precision shaft, final challange - balancing.

Hi, Ive got my 1.2M long 12mm shaft running quite well, the standard deviation of position from one end to the other on my encoders
is about 2x10-7 (less than 1 arc second). this is a reasonable improvment over my 5mm shaft wich was about 5 times this or more.
It runs quite well at <5000rpm.
The shaft is supported at each end and the motor in the center also supports the shaft but it is rubber mounted to reduce noise and torque pulses.
Despite being a linear shaft it was 0.2mm out of true in the center, I dont know how acccurate its suposed to be, im waiting for a reply from the stockist, however I managed to straighten the shaft to within 0.01mm at any point, much less than 0.01 in the center, if I try and get it any straighter it tends to end up further out of true the opposite way.
I managed to find some 12mmID 18mmOD bearings on ebay, managed bore out the casing with less than 1mm to spare.
however the rotor in the center obviously needs balancing, after I drilled out the center to take the 12mm shaft, although it looks true, it vibrates at moderate speed and resonates at about 6500 rpm, at 5000 rpm it isnt too bad but My aim is to get it as smooth as possible. final running speed will be the optimum of (sd of angle/speed^1.5)
So Ive drilled and tapped 3 holes to put grub screws in to act as variable weights, dont worry il eventually glue them or something to make sure they cant fly out.
however ive been playing around a bit now and cant seem to make things better, so I googled for dynamic balancing but all I can find is mostly waffle about what it is and how people use equipment to do it and how they use weights, but not the hands on nitty gritty.
So, so far im using a felt tip pen to identify the side of most deflection, however I realise that position is a result of integral of velocity wich is in turn an integral of acceleration wich is governed by the balance weight, so is it going to be 180' out of phase to the side of least deflection ? or closer to 90 if the the rubber mounting is resisting deflection ?
and how does this change around the critical speed ?
thanks Colin =^.^
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
@ntworld.NOSPAM.com says...

That's right at the limit of Thomson's standard spec - .002 inches / foot.
Ned Simmons
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
says...

aha ic thanks. didnt realise the error added up to so much over 1M. so I gues even with a higher precision shaft it might be hard to get better then 0.01mm run out in the center.
Ive managed to balance the center of the shaft so it runs with very little vibration at 5000rpm, I just put a very large grub screw and noted where the high point was in relation, then worked backwards, initial problem was my grub screws were too small to make any noticable difference.
Ive had a further look round and ive noted references to acceleromters and polar charts, presumably this chart gives the relative angle from high point to where the weight is needed, my felt tip pen kinda does for the acceleramoter lol.
however I notice the middle of the unsuported sections are deflected widly at 6000rpm, I gues this is whirl mode, so I need to add some balancing here, whats a quick way of adding balance weights securly to a 12mm shaft ? I dont realy want glued on weights. maybe something like a circlip or two ?
Im thinking of making a loose fitting colar and grub screws to adjust the CG. I have a hefty roller from a photocopier wich happens to have a bore of 12mm. although its probably too hefty.
Colin =^.^
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 30 Apr 2007 14:34:18 GMT, "colin"

..
Perhaps this whirling shaft demo and manual might help?
http://usdidactic.com/refpdf/gunt/04062500%202.pdf
Brian Whatcott Altus OK
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Have you considered a strobe light for finding the angle that produces the most deflection? Richard
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

thanks, yes I have a strobe led driven from the motor controller, however it never seemed to show anything. but with the deflection so great it can be seen with the led, a felt tip pen works probably just as well.
from what ive read it seems I need an accelerometer or vibration sensor, maybe il have to buy/make one and hook it up to the controller. anyone know where to get these or if theyr expensive ? I could rig up a coil with a suspended magnet or do I need a qood quaility one?
would be cool if it dynamically self balanced itself.
Colin =^.^
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

ok so ive got some vibration sensors now, and a strobe led wich fires at the peak output from the sensor, the signal is processed with a dsp to generate the strobe.
the shaft is 1.2M long with 3 shaft bearing blocks mounted on 2"x4" stainless steel box section, wich is rubber mounted by 4 points to a 2"x4" alu section and I have a place to put sensors at either end and the middle, I can measure any 2 at once.
the vibration sensors are peizo sounders, atatched to the isolated beam, and a bolt screwed up from the lower non isolated beam to make contact with the inside of the piezo disc, so it gets a very strong signal of the movement of one beam relative to the other.
it resonates at 5400rpm, I find that the peak position moves about 180' from one side of resonance to the other, and the difference between forward and reverse is upto 180' depending on speed, however at some speed just over resonance I think the spot will be in the same place in either direction, also it does wander quite a bit as resonance builds up over a few seconds.
So how do I work out where to put the weight to balance the shaft ? im just playing about at the moment, and adding some weights notcibly reduces the vibration, but the position wanders about too much to make a connection between the spot and where to put the weights.
ofc I dont know what polarity of the piezo transducers are either so this may also be 180 out of phase.
however it does run quite sweet with just a couple of 3mm nuts taped onto the shaft with pvc tape although thats not very safely secured.
do I have to vary the speed from one side of resonance to the other in forward and reverse and process the vibration signal to find the mid spot ?
are there any pitfalls in trying to balance for all speeds in either direction ?
thanks Colin =^.^
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
...

Have you considered removing material, rather than adding material to balance it? when balancing a automotive crankshaft, the material added is actually mallory metal which is heavier than the original metal it replaces.

Interesting website I found: http://www.brad.ac.uk/research/mbdmst/rotordynamic/unbalance/unbalance.html
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

I dont want to remove metal as it would weaken the discs, and i might get it wrong , then i would end up with several holes, plus Im still needing to take it apart to modify it every now and then.

thats realy cool thanks, altough it covers the fundamental physical principals such as centripetal force etc wich im fairly familiar with, the links on that page dont seem to be working atm.
I think the case of the vibration signals is far more complex however as there is resonace etc. im only just begining to understand this, I initialy thought it was the shaft acting like a violin string but now realise its dependand on the mass and stiffness of the all the components involved.
they mention bearings mounted on springs, however ive got all 3 bearings mounted on a 2x4 steel beam, wich is itself rubber mounted, with a sensor on the beam under each bearing. I can feel the vibration but this probably makes it more dificult to analyse, im not sure if I can use springs as I have to dismantle the shaft to remove the springs and re align the bearings without the springs. however the beam its mounted to is also vibrating wich probably is rather unhelpfull, I gues il have to clamp it to something more solid.
thanks Colin =^.^
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 22 May 2007 15:30:33 GMT, "colin"

The "homebrew delight" method is to run up the speed just past the first resonance, then look to add weight opposite that point of max deflection. Varying the weight for minimal vibration comes next. Then spin the shaft up past the second resonance, and place a small weight at that point of max deflection on the shaft.
Repeat....
Brian Whatcott Altus OK
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote in message

cool thanks for that :) ive noticed if I get the speed right the spot is the same in both directions but its quite sensitive, its just over resonance just as you said. seems my led flashes at the point where I need to add weight at the top of the shaft wich is convenient.
however after ive added a few weights the spot just wanders round the shaft, yet there is stil a fairly strong vibattion. maybe my speed isnt totaly steady, its crystal controlled but maybe i need to tweak the control feedback loop a bit.
looking at the vib sensor there is fundamental resonance at 5400 rpm = 90hz, and at 1/3 or 1/5 that speed the vibration signal contains 3rd,4th harmonic of the shaft speed, wich makes it stil the same 90hz.
when you mean 2nd resonance would that be 2 or 3 times shaft speed ? I can manage 10krpm, 15krpm and its probably gona be singing quite a lot, although if I could get it running realy sweet at that speed it would be cool :)
i havnt run it much past this yet although it did run away and got to 16k rpm before I hit the off button lol ! the electroncs wil easily goto >30krpm, im not sure what speed it will fly apart. il put some sheilds around the discs.
my motor wuch is in the middle of the shaft is from an old large disc drive, it has 3 grub screws to centralise it on the shaft, and 3 more just to act as weights, the discs at the end also have 3 grub screws, and il think il try making different length screws to redistribute the weight.
Colin =^.^
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.