What’s causing these bands on a turned rod?

[SillyOldDufferModerator @sillyoldduffer]

Posted by Joseph Noci 1 on 21/06/2017 22:25:35:

Posted by SillyOldDuffer on 19/06/2017 11:02:02:

Posted by John Horne 2 on 19/06/2017 09:37:49:

…
I agree with Mikelkie, I think its a resonance/vibration thing, I also think it can have a number of causes which can be transferred to the job through a loose gib etc. but I think there will be a root cause like a vibrating belt or some numpty forgetting to tighten down the feet

Edited By John Horne 2 on 19/06/2017 09:39:49

Edited By John Horne 2 on 19/06/2017 09:42:47

I agree. I've stalled on this one for the moment. I have an oscilloscope that, in theory, can separate out the individual frequencies from a mixture of vibrations. I thought it would be dead easy to stick a microphone on the lathe, press a few buttons on the oscilloscope, and identify the root cause from the period of the strongest vibration.

Silly me. It turns out I don't understand the oscilloscope display even after re-reading the manual.

When it comes to setting up lathes I've not found much advice on minimising vibration. I'm starting to think vibration is the elephant in the room, perhaps causing more trouble than generally realised.

Dave

Dave, what type of 'scope do you have? Is it a 'modern' digital type? If so, it will most probably have some Maths functions, one being the ability to display the Fast Fourier Transforms of observed signal inputs. That will show clearly the amplitude and frequency of all signals seen by the microphone. You can then place the microphone on the toolpost and do some cutting and see exactly what the vibration frequencies and amplitudes are.

Signals displayed thus , in the frequency domain, show as a series of frequency 'bins' , starting from left, lowest frequency, going to the right, the highest frequency. The display would then draw a line showing the relative amplitude at each of the frequency bins. You would then easily see at which frequency the highest level occurs.

Looking at the signals in the time domain only, ie, such as a pure sine wave on the scope display will only show a mishmash of frequencies all mingled together at various amplitudes – the brain cannot FFT that! Impossible to single out a single culprit frequency, unless it is huge and dominant.

If the scope is not a digital type…there is an almost easier way!

Use the sound card input to your laptop, and feed the microphone signal in there. Download one of the many free Audio 'Scope/Spectrum analyser tools from the SphincterNet and you can easily see the FFT and anything else you want. I use that a lot for vibration analysis and it works very well. Google 'soundcard Spectrum Analyzer' or such like…

Regards

Joe

Hi Joe,

My scope is digital. It's a Siglent SDS 1072CML, which is a LeCroy clone I think. My problem with the scope turned out to be mostly self-inflicted, although the FFT section of the User Manual takes some of the blame. FFT is basic on this model and setting up is a little complicated. Most serious, I missed the instruction to centre the waveform on zero and not doing so wrecks the calculation. Secondly, you have to engage 'Math' as the signal source before the cursor functions work. I've had it working with a signal generator, but am temporarily blocked by losing my microphone. I put it somewhere 'safe'.

Using a PC soundcard with free spectrum analyser software is a really good idea. when the microphone turns up I shall try it.

I also tried making one of these, a moveable brass weight on a length of piano wire with a solid base, it's an upside down pendulum:

The idea is to put the thing on a running machine and then tune the brass bob for maximum wobble. At max wobble the pendulum should be resonant with one of the lathe vibrations, and the frequency can be calculated with the pendulum equation. So far I can't get it to work! It might be that the piano wire is loose in the base.

Dave

[SillyOldDufferModerator @sillyoldduffer]

After further investigation today I found the root cause of my banding and vibration problem. Careful measurement of the bands revealed that they are 15 revolutions wide which pointed at the compound gears.

Disconnecting the gears by swivelling the banjo reduced gear noise considerably. It also eliminated the vibrations visible in the 'saucer-scope' test.

Turning the chuck by hand showed that the gears had a tight spot. I had set the gears up using a single layer of thin photocopier paper between the teeth. I re-spaced the gears with a double layer of thicker paper and increased the gap to about 0.25mm. Now the gears are much quieter, the tight spot has disappeared, the lathe no longer vibrates and the bands have gone.

I guess one or more of my gears isn't quite round causing them to bind periodically as they turn. I suppose it's obvious that the forces involved between turning gears are considerable: there's no way I'd stick a finger in them. Even so, I'm surprised to find that a few gears being too close together was shaking the whole lathe (280kg) and spoiling the finish.

This is one up to common-sense. It's ironic that identifying and fixing this problem didn't need complicated high-tech methods involving microscopes, FFT analysis and an oscilloscope. No matter, playing with the toys kept me amused and I got there in the end.

Dave

[Author]

Posts