EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: RedLion on November 25, 2020, 02:52:26 pm
-
Hey all,
I'm trying to build an inertial dynamometer for small hobby motors, so I'm looking for an angle sensor or rotary encoder that's capable of high RPM, at least 75000rpm, more if possible, that's also as low friction as possible, so nothing that has detents like a rotary encoder.
The motor will be bringing a flywheel up to speed, the motors I'm currently working with are capable of over 60k rpm. To avoid friction, the motor would best be directly coupled to the flywheel and the sensor ideally would be too.
In terms of resolution or functionality, it's not that critical, as the area of interest is primarily the high RPM range. A pulse for every 10 degrees or so would be plenty accurate.
Ideally I'm thinking of those optical devices the old ball style computer mice used to have, or something employing hall effect or so; something with a wiping contact would probably not last.
Has anyone ever worked with such a sensor or does anyone know a search term I should employ?
Thanks,
Ivo
-
I'd think an optical encoder would be the right thing. Can you paint a pattern anywhere?
-
variable reluctance sensor should be OK. Kind of like a hall-effect but won't go down to zero. It's just a coil and the shaft needs some magnetic variation. Typically used with gear teeth or a top-dead-centre lug
https://duckduckgo.com/?q=gear+tooth+sensor&t=brave&iax=images&ia=images
If you use a gear alone you can't tell where it is in the rotation. Typically a missing-tooth algorithm is used (with one gear tooth removed) to detect a gap in the stream.
-
That might be possible, the inertia wheel is a solid piece of brass, but printing on it might be difficult.
I could maybe 3d-print a slotted wheel and use one of these light gap thinies like this:
https://www.mouser.lu/ProductDetail/Omron-Electronics/EE-SX4340?qs=u4fy%2FsgLU9PvYeKXUeqGdQ%3D%3D (https://www.mouser.lu/ProductDetail/Omron-Electronics/EE-SX4340?qs=u4fy%2FsgLU9PvYeKXUeqGdQ%3D%3D)
However I had hoped for an integrated solution that would save me from lining this up everytime I build such a thing, as other people are going to want such a thing eventually. I'd have hoped for something like this, but this one only goes to 3000rpm: https://uk.rs-online.com/web/p/optical-rotary-encoders/2632873/ (https://uk.rs-online.com/web/p/optical-rotary-encoders/2632873/)
-
Can't you look at the commutation in the power lead? Then nothing has to be on the shaft.
Steve
-
I have actually never tried that, I'll have to check with a scope. If that works it'd be marvellous, but these are those extreme high power brushed DC slot car motors. I expect the readout to be noisy as shit, plus I don't own a current probe.
-
I have never thought of the optical slot sensors as being particularly fast. Well, they're not! Maximum frequency of this model is just 3 kHz:
https://www.digikey.com/en/products/detail/panasonic-industrial-automation-sales/PM-K65/5962593 (https://www.digikey.com/en/products/detail/panasonic-industrial-automation-sales/PM-K65/5962593)
Even using the rise and fall times, the device is only good for 9 kHz.
There are many of these devices and this is the only one I checked but I would imagine the others are similar.
70,000 RPM * 36 Pulses/Rev * 1/60 Min/Sec => 42,000 Pulses/Sec, far faster than the sensor.
You might be able to get 2 Pulses/Rev - at least the rotor would be balanced.
70,000 RPM * 2 Pulses/Rev * 1/60 Min/Sec => 2,333 Pulses/Sec.
The slots are going to be 1/4 of the track, each.
-
Well it looks like I may have to settle for a lower resolution then. With a heavier mass I should be able to slow the motors down significantly, although having that much brass spinning so fast sounds quite scary.
-
I have found these on DigiKey:
https://www.digikey.lu/product-detail/en/tt-electronics-optek-technology/OPB615/365-1661-ND/1636989 (https://www.digikey.lu/product-detail/en/tt-electronics-optek-technology/OPB615/365-1661-ND/1636989)
https://www.digikey.lu/product-detail/en/tt-electronics-optek-technology/OPB616/365-1662-ND/1636990 (https://www.digikey.lu/product-detail/en/tt-electronics-optek-technology/OPB616/365-1662-ND/1636990)
They claim to be very fast, maybe I give them a try. At less that 2$ a piece, would be worth it.
-
Take a look at tachometers used for turbine balancing..
-
Where would one find such a thing and what would it cost?
Because I can buy a dynamometer with software for 400$.
I'm obviously trying to undercut that, I'm not doing this for the hell of it.
-
That might be possible, the inertia wheel is a solid piece of brass, but printing on it might be difficult.
I could maybe 3d-print a slotted wheel and use one of these light gap thinies like this:
https://www.mouser.lu/ProductDetail/Omron-Electronics/EE-SX4340?qs=u4fy%2FsgLU9PvYeKXUeqGdQ%3D%3D (https://www.mouser.lu/ProductDetail/Omron-Electronics/EE-SX4340?qs=u4fy%2FsgLU9PvYeKXUeqGdQ%3D%3D)
However I had hoped for an integrated solution that would save me from lining this up everytime I build such a thing, as other people are going to want such a thing eventually. I'd have hoped for something like this, but this one only goes to 3000rpm: https://uk.rs-online.com/web/p/optical-rotary-encoders/2632873/ (https://uk.rs-online.com/web/p/optical-rotary-encoders/2632873/)
lining up a light beam is going to be a walk in the park compared to trying to align and balance something running at 75krpm
an why the need for such high resolution? wouldn't single index be plenty? it also nicely avoids any variation in stepsize
-
I am thinking about 3D-printing the indexed wheels, with varying amounts of slots for different types of motors. The slowest we use at our local club have 18k RPM, the fastest can go up to 80k, that is with no load of course. As a rule of thumb, the lower 30% of the RPM range will see no competitive use and thus are not crucial.
I have access to an SLA printer, so precision should not be a problem. My remaining concern is that the resin needs to be able to take the load without flying apart.
wouldn't single index be plenty?
In that instance I would worry about the imbalance ripping apart the plastic.
-
I am thinking about 3D-printing the indexed wheels, with varying amounts of slots for different types of motors. The slowest we use at our local club have 18k RPM, the fastest can go up to 80k, that is with no load of course. As a rule of thumb, the lower 30% of the RPM range will see no competitive use and thus are not crucial.
I have access to an SLA printer, so precision should not be a problem. My remaining concern is that the resin needs to be able to take the load without flying apart.
wouldn't single index be plenty?
In that instance I would worry about the imbalance ripping apart the plastic.
skip the wheel and just paint a spot on the flywheel and use a reflective sensor
-
That might work too, but I'd lose some flexibility. I'll print some samples and spin them up to see what I can get away with.
-
That might work too, but I'd lose some flexibility. I'll print some samples and spin them up to see what I can get away with.
I didn' say to buy one but to study how they work. They shine a laser at rotor, you put a speck of paint on it and it detects reflection. At those RPM you need only one PPR, because rotational inertia makes sure there will be no speed variations in 20, 50 or 10 revolutions..
Good luck with flywheel, and make sure to design protective cover..
-
At those RPM you need only one PPR, because rotational inertia makes sure there will be no speed variations in 20, 50 or 10 revolutions..
The issue is that I want to go throught the entire band, and I don't know what resolution I need. On these motors, angular acceleration is particularly strong in the low range, so I would like to know for as low RPM as possible.
-
At those RPM you need only one PPR, because rotational inertia makes sure there will be no speed variations in 20, 50 or 10 revolutions..
The issue is that I want to go throught the entire band, and I don't know what resolution I need. On these motors, angular acceleration is particularly strong in the low range, so I would like to know for as low RPM as possible.
you must know what torque/power levels you can expect to know how much rotational mass you need to get reasonable acceleration, so you should ready have all the numbers
-
At those RPM you need only one PPR, because rotational inertia makes sure there will be no speed variations in 20, 50 or 10 revolutions..
The issue is that I want to go throught the entire band, and I don't know what resolution I need. On these motors, angular acceleration is particularly strong in the low range, so I would like to know for as low RPM as possible.
you must know what torque/power levels you can expect to know how much rotational mass you need to get reasonable acceleration, so you should ready have all the numbers
As I said I want to make several inertial wheels, since the motors we use are all over the place. The most I ever heard is about 1kW electrical (that is a thumb sized motor, mind). Who knows what the efficiency is on these, also this has not necessarily been done before. I expect not to get more than 250W on the shaft.
-
I'd be really impressed if you're able to 3D print anything that can spin 75k RPM and not blow up.
If you can buy something off the shelf that will do the job for only $400 then that seems like a no brainer, unless you need enough units for the development effort to pay off.
-
I don't mind effort. I got nothing but time.
The 400$ is what I'm lacking at the moment.
-
50KHz should be much of a problem assuming you need that resolution. With your inertia dyno and being able to directly attache the large flywheel to the motor, assuming you can get it balanced seems easy enough. The non-contact sensors you found should work nicely.
I have no interest in the hobby but I did help out on a couple of larger dynos. For the last one, I made up a small model dyno so I could work on the software and electronics for it. One of the motors I was playing with could turn at around 20K RPM. The first attempt to make something from a bunch of junk was a total bust but the second attempt worked fairly well.
https://www.msuk-forum.co.uk/forums/topic/81512-self-built-rc-dyno/?do=findComment&comment=1014192 (https://www.msuk-forum.co.uk/forums/topic/81512-self-built-rc-dyno/?do=findComment&comment=1014192)
-
Wouldn't that require a pretty good knowledge of the load motor and its efficiency? That's why I went the inertial route.
A load motor would probably be the better way if all that was required was a comparative result. However I'd like the exact numbers, also I have seen inertial dynos before, so I designed it that way as well.
-
Can't you look at the commutation in the power lead? Then nothing has to be on the shaft.
Steve
I have actually never tried that, I'll have to check with a scope. If that works it'd be marvellous, but these are those extreme high power brushed DC slot car motors. I expect the readout to be noisy as shit, plus I don't own a current probe.
I can confirm that measuring the current to the motor works, but the waveform is noisy and can be challanging to pick-up. It looks like a noisy rectified AC voltage waveform, with the number of phases equal to the number of poles on the armature, assuming there's only one pair of magnets on the stator and brushes on the commutator. For example, if the motor has three poles on the armature, the frequency of the ripple will be six times the number of revolutions, per second. It's fairly easy to decipher with an oscilloscope, but a little more challanging to design a circuit to do it automatically.
-
>> It's fairly easy to decipher with an oscilloscope, but a little more challanging to design a circuit to do it automatically.
Made much more difficult because you can't easily low-pass filter it : the noise and distortion all scales with RPM, so a perfectly respectable detector at 1000 RPM is hopelessly wrong at 20K.
-
Probobly useful for some hand measurement on the fly, but nothing that can get me a nice plot in one go
-
Wouldn't that require a pretty good knowledge of the load motor and its efficiency? That's why I went the inertial route.
A load motor would probably be the better way if all that was required was a comparative result. However I'd like the exact numbers, also I have seen inertial dynos before, so I designed it that way as well.
The absorber has an arm attached to it that acts on a load cell. I this case, a 5 place balance. Knowing the length of the arm, the load cell and the RPM, we can calculate everything. For the larger system, I did a forth order polynomial for the the load cell. We used several weights then to calibrate it. The absorber (load motor) needs to be sized for the motor we are testing. For small motors like I was using with the model, all friction was a concern. This is why you see the ceramic bearings, pillow blocks, ground shaft to align it and the high strand wire to interface with it.
This type of dyno is commonly used to make absolute measurements. There are some benefits besides the accuracy. One is we can run steady state. The dyno can run constant torque for example regardless of motor speed, or it could run constant speed. We used to have motoring dynos which we could then drive the motor as well as absorb. So for example simulate the car going down hill versus up.
-
You are starting to scare me!!!
First off flywheels can be dangerous simply due to the wrong material being used and exploding at high RPM's. As such any significant use of a flywheel will require and evaluation of it ability to hold together at high RPM's. Brass may or may not be suitable this is really a mechanical engineering question. 3 D printer objects, especially on cheap 3D extrusion based printers do not inspire me as being safe at all There are two issues, one the integrity of the prints and the second is the balance. An out of balance shaft connected item can be as destructive as something coming apart from excess speed.
As for encoders high speed suitable units are fairly easy to find: https://www.maxongroup.pt/maxon/view/product/sensor/encoder/ENX/ENX6MAG/ENX6MAG02. (https://www.maxongroup.pt/maxon/view/product/sensor/encoder/ENX/ENX6MAG/ENX6MAG02.) However you need to watch the specs very closely as you might not be getting many pulses at all per revoluiton. Here is another option: https://www.controlinmotion.com//bm~doc/lb-minicoder-brochure.pdf. (https://www.controlinmotion.com//bm~doc/lb-minicoder-brochure.pdf.) You will need to look around to see what fits your use case best. Do understand though that the data rate can be pretty high from some of these high speed encoders so your reading electronics and possibly more important the interconnects may need to handle MHz signals.
Also don't expect cheap.
I am thinking about 3D-printing the indexed wheels, with varying amounts of slots for different types of motors. The slowest we use at our local club have 18k RPM, the fastest can go up to 80k, that is with no load of course. As a rule of thumb, the lower 30% of the RPM range will see no competitive use and thus are not crucial.
I have access to an SLA printer, so precision should not be a problem. My remaining concern is that the resin needs to be able to take the load without flying apart.
wouldn't single index be plenty?
In that instance I would worry about the imbalance ripping apart the plastic.
-
First off flywheels can be dangerous simply due to the wrong material being used and exploding at high RPM's. As such any significant use of a flywheel will require and evaluation of it ability to hold together at high RPM's. Brass may or may not be suitable this is really a mechanical engineering question. 3 D printer objects, especially on cheap 3D extrusion based printers do not inspire me as being safe at all There are two issues, one the integrity of the prints and the second is the balance. An out of balance shaft connected item can be as destructive as something coming apart from excess speed.
When I was a teenager I attached an AOL CD (remember those?) to the shaft of a rotary tool and spun it up to roughly 30k RPM. After a moment a vibration set in and then the disc suddenly and violently failed. Thankfully I had the sense to hold it up over my head because there were shards stuck into the sheetrock of all four walls and I was finding small fragments the rest of the time we lived there. 30k is scary, 70k is terrifying. That's a LOT of kinetic energy, think of what happens during an uncontained failure of a turbine engine.
-
First off flywheels can be dangerous simply due to the wrong material being used and exploding at high RPM's. As such any significant use of a flywheel will require and evaluation of it ability to hold together at high RPM's. Brass may or may not be suitable this is really a mechanical engineering question. 3 D printer objects, especially on cheap 3D extrusion based printers do not inspire me as being safe at all There are two issues, one the integrity of the prints and the second is the balance. An out of balance shaft connected item can be as destructive as something coming apart from excess speed.
When I was a teenager I attached an AOL CD (remember those?) to the shaft of a rotary tool and spun it up to roughly 30k RPM. After a moment a vibration set in and then the disc suddenly and violently failed. Thankfully I had the sense to hold it up over my head because there were shards stuck into the sheetrock of all four walls and I was finding small fragments the rest of the time we lived there. 30k is scary, 70k is terrifying. That's a LOT of kinetic energy, think of what happens during an uncontained failure of a turbine engine.
https://youtu.be/WEEDJiC8Jqc
-
OP, if these motors are not custom made, can you provide a link where they can be purchased?
-
I don't know if this is applicable or not. I guy I went to for my motorcycle tuning had built a dynamometer he called an "Eddy current dynamometer". The bike would sit on a contact wheel and essentially turn a generator. As you dialed in load on the generator, the load would be transferred back to the bike. You can get RPM by measuring the frequency coming out of the generator, of course. You can probably scale this down to the point of being able to dial in very small loads. With the right bearings and foresight to create the generator coils correctly, I would think you can probably scale it in RPM and load.
The interesting thing was the way he measured the bike curves. He would bring the throttle wide open and then load the bike to reduce to the measured RPM. This provided a measurement more like that under race conditions where you are coming out of a corner and wack the throttle wide open. Of course it took some math to convert it all to the appropriate units and calibration but it worked and produced very accurate and repeatable results.
There are probably limitations that I'm not considering and all this was explained to me years ago so I can be completely wrong in the description but it all made sense at the time. His numbers were generally lower than those measured with other technology at that time but the tuning results were always superior.
Jerry
-
That's typically how engine dynos work. The dyno is engaged at idle and then the throttle is held wide open as the dyno dynamically adjusts to allow the RPM to rise at a linear rate. The software then plots the torque vs RPM curve and calculates things like peak horsepower.
-
75k RPM is just over 1kHz and that's peanuts for some electronic sensor.
75k RPM is starting to get a bit hard on mechanical stuff though. Balancing becomes important and a 3D printed slot disk may fly apart because of it's own centrifugal force.
I once stuck a small NibiDibbiBibbium magnet on a 16mm shaft and tried to measure RPM with a simple HALL sensor but it flew away at around 18000RPM.
-
Look at Monolithic Power System's MA730, MA732, MA782 and compatible devices. This speed is well within their range.
-
I don't know if this is applicable or not. I guy I went to for my motorcycle tuning had built a dynamometer he called an "Eddy current dynamometer".
...
Jerry
Could have been similar to my model where it measure torque. We used water brakes for the two home made units. Rather than a generator it's a pump where you can control the head pressure to change the load. The generator removes all that valve control.
I came across some students who made one for a class project. They were using the platters from old hard drives. I think they may have been trying to characterize a small gas motor.
-
For RPM sensing at that speed please allow me to strongly, STRONGLY recommend optical as your only reasonable sensing technology.
As others have stated, 75K+ RPM involves serious energy even if the spinning mass isn't all that "massive". If you add anything mechanical to that you now have to (re)balance it perfectly - 75K+ RPM will reveal the smallest imbalances and could do so with catastrophic consequences.
You've already realized you need a non-contact approach. Magnetics requires at least one magnet (probably at least two so you can keep things balanced) and then you have to balance the magnet(s) mass into the system.
The advantages of optics are 1) the device being measured gains zero mass (just a reflective/non-reflective patch, perhaps as simple as a Sharpie marker or dab of reflective paint) so you don't become responsible for imbalancing and then rebalancing the system; 2) optics won't be the slow part your system; and 3) interfacing optical sensing to your electronics is straightforward using off-the-shelf components.
As someone else noted, one pulse per rev at 75K RPM is 1250Hz, or a period of 800uS. That's slow enough to be directly handled in firmware, let alone if you use an on-chip peripheral like a Timer/Counter to offload the "fast" part to hardware.
Finally, I'll share a related story like others have in this thread. I wasn't there that day, but my wife and (then) toddler son visited a friend's woodshop. He put a bit in his table router and turned it on. My wife said it made a "different" sound while it spun up part way, then there was a house-shaking THUD and the RPM's skyrocketed. Our friend turnoff the router to find its bit was gone. They finally found it embedded in a wall at a vector that meant it passed within a few feet of our son's head. They had a hard time removing it from the wall. My family left immediately, and our friend sold nearly all of his woodworking equipment shortly afterward - it scared him that much. In retrospect, he probably didn't secure the chuck properly... a simple error. But routers are under 10K RPM. You're talking about many times that. Do you really want to attach your own unbalanced load, presume you can balance it adequately, and take the chance? I wouldn't, unless I had some seriously accurate machining T&M equipment and the skills to use it properly. I'm honest enough to admit I don't. Maybe you do, but the optical solution sure sounds like a good insurance policy.
-
Seems a bit extreme to sell everything and quit. I've had a few scary close calls where I came close to meeting my end or losing parts, every one of those was a teachable moment that left a lasting impression. I bet he would never forget to double and triple check the chuck every time in the future, and wouldn't ever ignore an unusual sound.
-
My wife said it happened very fast, no time to react to the "different" sound before the bit went flying.
I think he DID take it as a "teachable moment". He was an older gentleman - perhaps in his 70's - and perhaps he viewed it as an indication that his age was starting to affect his attention to detail. I cannot ask him because he has since passed away. However, I can say that he was as passionate about woodworking as we here are about electronics so it would not have been a decision he made lightly. He'd been at it for decades and had a fully outfitted professional grade woodshop as part of the house that he and his wife had built and in which they intended to live out their lives... that discipline's equivalent of the "home lab" to which we all aspire. I still have pieces, from handheld to furniture, that he gave us for Christmas, birthdays, etc. Their craftsmanship is like a Swiss watch.
-
You've already realized you need a non-contact approach. Magnetics requires at least one magnet (probably at least two so you can keep things balanced) and then you have to balance the magnet(s) mass into the system.
Place the magnet next to the hall sensor and drill divots into the steel flywheel. Most likely they will need to drill divots into it anyway to get it balanced.