the mass by drilling lightening holes or screwing a steel ballast to the side of it. (As you noted. Would ensure an apples/apples comparison).
Very interesting. I sold motion equipment for years and the only consideration for inertia was break away torque. Once break away is achieved the only other issues were stopping or holding. (Example: Overhead crane.) I am comfortable of your theory and testing. Great job!
To avoid variation on tugging the rope, you could use a weight to make it constant. I don't think it's necessary, but here is the setup:
Run the rope horizontally to a pulley and have a little bit of rope hanging down.
You attach a small weight to the loose end of the rope.
At the start, you hold the weight up a centimeter or two.
Then just release the weight.
Fantastic video James as always! Based on my understanding of all things relative I personally think the only thing that would change depending on the mass of the drum is the time it takes from application of force to realize a stall condition. As has already been mentioned flywheels store kinetic energy. On a side note, USB to Serial converters always seem to be super fickle in my experience, sometimes the $10.00 one works way better than the $200.00 one… at least that’s what I have run into. Keep on keeping on sir! great content!
add a chunk of steel to the side of the 'aluminum' pully, that way your other variables stay the same, only the mass and subsequent inertia changes, not friction
Hi James, Good bit of education. I think there was an easier way. Just Superglue the extra aluminium disk on to the first one - keeps things more similar. But, you got there despite the extra 'stiction'. My thinking was the inertia did not matter because you were measuring the stall torque. The spool disk was no longer rotating and therefore not applying any extra force. BobUK
The inertia could be a problem: let say as a thought experiment, the inertia was infinite. when the motor stops, because it's overloaded, the wheel would continue to spin. You could then pull harder on the string, as hard as you can and the wheel wouldn't stop.This makes the frictional force larger, and then also increases the reading on the scale.
If however, you stop pulling immediately when the motor stops producing torque, the force reading will be accurate.
wicked video. My conclusion was also no inertial force but from a different position which worked well with the video. Stall occurs when static holding friction applied to drum when the servo can no longer track or apply torque. The inertia is irrelevant as its a static problem but further it is a dwell in force and not a change in amplitude. Glazing or caming (drum break force multiplication) both occur in the video is a change in coefficients of friction which again as a static holding friction problem.
Great video. Wrapping the petg with capton tape taking note of the direction of wrap so the seam doesn't lift up might have provided better results. Overall you got the date needed anyway.
Solid. in the noise. agreed.
I really hope you do a full review of the servo including software and tuning, i can't seem to find any in english, only in german.
Thanks for making this video. I was one of the people that made a comment on the last video. From my perspective, when the motor shuts off it looks like the rope bites and stops the system. As you showed today, this is not the case at all. The electronics are cutting out and the force on the meter should not change during the interval between motor shutdown and the system coming to a stop.
By the way, I would love to see some more of your videos on developing software. It is an area that I really can't find anyone on TH-cam with a lot of experience, specifically dealing in professional development of real time control systems.
Thanks! I'm glad it was helpful. On the software front, my expertise is in cloud software at scale, which is just about as far from real-time control as you can get. :) I do this stuff as a hobby--in part because my day job is so intangible. Making something that sits there physically on the bench at the end of the day is very satisfying.
Yesssss, new millllll 🤗.
Great video. Perhaps you could have wrapped the plastic wheel with some aluminum tape to make the u coefficient more similar, and reduce the melting. Loved the demonstration. Ciao, Marco.
I can only think of 2 things related to inertia:
If the plastic pulley bonds to the cord instead of the coupling being friction it would be a mechanical bond, so the inertial energy would be 'dumped' into the load sensor making the reading spike at the end.
Higher inertia might make the motor control loop more stable so a potentially higher peak torque, while less inertia would be more venerable to perturbations affecting the control loop making the motor controller give up sooner, but with the load being slowly and steadily applied this shouldn't change unless the controller is 'really' unstable.
Very good insight and effort. I'm still thinking about the test and the theory behind it. One thing should be took in consideration: when the motor stops you have a different friction coefficient (static) while slipping was a dynamic coefficient. Those are different. Elasticity of the wire partially compensate the raising force. I think the most accurate way to calculate is with energy equations (Huygen Steiner and 2nd law of dynamics). You need an infinite force to stop some finite (non-zero) mass in a zero time because of F=ma. If the acceleration is very very high the force could be high. BTW, in my opinion, before doing some math, i guess this is a very small contribution to the measured value because of the wire elasticity which "flatten" the curve. Just a tip for further investigation: to make the same experiment with a huge and heavy spool with different friction coefficients.
Excellent revisit of the subject to dispel the misconceptions James.
Sometimes intuition can mislead us, and cause us to include factors that can irrelevant. Is this wrong? No. It implies critical thinking. However, this can lead us to over complicate rather than reduce the problem. If the results do not agree with what the theory would predict, only then should other factors, and inherent complications be considered. This point is rather serendipitously demonstrated with the plastic drum in this video where the material properties were found to have changed significantly due to temperature and melting.
The original problem can be reduced to conservation of energy. ie no more energy can be removed from the system than that originally input. Any residual inertia stored in the drum is any excess energy that is not directly acting against the brake. In this demonstration, the inertia is zero.
Please bear with me while I attempt to explain in my terms how this might appear counterintuitive to some -
Assuming an ideal motor and bearings:
The amount of energy required to spin the drum/flywheel at, say, 1000 rpm, might be, say, 100W. That is 100W of electrical energy used to apply a torque to convert a stationary flywheel into a rotating one by overcoming the inertia to attain 1000 rpm.
Once at 1000 rpm, the torque required to accelerate the flywheel will be zero, and the flywheel will now contain 100W of energy in the form of inertia.
If the motor is switched off, and because we are assuming an ideal setup, the flywheel will continue to spin indefinitely, but will retain the initial 100W of input energy.
Applying a braking force to the flywheel in the same manner as James' dynamometer, the stored energy in the flywheel is converted into heat via friction. This can only ever be a maximum of 100W of heat energy and comes (converted) from the stored inertial energy in the flywheel. The resulting output torque from the residual inertia can be measured indirectly from the force required to create the friction that converts inertia into heat and brings the wheel to rest.
So far, inertia seems to be an important consideration doesn't it? I would suggest that this may be at the heart of the confusion or misconceptions. The difference is in how the system as a whole operates, and most especially how servo/stepper motor is controlled in the presence of an external factor.
Taking the previous setup, but not switching off the motor before applying the braking force.
The motor control loop will continue to apply energy to increase torque to maintain the target 1000 rpm as the braking force is applied. This is converted directly to heat via friction in the brake. As the amount of energy to produce enough torque to maintain 1000 rpm nears the original 100W, less and less energy is being stored as inertia in the flywheel. Torque has no sign, but can be characterized as input and output, but never both at the same time. If all of the available energy to produce an input torque (100W arbitrarily chosen here) is now being converted to heat via the braking action, inertial energy in the flywheel is now reduced to zero. The motor no longer has the energy headroom to produce torque, so therefore stalls. The flywheel/drum stops rotating immediately.
Inertia being cancelled out may seem counterintuitive considering the importance of inertia in the first part of the treatment above, and even when considering the motor to provide all of the braking and perhaps reversing direction, but the whole system's available energy should not be overlooked. The important takeaway is that while inertial energy is important in a pure flywheel scenario, it is effectively cancelled out in a servo scenario where there is part of the system acting as an energy sink - the brake acting against the motor in this case. The whole system can really be reduced to the amount of torque that the motor can produce regardless of any other factors. This is exactly what James demonstrated in the first experiment video.
Personally I would have tried using a hub made from steel with a similar surface finish. I really enjoy you methods and the gear you are using. Yes, I’m a test geek and automation type. I wonder if you serial converter is RS-485 or which serial format?
Forces on the string are calculated from the capstan friction equation.
Oh, cool. The concept of normal force still applies, but the capstan equation is a much better fit for what I'm doing here. Though I guess there's still a mismatch because the capstan would assume static friction.
@@Clough42 Many Capstans are driven and used as winches. The same equation applies just as well to dynamic friction. It is also a useful prediction of friction in control cables, etc.
My suggestion for testing the inertial question would be to add an additional mass to the side of the pulley, such as a thick steel disk screwed onto the open face. This would leave the original surface that interfaced with the paracord as the original aluminum spool. One less variable involved, and no melting introduced. It would also allow for raising the mass twoo to three hundred percent over the naked aluminum spool. Just a thought.
Question not on topic of this video: could your ELS be used for cutting gear teeth on a dividing head without using the plates, sector arms, etc? That would be hugely useful if you're not cutting gears all the time but do have to do it on occasion! Does the electronics have a function that could help figure out the math? If not, could that be added on version 3.0?
The inner control loop of the motor, likes the averaging filter of the additional angular momentum from aditionals mass. Put a scope on the output of that amplifier.
I suggest to use the same alu part with some holes on the body to decrease inertia.
Chuck the Al pulley back up and turn a deep dish in one or both faces. You could easily reduce mass of pulley by ⅔. Easier than drilling holes and will maintain balance of pulley. Exact same coefficient of friction, and no melting of plastic.
@@Zircon10 This is the idea.👍
I agree with your thoughts about inertia not mattering too much in this test, but this is because of the way you are carefully applying the load. Also the reason that momentum or velocity isn't counted in the friction calculation you are using is because that is a static friction calc :)
I have had some fairly good results placing metal Duct tape over the wear/impact areas on prototype pulleys and gears, before making them in a more durable material, better than I have with JBweld coatings. Just throwing that out there.
It's unfortunate that these videos don't seem to get as many views, I love the testing for both comparing the motors and the correction showing how the physical system works.
Yeah...this one will probably never amount to much, but I found it interesting. If I were doing this purely for the views, I would have already quit, or switched to a high-energy, fast-paced style that insults the intelligence of my audience, like so many others.
@@Clough42 well, I'm glad that you experiment with some of this more "boring" intellectual content.
I agree, I don't think inertia plays any significant role. Thanks for all the great videos James!
Not significant, but it plays.
A possible inertia benefit at lower speed is the smoothing of the step motor's cogging effect which could help it get a final few more steps in before giving up 🤷
What if you put something like painters/duct tape around both spools to have the same surface friction for the testing?
The SAME Spool was used on the various motors. So each of the motor used the same spool. So your experiment is valid.
In my opinion, the way this torque testing procedure is being done is basically scrubbing all the inertia away as the braking is applied so I think these results are pretty accurate.
Don't get me wrong I'm a grunt and in no way an engineer but looking at the test with a more basic eye I imagine the motor spinning at say 1500 and then the braking is applied. The inertia of the wheel gets scrubbed off and the rpm starts decreasing and the torque of the motor is fighting you. Now your down to 1200 rpm and the energy of the wheel from the inertia it had at 1500 rpm is now only the energy available at the 1200 rpm level. Once the electronics of the motor detects the shaft of the motor is falling behind the signal sent as the rpm decreases even further, it quits and the meter has already recorded the maximum torque that was applied. Whatever energy from inertia was still available is now stopped by the braking mechanism.
Well, if you want to prove one way or the other, then go to an extreme. Maybe add a steel flywheel of 8" to 10" to the aluminum pulley or whatever size is easy to cut on the lathe so its very close to being balanced. I don't know what a 10" disk will be doing at 1500 rpm but maybe just low speed tests would clear a lot of thoughts on the original test of motor torque and the different effect inertia has on that test. Of course the flywheel will not change the torque of the motor but it will show the effect on the braking and acceleration. Its like a bulldozer with its blade an inch off the ground and driving into a level pile of dirt. Initially the engine RPM maintains rpm due to the inertia of the flywheel and travels into the pile at a fast rate. Then the inertia starts getting scrubbed and the rate of travel decreases as the rpm comes down to the point the torque of the engine is doing all of the work including turning the mass of the flywheel. The rate of travel is maintained at that speed until traction is lost.
In my opinion the numbers are too small for clearing up the argument of the inertia effects in these motors. It might show a difference on a "Clean Jerk" type of braking. The aluminum wheel may actually send a higher spike if you use more wraps and do an instant stop on both materials. It may require a graph type meter instead of a digital one to actually see the effects.
EDIT: Sorry I just came across these two videos and didn't realize they are over a year old.
I really like them and will be bingeing your channel to see whats up. Thumbs Up of course.
James, did you swap out the PM 940?
Was it not rigid enough? I remember you had commented in an earlier video on that.
Wouldn’t adding weight to the original pulley be a more consistent test? Either way I would think the extra inertia would only help with very transient loads. Could that be helpful? Perhaps for certain circumstances?
Why not make second aluminium wheel but drill out few holes or turn some material away.
Yeah, this one. This is certainly the best method to keep everything but the inertial momentum the same.
Very interesting video, thanks for the followup. I do think you were 100% spot on and think the results mostly corroborated that. I do have a minor quibble with performing the experiment until you got the "desired" results. I mean, obviously you aren't creating le grande k or anything here so its sort of a non-issue, but still as someone who has to test things professionally, it sorta hurt to see. Either way, I really enjoyed the technical details.
Of course. The inconsistent results were only at 500RPM, oddly enough. 1000 and 1500 were consistent, which is why I went back and retested at 500. It didn't fit with the other data, so I assumed something nonlinear was happening. Given that the results were inconsistent, I retested until I got values that proved the hypothesis was reasonable. The next step would be to design a new experiment that works around the discovered issue and do it all again. Feel free to do that if you like.
@@Clough42 I'm not trying to say your results are somehow invalid or that the process was was a sham, far from it. The rigor of your testing was impressive and not only that the fact that you even did a followup to address specific concerns was commendable (lol, as if commendations from a rando on the internet mattered, but anyways). I too definitely thought that the inertia of the wheel would make a difference and I thought that was what explained the results from the first video when you got some results that exceeded the manufacturers results. The obviously demonstrated conclusion from the second video is that it is in fact a non-issue.
So sure, you could have made the experiment blind, or used some sort of mechanical device to trigger it removing the human element. Thats all true. All I'm saying is that what you did could have influenced the results (likely not in this case, sure). Its a problem in many scientific studies and experiments as well and has led to a call for people to preregister their papers. To quote from Center for Open Science (COS) "Preregistration separates hypothesis-generating (exploratory) from hypothesis-testing (confirmatory) research. Both are important. But the same data cannot be used to generate and test a hypothesis, which can happen unintentionally and reduce the credibility of your results."
So yes, its a real problem in highly rigorous scientific papers. Its a "mistake" you made(I'm hesitant to even call it that because your goals were not necessarily one of extreme rigor here, but rather more of a demonstration). You are not purporting this to be some rigorous scientific test so I don't see any problem. I'm not criticizing you for doing it. But it is an interesting topic and thing to think about and seeing it play out in your video was, well, interesting.
@@isavedtheuniverse yeah... Mistake isn't the right word. I realized as I was doing the test that the value of the results had been impaired the moment I started getting inconsistent readings. There was essentially no way to recover at that point. Claiming the hypothesis disproven would also have been weak, and likely invalid. I continued to see if I could learn anything, and I published the video as a matter of personal integrity. You got to see what happened. And you're right, I moved back to the hypothesis phase, and a new experiment is in order to test the new hypothesis.
I kinda expected you to add some inertia to your aluminum spool or to drill some holes into it to make a comparison rather than making a whole new wheel.
Your argument is not wrong. The important part is that you are measuring torque at which the motor shuts down, not torque generated by the stopping spool while the motor is still forcing it to rotate. The second one would definitely be affected by inertia.
Won't the driver current limits and driver supply voltage significantly affect these results ?
Try a second pulley out of steel or one out of aluminum that has been hollowed out to reduce the mass. Aluminum lightened pulley would likely be the best since the coefficient of friction would be the same. Any tension on the pulling end of the string need to be subtracted out also. There is energy in the pulley. The big issue is that it has to be dumped at a rate that effects the measurement. With that you would need to know the change velocity / deceleration rate.
I would have thought that the greater the rotating mass of the spool (flywheel) would require more friction to stop it rotating when it overcame the torque provided by the motor.
To every action there is an equal and opposite reaction.
I think the reason that you got the same results is that the motor you're using is closed loop. So when the drive can't keep up with the torque requirement it actively brakes\locks the motor meaning the inertia of the ali spool is pretty much negated. It'd be interesting to try the test with an open loop stepper and heavy steel vs lightened ali pulleys.
My intuition still says you'd find a difference. Imagine doing it with a DC motor... If you spin it up to 1500rpm and disconnect one of the power wires before pulling on the string then you'd be able to measure torque even though the motor is not providing any.
I know the viewership doesn't warrant it (unfortunately), but a relatively quick fix for the melting potentially influencing the result might be to just wrap (and/or glue) a strip of aliminium foil to the friction surface of the PETG rotor. It might not be the exact same surface, but it would certainly get us into the same realm.
15 years ago I designed a cricut to start a jet engine with a pratt whitney specified torque vs rpm profile. In testing up to 3000 rpm starting, I learned there is no flooded engine like a flooded jet engine. To rev it up with starter and no fuel or ignition it shoot a fog of fuel and air out on the tarmac.
They showed us a 16mm film about the coefficient of friction in the 2nd grade in 1958. I am not sure everyone understood it quantitatively, as none of us were multiplying yet..
A torque wrench doesn't care about inertia, so neither would any other torque measurement. If you think about it, the more accurate clicker style wrenches remove(reduce) inaccuracies caused by tightening too fast(velocity). Inertia is dependent on velocity, torque is not. In this case inertia would play into how fast it stops, or the power the motor would need to stop itself. Basically, we're talking about the difference between torque and power.
Maybe wrappig masking tape around both would prevent the plastic spool from melting and make mu pretty much identical, Then tje only variable is mass (and angluar momentum);
In any case, once the spools have stopped, we've changed from a dynamic set of forces to static and, since w is then zero, there is no iniertia,
Granted, intermia may have cause a spike, but not for long...
Opps, sorry. Missed that others had suggested following out the aluminium pulley. BTW, bought your board and built an ELS for a Craftsman (Atlas) 6 inch lathe. Works great.
Great.
My intuition says inertia does have an effect but its insignificant. In the earlier tests it is a consistent value applied to all the the test segments so it can be negated. To prove inertia does contribute imagine we repeat a high speed test but just before pulling the string uncouple the motor. You will get a torque reading
easy to disregard. Also the motor when going out of sync it acts as a brake "cancelling out" the intertia of the spool... ?
Based on your assumptions (that the flywheel is not accelerating/decelerating equating a to a static equation/test), you are mostly correct. You are excluding the very small effect of aerodynamic drag of the wheel in the air, however, this force should be so small as to be ignored. However, we know this is a dynamic equation, and the flywheel's inertia resists any acceleration, emphasis on any. The way you are testing the torque, there HAS to be some deceleration in order for the timing to go bad and the driver to detect a misstep. As we know, F = m*a. Based on your results, and as expected, it appears that it the deceleration is small. Additionally, I think you will find that the results will become more skewed as the RPM goes down due to inertia being a result of angular velocity squared; as RPMs (angular velocity) increases, the weight becomes a smaller component of the equation. Keep up the good work.
This is an interesting line of thought. I think in general, there's only a small deceleration and the pulley ends up running at a slightly lower speed than commanded, which causes the error to grow even though the pulley speed is almost constant. This rabbit hole goes really deep, and I need to try to stay focused on the original question: Which motor combos are a good value for the ELS. :)
Inertia and torque are two different animals. Torque is the twisting effort that in this case is trying to overcome the Inertia of the spool. Inertia will dampen torque irregularities. What you have built is what is known as a Prony Brake dynomometer. A very common old style of engine dynomometer. Providing you load the motor using the brake, as opposed to brake the motor using the brake, then Inertia will make no difference to the actual motor torque. Put another way, you probably need to be stopping the motor over a period of several seconds to eliminate the effect of the spool Inertia. Great video anyway. Greg
Is stall torque really interesting for ELS? As soon as the applied torque slows down motor RPM the motor is no longer clocked to the spindle...
Stall torque defines the performance envelope where the carriage stepper IS still clocked to the spindle, yes? That envelope dictates what feeds & speeds are viable, so the point of plotting it out is to keep within the working limits where the ELS motor doesn't lose clocking.
It doesn't really seem much different than determining the capabilities of any other hobby lathe, except instead of slipping belts or breaking shear pins or gear teeth, it loses steps or stalls.
Not really relevant. However, it's the figure that's usually advertised, and I was curious to see how they matched the specs.
Use your own judgement, but you _could_ superglue (or otherwise attach) the three drums together.
I considered it.
The inertia of the pulley should not matter because the forces are steady and thus reach an equilibrium. The very slow increase in tension results in a virtually steady state.
A sudden increase in friction (like the plastic pulley melting) will result in a larger reading from the scale. In this case the inertia does play a role. You did not see this increased force peak because the digital scale is not fast enough to register it before the system is stopped.
Good point. It's counterintuitive that the reading would be lower. I think you're right that the meter was too slow to catch the peak.
For the simplified formula... μ for kinetic friction assumes constant velocity. Depending upon the two interfacing surfaces and velocity ...kinetic friction changes (take this to the extreme... i.e. the difference between static friction and kinetic friction). Like the slip/stick nature of interfacing plastics. I agree that if you pull the cord slowly it will minimize this. But because μ us not constant you need to account for F=ma for deceleration. Changes in a will skew results. This would likely account for the reading variations you were seeing on the nylon to petg setup when surface welding lead to higher deceleration when it grabbed.
Great video as always... But confirmation bias seems to have played slightly into your conclusions IMHO.
I did some reading on this, and found two classes of answers: 1) there is some variation of sliding friction with speed, depending on the materials and surfaces involved, and it can go either direction, and 2) as mechanical engineers, we really don't care much because we generally try to avoid sliding friction when designing mechanisms.
That was a very diplomatic way of saying: "If you don't like the results, keep trying until you get the results you do like".
It's easy to say what should have been done....so I'll say it :-) He should have used a second motor as the "Brake", which is what most proper dynamometers use.
Friction braking can be very difficult to repeat accurately and consistently and can produce widely varying deceleration rates.
Great to read a comment from someone who understands what's really happing in this test!!
You could, with a bit of machining, make a plastic wheel with an aluminum "tire" so that the surface that the string bears on is the same. You could also just drill or mill out a bunch of "lightening" holes in the aluminum wheel.
Yeah, I considered all of these and then tried the easiest thing first.
You could have superglued one of the other spools to the one mounted on the motor. Instant doubling in mass, no change in µ.
That worked about like I thought it would. I'm not surprised that the plastic melted but I wouldn't have predicted it
Trepan out the center of the aluminum disc and fit a plastic hub - same mu with less mass.
Just measure your tugging force and substract its value from the torque force measuring. And if you will want it very right consider the weight of the string hanging down.
So fun fact, with the friction equation you used (which assumes negligible deformation), mu doesn't actually have anything to do with contact area
I would think that the ineria of the whole servo/stepper rotor is at least in the same range as the pulley, if not higher. While it's diameter is smaller (moment of ineria grows with r²), its weight is for sure larger (all iron and copper, assuming a synchronous motor, not an induction motor). So you didn't change the total inertia from 350g to 69g, but rather from lets say 700g to 419g (numbers just for pointing out the topic, but in reality we should use gmm² or so). This is a much smaller change than perhaps assumed.
Secondly, your derivation sounds correct, but only for steady state. However, your servos/steppers stopped quite abruptly, dumping all that energy in the string within very short time. And as F=m*a, or here M=I*alpha, the faster you deccelelrate, the higher the force. This might therefore have lead to higher forces as while the servo was still running. Additionally, you were slightly startling when it did (who wouldn't?), so this can also contribute to measurement error.
IMHO, you should have done each speed at least 10x, and then use mean and stddev. to include measurement uncertainty. But, man, we are all too much used to the high precision and reproducibility of electrical measurements. :-)
BTW, I wanted to measure the power vs. torque of my milling spindle (900W motor), and had a bicycle disk break in mind. Then I saw your string idea, which is much simpler, so I changed my plans. But after seeing the melting of the string, I think a disk break is better suited. :-)
How about just threading some holes on the aluminum pulley and bolting weights on it?
My choice also.
OK, at 7:10 I’ll note that the explanation of friction seems to have overlooked the difference between static and dynamic coefficients of friction. Static coefficient > Dynamic coefficient generally, so when you increase the angle of a ramp, a box at rest begins sliding at some point (and doesn’t stop sliding till it hits the bottom. However the angle at which this happens is steeper than the angle at which the box will slide if you give it a little push (going from static to dynamic friction. No idea if any of this makes a difference, so back to the video...
I did skip it, because this system is in motion. Static friction only comes into play once it stops and the measurement has already been taken.
What about using abs and doing an acetone vape to smooth the outside. It will be less prone to thermal issues as well which should provide more consistent results..
Maybe. I print ABS and PETG at the same temperature, so I would expect that for similar amounts of friction, they'd reach similar temperatures. Though ABS is slicker in general, so it would be spread out over more wraps.
@@Clough42 with abs the vapour smoothing how ever will make the surface glossy and smoother than PETG, plus the glass transition point on abs is higher, meaning it shouldn't be as quickly affected by the heat generated by friction. Plus the acetone vapour smoothing process tends to make the surface more rigid. Just a thought.
@@Clough42 by the way, really enjoyed the vid. About to start building your ELS.
you can also wrap an aluminum foil around the plastic pulley
You could wrap a layer of foil tape on that PETG wheel to fix the melting issue.
The inertia of the pulley is definitely a factor that needs to be considered in the experiment design, you are after all braking a rotating mass which will contribute to the reaction force you measure.
That being the case, measuring the force at a constant rpm does overcome the inertia problem. If the rpm is constant and the mass constant the inertia remains constant by definition.
So, the issue with your experiment is really just working out if that rpm is constant at the time of measurement. With the equipment you're using I don't think you can.
You would need to measure the force and stop that measurement the moment the motor is cut off.
Alternatively, you could measure winding current and compare that to a torque vs current chart. Current will obviously go to zero when the motor is cut out allowing you to side step the inertia problem.
But, how much of an issue the inertia actually is is a whole other argument. It may not be contributing any error at all. I'm sure the rotor mass is also quite high so changing the pulley alone isn't necessarily a good way to tell.
Basically, I don't think you've demonstrated it either way and if this were to be peer reviewed it'd be a massive fail. But this is your home workshop so that's not the standard you're working to. You're working to the your happy with your results standard, and it looks like you've done that very successfully. It's certainly a lot more effort that I would ever go to, I'd just look at the spec sheets.
You're participating in the peer review right now, except that you haven't added any new experiments or data to the conversation-only expressed your doubts based on intuition. That's a good place to start. If you look through the comments, you will find someone who reanalyzed my data as you suggest (stopping at motor cutoff) and confirmed that the difference is negligible. I'm totally open to learning more if someone else does a better experiment with better equipment and produces different results.
I think you've already done quite a good job, and while there's certainly the opportunity for more rigour I don't know that it would make it a better job. Just more complicated, and there's no end to that particular rabbit hole.
And I want to apologise, I reread my comment and on second reading it seems critical. That wasn't the intention.
My intention was to say that I don't think you've proved it either way, but that for what you're doing I you don't think need to in the first place. It's just guilding the lilly for no particular reason. What's more it's an opinion and I reckon I've been wrong more times than I've been right. Don't take it as a criticism of your work, because it isn't.
Depends on how fast it is stopping.
James, it doesn't make sense,
if you are trying to take aluminum pulley under consideration, so what about the weight of motor rotor itself which is much more heavy then aluminum pulley?
I would conclude that it also doesn't make much difference. Not much you can do about it anyway.
Paused the video at 2:00 to say that the inertia would affect the time it takes to get a reading but would not affect the amount of force needed to stop the motor, which is what is being measured. If you use perfect materials, regarding coefficient of friction, wear, heating etc, then you will be measuring the torque in the motor. It may be that a long time spent slowing may heat up the spool and wire leading to odd readings.
I'll now let the video play through and see if I am right.
OK, paused again at 4:20 and the test would be much better if you had a device to apply slowly increasing force to the string, not a human. That would allow a much more standardized test.
Unpause.
OK, paused again at 5:11 when I realized that the device you want, and that your diagram looked like to me, is a freely rotating drum brake that slowly increases friction in a known way between the rotating disc and the part attached to the scale and applies the force needed to stop the disc in a reliable and known fashion every time. You are almost literally re-inventing the (car) wheel.
I think your theory is sound (inertia does not affect the reading), but your implementation could do with some more work because your test rig has too much human input. Not going to pause again, this post is long enough and is what I think.
I'm reinventing the Prony Brake. It does exactly what you describe, with the free end of the band anchored to the lever frame.
My new digital force meter came in yesterday. I suppose that I will NOT be making my pulley out of PETG, as planned.
From my career of tightening billions of caps the weight of the cap chuck inertia is and issue and Aluminum is favored as the steel units cause excessive cap torque especially as you go into higher speed single head machines
But your string is much more gentle and is probably only going to make a small peak difference
That's a different dynamic scenario, as the inertia of the chuck is being dumped into the cap as it shoulders out. In this scenario, there is no hard stop. It only looks like it because it happens in a short period of time.
Too bad this video gets less views, as it's a great learning moment testing these things.
Use the original aluminum blank, drill holes or remove material to reduce mass.
New PM mill in the background......what happened to the PM940......I wonder if he has already sacrificed the 940 to the CNC gods....if so, we have some great vids to look forward to.
Good eye and great question
Oh, that is weird. It kind of looks like a different mill in that shot.
The reason I don't think it matters is because the difference in speed when the spool is being slow down is too small so the difference in energy is too small. Is not like you're stopping instantly the spool and measure the force.
I enjoy the inertia the ELS topic has generated 😇
I didn't have any issue with inertia, for the exact reason yo9u explain: it's not a variable in the relevant equation of motion. HOWEVER I do have an issue with you changing the number of turns of the cord. I understand why you did it, but the more turns of the cord, the larger the friction, thus the greater the COF (mu), thus the less force you have to apply to counteract the torque of the motor, thus the measured torque is lower than it really is.
Even if the weight of the pulley had a little influence isn't an issue since the motor will need to be attached to something anyways. It seems that having the motor attached to the pulley for testing will give a more usable and realistic measurement/curve anyways. If there is a way to test an unloaded motor, the numbers would not reflect real word application. And there is a whole other side to understanding how weight effects round objects. Weight location can noticeably effect the centrifugal force and momentum. Your pulley design is the best for testing since the weight is more evenly distributed and would have less of an effect anyways.