Up grade milling machine

A video supplement to Jason's valuable contributions to this thread can be found here.

Yer, I had bought one cutter - 3 flute as I read in someone's book that it was a good all rounder

So many time people discuss rigidity of machines and spindles but strangely no one attempts to measure this significant attribute in order to make the discussions more objective. Its not hard to do. I have done it on my own Wabeco mill and the result is shown below. All I did was put a short stub in the collet with a DI earthed to the base. I then pulled the stub with a spring balance in the direction I wanted to measure the rigidity, then plotted a graph. You can see there is a considerable difference depending on the direction of pull.

The reason I did this was that I planned some simple improvements to the machine and wanted to make sure I was getting a quantifiable return on my investment. Unfortunately, with so many interesting things to do, this work is still in progress. One day I will report the conclusion but I can tell already that some re-machining of parts of the machine has improved my ability to work with tougher materials.

But as has been said in this thread, high rigitity is not crucial if one is prepared to compromise on the time it takes to do a task. Its more of a 'nice to have'. It would be good though if suppliers made some attempt to quantify rigidity of their products to help in comparing machines at purchase time, instead of us relying on folklore and hearsay.

Gerry

The deflection of the spindle should be the same in all axis when measured. Any excess difference could be due to movement in the housing against ts mounting. When spindle accuracy is measured it needs to be on the tool mount face/bore when the spindle housing is sitting on surface plate. Any movement can then be attributed to spindle bearing clearances and nothing else. When seeking true runnng the final piece is the tool holder mount, some of which are designed to be capable of setting in situ on the spindle to get near zero run out. Where I worked all spindles were measure before building, then again when built, then finally after extended running. Final accuracy can only be determined after getting the assembly up to its normal operating temperature as expansion creates a totally different set of results, though hopefully only a minor change in any direction.

That's true for determining if there's an issue with bearing/quill fit, but rather misses the point of considering the rigidity of the machine relative to the workpiece.

Gerry's experiment was to measure the total deflection of the spindle relative to a workpiece when the overall machine structure was loaded.

Because the machine's body has different levels of rigidity in different planes as an inevitability of the asymmetry of it's construction, it would be fully expected to show a measurable difference.

​

The differences highlighted by Gerry's experiments on static rigidity are only the start of the story, too the effective rigidity when exposed to dynamic loads which induce oscillations is influenced by many more factors than just geometry, mass and young's modulus as you might expect for static loading.

But those kind of dynamic rigidity measurements are much more complex to undertake; needing reliably synchronisable time series data from several strain gauges, a method of measuring vibrational frequency in multiple planes, and a way to induce said vibrations, then data processing to marry up of all that data and disentangle the various effects from each other.

It's a major area of research in the design of high speed and ultra-accurate machine tools, with significant amounts of manufacturer resource poured into better understanding and modelling of the underlying phenomena and the ways machines can be designed to avoid (or work with!) them.

Jelly, yes a static deflection measurement is just 'part of the story' but both static and dynamic deflections in a given structure are of course intimately linked. If I could halve the static deflections in my Wabeco mill spindle, I would get two benefits. The first would be achieving a given dimension on a work piece with perhaps fewer passes. The second would be improved surface finish, as a consequence of lower dynamic amplitudes.

Its clear to me that from the magnitudes of the deflections in my mill, the weakness is not the spindle bearings, the spindle flexing or clearances between bearings or housings etc. Its mainly the supporting structure. I'm sure if the manufacturer had done a little bit of FEA work on this he could have optimised where he put the metal and given me a much better machined for the same cost.

Gerry

Fully agree, the spindle assembly is very likely the most rigid element with respect to lateral deflection in the whole assembly, whilst the head-column slideway and column-bed connection are very likely sources of deflection.

I'm pretty sure there's a Stephan Gotteswinter video about trying to eliminate some of the deflection in his old Optimum MB4 mill (which is very similar in design to many of the Wabeco mills) using epoxy to improve the fit and bedding of some parts, which suggests that the inherently compromised design is exacerbated by "good enough" machining of mating surfaces too.

Reducing static deflection would (usually, absent unlucky resonance effects) result in reductions in dynamic deflection too, although I was mentioning it in an attempt to illustrate to Howard that if you were measuring that kind of deflection statically, there was potentially even worse real world deflection at certain cutting speeds which were unfavourable (the same being true of much bigger more heavily built machines as well)

I would be interested to see a comparison of the dynamic deflection in a mill like yours to a more old school milling machine design (which didn't have the benefit of FEA, but does have lots of material to increase stiffness and/or simply absorb and dampen vibration)...

I have a suspicion that the older design would still have dynamic deflection, but simply with the periodicity shifted such that it had less impact on surface finish in the anticipated operating range of RPM/Feeds

Edited By Jelly on 14/12/2022 00:09:54

Having done a lot of vibration testing in my time … I am watching this discussion between Gerry and Jelly with great interest and enthusiasm.

It’s good to see Gerry doing some well-planned practical testing of his machine.

Allow me to just introduce one ‘benchmark’ that demonstrates what the structure of a generic small milling-machine could be. … Have a good look at the BCA Mk 3 : it’s all pretty light ‘at the sharp end’ but that foundation is superb.

**LINK**

http://lathes.co.uk/bca/page4.html

MichaelG.

Yes, the lack of a quill on machines like the BCA will go a long way to eliminating movement at the tool. Combine that with no joint between bed and column and the two main sources of movement are taken care of.

I doubt the modern benchtop machines have much flex in their actual columns and most of what can be detected is due to the joint between the column and base.

I used to live with one of the Senior Applications Engineers for a company which delivered fatigue and durability modelling of products based on sensor data, and one of my missus's friends is a specialist in vibration analysis for advanced machining at Boeing... I'm working with knowledge which is the scraps from their tables so to speak.

In my professional engineering career the extent of vibration analysis has tended to extend as far as:

• "Pump's loud, can you get a fitter to come look at the bearings on the back shift", and
• "Pump sounds funny I think that's cavitating, radio the board operator to find out why upstream pressure is dropping".

​

However, I'd definitely be interested to see if I can source reasonably priced components to cobble together a basis vibration analysis rig, although it is decades since I last worked with strain gauges during Physics A-Level so there's probably quite a lot of basic learning to re-do as well.

.

I’ve been pondering, for a while, how we might get best value-for-expenditure [of not just money] by using consumer-level sensors and instrumentation … and I think the best area to explore might be ‘modal analysis’

May I leave that thought with you ?

MichaelG.

.

Skip lightly to this section:

https://en.wikipedia.org/wiki/Modal_testing#Impact_Hammer_Modal_Testing

… and think how relatively simple the necessary kit is.

Edited By Michael Gilligan on 14/12/2022 16:10:18

Thanks Michael for the link to the BCA jig borer. Very nice neat machine and because of its small dimensions very rigid I should think. A very limited size of workpiece goes with it of course, probabaly too small for most hobby engineers.

Looking at the larger of my milling machines I often have to resist the temptation (in order to get other things done) to do something about the collumn. It's a channel section casting open at the back but closed by a thin plate held in place by some small screws tapped into the casting. The holes in the plate are clearance. From courses attended many years ago I know (without having done the sums) that the torsional rigity of the collumn about its vertical axis would be increased many times if the casting had been a box section (a la BCA). This obviously would in turn have a major impact on tool/workpiece stiffness in the X direction. So I ask myself does the thin plate have a measurable stiffening effect and would it be worthwhile cooking up something more robust to replace it ?

My feeling re. relative stiffnesses of joint and collumn has been the opposite of Jason's but it would take a lot of work to get at the truth of the matter. This question will undoubtedly have been addressed by makers of professional machines but less likely for our sizes and even less likely to have been published. But I am happy to be proved wrong.

For Gerry's test what would be a 'reasonable force' to apply to a mill by size eg WM14, Minor, Major, Bridgeport ?

On the round columns it might be interesting to see if the head suddenly rotates!.

Would a microphone and a spectrum analyser be of any help, perhaps revealing a resonant frequency that equates to rigidity?

.

If … and it’s a very big IF … a thin plate is firmly attached to a channel section, it can make an enormous difference: effectively the diaphragm provides an infinite number of ‘spokes’ in tension [and an infinite number in compression, but those are of little use] … so you have instantly added multitudinous triangulation!

For a very simple demonstration: look at self-assembly book-cases … the back-panel does a lot of the work.

MichaelG.

Yes I think small mills have two problems. Jason touched on one, the joints. The other being the basic structural rigidity of the machines elements. It was joints that got me started on this project. Under certain cutting conditions I could almost see a tendency to rock where the head fixed to the column. Dismantling showed a pretty un-flat and poorly machined surface. I thought this would be a quick and easy fix and took the offending part to someone who agreed to grind it flat. When I collected the part, I found it had been fly-cut not ground! I wasn't too happy about this. It certainly looked nicer but when measured at home it still wasn't flat. Not only that but it appeared to be more out of square than it was originally. As happens so often with me, I decided that if you want a job done properly one has to do it oneself and it was this that drove me towards the purchase of my J&S grinder. So before I can investigate and sort the mill out properly, I have to finish the grinder refurb. Before I finish the grinder refurb I have to get my new workshop built. Hopefully 2nd time around with our local planning people will see this kicked off early this year and everything should then fall into place Therefore, as regards the mill issues, as Captain Oates said, I may be some time.

Getting back to the point, in other parts the Wabeco mills claimed 'accuracy' appears to come from the insertion of tiny shims within the joints to ensure good geometry between the various machine elements. Under light loads this is fine. Unfortunately these shims have not always been very strategically placed and have been closer to the neutral axis of a bending plane, instead of close to the extremes of the section of the member. This compromises the stability of the joint.

To do the shimming job properly would require a lot of measuring, a whole bunch of graded shims along with huge amounts of time and attention to detail, so its not at all surprising that it doesn't get done well on a hobby machine. I will therefore try to eliminate a lot of this shimming by more careful machining when I 'up-spec' my mill. Where I still need to add shims I will try to optimise their positions for stability. Once I have got this primary failing sorted, I will look at Gotteswinter style infilling with a granite~epoxy.

Of course, none of this effort is really necessary, I could just cut slower instead. But like many on here I'm just a bit funny in the head and such things just have to be done. I'm quite sure though I will never get my Wabeco to the standard of those BCAs. They are such elegant machines indeed.

Gerry

"For Gerry's test what would be a 'reasonable force' to apply to a mill by size eg WM14, Minor, Major, Bridgeport ?"

This is a very good question indeed Bazyle. In some ways, perhaps it doesn't matter because one is trying to measure the rigidity, not a deflection. So providing the rigidity is linear (as mine was) using similar load levels to mine should give us an interesting comparison.

On the other hand I can imagine some big machines being so stiff that you might only get a reasonable deflection (say 25um or 1thou) if you apply a pretty hefty load to the spindle.

A more technically-based suggestion could consider the tangential cutting forces that the machine can generate, which would depend on motor torque and gearing.

I think would tend to go with a suck-it-and-see approach initially.

Gerry.

Gerry, you have a PM.