precision shaft engineering

Please note that there is a world of difference between 'precision' and ordinary ball bearings. Most precision types should have no problem with 10K rpm and contribute very little vibration or friction. Paul Mathews

First off, does the shaft have to be 1m long? If you can make it shorter, that will help greatly.

Play, combined with the shaft not being true, it probably going to kill you...you might get it to spin fine, but I'm not sure you'll be able to eliminate the vibration enough to not swamp your measurements.

All other things being equal, a larger diameter shaft will be stiffer and therefore have a higher frequency excitation mode...if you can get the excitation frequency higher than what you're driving the shaft, a lot of your vibration should go away.

Can you head the bearings or cool the shaft and do a thermal fit?

I think you'll probably have to fix this before you do anything else. If the shaft it out-of-true you'll need play in the bearing or the shaft will bind. But if you have play in the bearings you're not going to pin the shaft's excitation modes down enough to prevent it from whirling.

Good luck! Tom.

long 5mm dia and

nothing loading the

drill, and a dremel etc,

experiment is going to work

something, although

to have any of

speed, wich is an

understanding of it, so ive

play and unequal

as both of these

housing the bearings

problems i dont

preloading increases

the play/friction ?

have inner races.

inside of the motor

shaft diameter if

of the shaft where

still be close fitting

totally straight over

wonder about a tapered

if these are any

sure I could afford

I suspect your length to diameter ratio is a significant contributor to your problem. You may be experiencing a vibration situation by using such a small diameter. I would suggest you replace that shaft with a larger diameter hollow shaft to reduce unbalance which is magnified by rotation. You can use a large hollow diameter shaft (25+ mm OD x 2 mm wall) with plugged stub ends to step down the diameter at the bearings. A larger shaft diameter probably would be cheaper than precision bearings.

In addition, there is an old "rule of thumb" that states the length of a shaft should not exceed 20 times the diameter. Granted, this applies to torsional twist, but that too could be adding to your problem. Consider a larger shaft diameter - hollow to reduce weight.

Jim Y

Dear colin:

You describe an impossible situation for a slim shaft to do what you desire. And a larger cross section may present more stress than normal materials can take.

Why don't you simply drive a couple of motors at the point of application (stepper or servo)?

David A. Smith

A long skinny shaft of this kind is called a quill shaft. It can soak up some angular vibrations, but will whirl in several modes. Better to use a thin wall tube if you can. Then, you could just use a common or garden ball race at the ends.

Brian Whatcott Altus OK

thanks.

I searched for 'precision' on the website of a fairly large distributer here and it drew a blank, are they only available as special order? sounds if its probably to expensive and if they are ball races they stil have the problem of preloading, wich is difficult over a long shaft.

the other problem is the grease cuases a significant amount of friction, well drag anyway. I tried removing the grease and although the friction was much lower they destroyed themselves. some light oil may have helped but seems to also create significant drag. looking at the scratches inside the ball race maybe a needle roler will survive better with little or no lubrication.

I found some 2 row angular contact ball bearings, although a bit bigger than what I want, do these have any play without preload ?

Colin =^.^=

no, the longer the shaft the better as the stronger my measurement is.

wel I just need to reduce it by an order of magnitude or more, given that its a prototype made with the most rudimentry of machine tools, and not much experience of bearings or a great deal of thought I dont see that with a bit of expert help I cant acheive this :)

I think I pass through the excitation frequency at about 5000rpm, it sounds awfull and requires a lot more power to accelerate through this band, but above this its relativly smooth.

Im not sure what exactly the frequency depends on, the whole device is suspended. I did consider rubber mounting the bearing blocks.

its difficult as some of the bearings are inside the motor and to assemble it takes a lttle more time than the shaft would take to heat up. I have a batch of rods some of wich are exactly 5mm and cant slide the bearings down, another shaft they slide easly enough, ive scored the place where the bearings sit, and this removes play, but ive found these marks flatten out after a while and no longer serve any purpose, a center punch mark is far too much. initially I thought I could do something to coat the shaft with a hard substance, like electro plating or anodising wich on aluminium increases the thickness. but its not that easy.

I spent ages trying to straighten the shaft, in situ spining it to find the high points and just bending it in the right direction, this got it a lot straighter, the shaft easily bends under its own weight more than there is actual bend on its own. and so is unlikly to bind the bearings.

balancing the shaft and its attatchements wil be a lot harder, just rotating a tiny circlip on the motor with no shaft atatched makes a huge difference at 10krpm, at one point its in such good balance theres hardly any vibration or noise

I rigged up a strobescope assuming this would be useful but as yet I fail to see how exactly. I gues I need to combine this with a vibration sensor ?

thanks

Colin =^.^=

You might want to check around a salvage yard or ebay for a old supercharger. The Eaton M90 is a popularly available model used in ford thunderbids, various GM products, Jaguar etc...

These have a input shaft speed set to max out at 18,000rpm with a significant load and low if any vibration. At the very least you could get a shaft and some good bearings.

I think you're going to have to go to a larger shaft then. Your combination of shaft length, diameter, and RPM is going to be very hart to stabilize.

That definitely sounds like going through one of the vibration modes...since you're hitting it at 1/2 of your target RPM, I think you can safety assume this system won't work as currently designed. You need to drive the fundamental frequency up above 10000 RPM...either more bearings or a stiffer shaft, or possibly more rigid bearing mounts.

Frequency will depend on the shaft stiffness and the support condition at the bearings (how rigidly the bearings hold the shaft). Putting the bearing blocks on rubber will probably make it worse, not better.

If the shaft is bending noticably under its own weight now, it's not nearly stiff enough to spin at 10000 RPM.

Balance would be good, but it's not your main problem here. Even a perfectly balanced shaft is going to vibrate if it's not stiff enough for the speed and and supports.

The stroboscope should be able tell you exactly at what speed you're hitting the resonant frequencies.

In theory, that would allow you to know which vibration mode the shaft is entering and you could guide your fix appropriately. However, given the prototype nature of this project, I'd just forget this particular shaft and go nuclear...get a very large diameter hollow shaft that's way stiffer than you need and just make the problem go away. Or get rid of the shaft entirely and use synchronized motors.

Tom.

Aha! A dose of reality sets in. It's not like 10 krpm is anything stratospheric. Your Daddy's Buick's prop shaft went to maybe 6000 rpm in top. And that featured a 2.5 inch diameter five foot long mild steel tube welded to universals at each end. The balancing consisted of a steel tab spot welded on at the appropriate spot.

It seems like the skinny quill shaft in question may be made of aluminum. Replace that with a thin wall steel tube of twice the diameter, and half the weight, and that's it! All done.

Brian Whatcott Altus OK

The shaft is stainless steel and is supported by bearings at 5 points along its length. The thing does spin at 10krpm quite happily, but one of the the problem I am trying to get rid of is that of play, and also friction as much as possible, although a larger shaft might reduce the twisting due to friction wich is the problem, im unsure if the larger bearing will be any better with regard to play. particularly given that preloading ball bearings is proving difficult. ive seen 4 point contact bal bearings, will these have no play if used with no preload?

so alternativly I could just get an old rwd car, turn it upside down and use its propshaft !

Colin =^.^=

I have bearings in 5 positions along the shaft, so effectivly ~250mm between bearings. I made the distances unequal to eliminate 2nd,3rd harmonics etc.

Im considering going to a maximum of 12mm shaft as this still allows me to still fit needle roller bearings inside the motor. I can get 12mm tube but this I think is only tube meant for liquids not shafts lol. Ive no idea if this will do, there is some cheap welded or some rather expensive non welded stainless steel tube. with wall thickness of 1.5mm.

Or I can get a solid 12mm shaft, but this is just stainless steel rod or bright mild steel rod or silver steel rod, im not sure wich of these has more rigidity. the torsional stiffness goes up with diameter is it 4th power ? thats 33 times, but how much would the drag/friction from the bearings go up by ? and is play proportional to diameter too?

there is no torsional forces other than bearing friction drag etc, I have considered concentric shafts so that the inner shaft has zero torsional forces on it, but this is a bit more of an engineering challange.

The main thing is I need is some thing wich is easy to make, with easily available parts etc .

Colin =^.^=

I dont see how you can say its impossible, I have it spinning at 10krpm, true theres a bit of vibration.

dont forget I said I had to put a few more bearings along its length to avoid it wobbling like crazy.

however I have just found that at 8krpm there is a realy nice sweet spot, virtually no vibration that I can feel, 500rpm either side and it vibrates a fair bit. im not sure if theres other sweet spots yet.

the point of the shaft is to ensure both ends are in perfect sync, wich would not be acheivable to anywhere near the same acuracy with stepper motors or with servos.

the shaft is driven from the center. Currently I am looking at less than 1 minute of varation of angle between the two ends, when its working well anyway.

Colin =^.^=

Dear colin:

Harmonics, like in music. If you could hit 16krpm (and it didn't thwack you getting there) it should be "sweet" again.

Adjust the spacing between bearings and you'll adjust the sweet spot. Just like a guitar string. Of course, you'll need to add more bearings.

How many bearings do you have now?

There will be torsional oscillations also. Driven by the bearings. Do you have encoders on both ends, so that you can be sure of what you say? Because you *don't* have perfect sync, unless you define "perfect" as "I can't see / measure any difference".

And a stiffer (larger diameter) shaft will decrease the torsional oscillations. Just probably not the energy stored in the member doing it.

What are you driving with this shaft?

David A. Smith

I suggest you search on "precision ground ball bearing" or a variant. I had no trouble finding sources, references. Here's one source:

Talk to apps engineering about your application. I hope that your shafting does provide a zero-clearance fit to the bearing ID, since that's essential in this case. I agree with the other posters who suggest hollow shafting, but I don't see it as essential. Another useful tactic: mount your bearings in energy absorbing material such as sorbothane, isoloss, etc. A little bit of damping goes a long way. Paul Mathews

By the way, 'precision' bearings are usually only 30 to 100% higher cost than garden variety ball bearings. Paul Mathews

Is it as simple as the shaft flexing like a guitar string ? does it do this in 1:1 time with the shaft rotation ? or is the heaviest point flung all the time so the shaft isnt actually bending? or does the shaft rotate and the flex stay effectivly stationary ? or maybe the flex rotates in the opposite direction to the shaft ?

I appreciate if theres resonance then there will probably be harmonics too under certian conditions anyway.

I have one set of 2 in the middle, 30mm apart a set of 2, 45mm apart at each end, and 2 single bearings between each end and the middle,

so the 4 longest sections between bearings is about 200~220mm

B-B----------B------------B-B--------------B--------------B-B

Its an optical experiment, the optical devices at each end act like encoders so with processing I can tell the difference in angle to much less than 1 second of arc, ofc this is after averaging over many revolutions, so torsional oscillations are rejected. the parameter being measured is aranged to be cyclic so absolute errors due to difference in angle can be eliminated over time. if so this would be what I describe as perfect sync.

however the measured difference in angle varies considerably at the moment such that it would take a long time to average out.

a larger shaft would be stiffer and so reduce error due to twisting, but it might also make the errors due to play/vibration/friction worse. not to mention put more vibrational energy into the supports and the rest of the system.

a slightly larger hollow shaft would probably be my next step. but how well would stainless steel tube/pipe meant for high pressure suffice. and still the issue of wich type of bearing is best.

Colin =^.^=

Dear colin:

...

A rotating guitar string, but yes.

If there are no "sweet spots" between 0 and 8k rpm, yes.

No, the shaft is always bending. And your "sweet spot" simply implies the energy to counter "off center momentum" comes from within the rod. In other words, it is "n+1/2" of resonance.

All sorts of things are possible with a slim shaft.

Yes. Safety glasses, and keep the heck away from it. You'd be lucky if it just beat the heck out of you. Quite an impulse available if / when something lets go.

Thanks. Double bearings at the middle "B-B" is where the driver is...

...

You might want to analyze the performance within the "many revolutions", to see how much the system is "ringing".

Then servos or stepper motors would / could do just fine.

You might look into a centrally located mass (between the middle two bearings) to act as a flywheel. This should help "quench" oscillations.

What is the driver, and how is it coupled to the shaft? And you realize that an electric motor applies torque in pulses, and these are transmitted into the shaft also...

No bearings and no shaft, with servo motors on each "end" is

*best*. Then you can stretch your optical workbench. Many early speed of light experiments involved synchronizing a slotted wheel to a "constantly rotating" faceted mirrored drum many kilometers away.

You might want to simply do a fourier analysis on what the angle difference is over time, rather than trying to smooth over it right away. Knowing the frequency contributors might help you locate and minimize their effects.

But a large diameter hollow shaft could be a big improvement. Be sure and get seamless tubing. Now understand that you will be adding another oscillation mode, namely "periodic ovoiding", which normally would be much higher frequency than "guitar string oscillation" (on the order of the torsional oscillation)... but will be driven by this type of oscillation.

David A. Smith

On Mar 10, 6:59 am, "colin" wrote: ...

...

The point of the supercharger is the problems you're describing have already been solved in these things. High speed shafts taking up to

40hp to turn that maintain tight tolerance and a minimum of vibration

Here is a link to a picture of the input shaft for a blower and the bearings used in the rebuild kit for them. these do run in a gearbox that contains a synthetic lubricant with tight control on the amount of expansion do to heat since the case is sealed..