Testing; Choosing the right sensor for RPM 1

[Zarathuztra]

I have no previous experience collecting RPM data so would like advice from those with previous experience. I need to collect RPM data from my machine and record it's behavior as I apply and remove power. I'll be engaging and disengaging a load. So I'm trying to find out how long it takes for it to come to a rest when I remove power. There are a few pulleys and spindles I could have a sensor observe to collect the data from.

My question, what is the optimal sensor for this job? I've come across information to suggest a hall effect sensor or an optical lazer sensor could fit the bill. There will be a fair amount of vibration so I'm not sure if that creates a big disadvantage for either sensor.

-Max RPM is under 4,000
-time for the output to come to a rest should be about 2~10 seconds.
-sample rate only needs to be about 100 Hz. (would like to reliably know within at least 0.1 s of output stopping)


Anyone familiar with this type of test have some recommendations??

Thanks

 

[MintJulep]

It might be important to know:

What is the expected max speed?
How long do you expect it will take to "come to a rest"? Minutes? Seconds? Micro-seconds?
How accurate do you need to be for the overall time?
How accurately do you need to know what happens while it is coming to rest?

Many of those question also affect the choice of data logger.

 

[Zarathuztra]

@MintJulep - Thanks, I added the information to the OP

 

[MFJewell]

Only thing I have used is an encoder on a shaft. I designed a control system for a small prototype turbine spin up test. It worked really well. The turbine was connected to a air motor and all of it was bolted to a steel test bench.

 

[IRstuff]

Your collection rate seems to be in conflict with your stated speed --> 4000 rpm = 66.6 Hz, so 100 Hz sampling will potentially given you weird results at speed, unless you don't care. I would think 400 Hz might give you better data for steady state.

I assume your vibration is not specifically the shaft against its mount.

There are lots of possibilities:
> Hall effect
> Absolute encoder
> Back EMF measurement at motor windings
> gyro



TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[MintJulep]

And you'll need some way to indicate when to start timing.

 

[Zarathuztra]

@IRstuff, max RPM will be less than 4,000. It will always be between 3,000 and 4,000 for max (usually somewhere near the middle).

For the vibration, my first test is gonna be a lawn mower, so the whole chassis of the mower deck should be vibrating a fair amount, but I'm not sure how much movement there will be from the shaft relative to the deck.

@MintJulep, When I remove the power it should be the start of the timer, since it will need power to remain engaged.

 

[IRstuff]

Nevertheless, 3000 rpm = 50 Hz, so 100 Hz is barely two samples per cycle, which will make it difficult to see what's going on at the start of the spin down and the end of the spin up.

The shaft had better not be translating relative to its bearings, otherwise, you have bigger problems than starting and stopping times.



TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[MikeHalloran]

For engines with self-starters, the industrial standard seems to be a magnetostrictive sensor affixed with its tip close to the starter ring gear teeth. There are a large number and variety of transducers to convert that N pulses/rpm signal into whatever else you might like. The typical number N of say 70-200 can give decent resolution of rotational speed, even within a single resolution, but it's not the sort of thing you can do directly with an Arduino or similar.

Well, maybe you can; I haven't worked with an Arduino, so maybe they are fast enough, but time-based sampling, as noted, is not going to give you what I think you want. You need something that produces a fair number of pulses per revolution, and measures the time between successive pulses, then computes RPM from the pulse rate.





Mike Halloran
Pembroke Pines, FL, USA

 

[dgallup]

Almost every engine with electronic controls has a crank position sensor and / or a cam position sensor. The sensor is usually some form of magnetic field sensor, can be variable reluctance, hall effect, etc. (I'm not aware of anyone using an optical sensor these days but it could be done. I think they are overly sensitive to dirt in many environments.) These are generally reading a toothed wheel with 60+ poles. Usually one pole is missing so you can determine where you are in a revolution. Automotive wheel speed sensors for ABS and traction control are the same but generally don't have a missing pole. There are 10's of millions of these made every year, they are cheap & reliable. They will withstand heat/cold/vibration/shock.

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[Zarathuztra]

@MikeHalloran & @dgallup, in order for that to work, I would need a gear tooth that I attach to the shaft, correct? (currently there is none)

 

[MikeHalloran]

Yes, you need something that looks vaguely like a gear, preferably made of steel.

I'll possibly save you some time here; plastic gears exist, of course.
If you wish to use one with an optical sensor, face it with metal foil or brass.
Even 'black' plastics are translucent at the infrared wavelengths used by optical sensors.

Worst case; you start with a low-gain sensor that can't see through plastic, and
later you get a batch of sensors with higher gain, same part number, and
they can see through your gear or interrupter flag or whatever, and
you get to retrofit truly opaque parts because you can't get a sensor
that's reliably bad enough to work. Happened to me a long time ago..



Mike Halloran
Pembroke Pines, FL, USA

 

[3DDave]

Without knowing the exact motor there is no useful suggestion. What's the reason for caring how long it takes to stop?

 

[Tmoose]

https://monarchinstrument.com/

with ttl output.

https://monarchinstrument.com/collections/tachometers/products/pocket-laser-tach-200

 

[GregLocock]

Optical sensors can be a bit easier to package. You can get dead posh ones that don't need stripes but the ones that pick up off reflective tape are good and robust. Beware of resonances in the mounting bracket for the probe.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[CWB1]

If you only need short term durability I would lean toward an optical sensor with a reflective tape or other marking, its cheap and easy to install. If you need durability then I would recommend any of the various timing wheel schemes mentioned above using hall-effect or other sensors.

 

[Zarathuztra]

@3DDave, sorry if my OP wasn't clear.. Engine is not a factor, since I will be only looking at the output load. I'll be engaging and disengaging that load from the engine, but I'm only concerned with the time it takes the load to come to a full stop when I remove power from it.

@GregLocock, dead posh? I'm not familiar, could describe?

@CWB1, Since I'm trying to be cost effective, that sounds in-line with my preferred approach. I'll probably utilize this on multiple machines, so if it is easier to setup, that will be a bonus too. Do their tend to be calibration issues with the reflective tape option?

Thanks

 

[GregLocock]

dead posh = very nice and expensive

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[3DDave]

If an item is not a factor, don't add it as a factor in your problem description.

 

[Zarathuztra]

@3DDave, I didn't. But if I'm somehow missing where I added it as a factor, please feel free to highlight it for me so I can correct that.