[sillythings]

Thanks Eldon!

Like most problems of this nature, I am positive that the mini mill column flex problem also follows the 80/20 rule as well. By that I mean the bulk of the rigidity gain can be achieved with a small amount of effort to strengthen a few key locations. Besides, I am lazy by nature and I derive extra pleasure out of avoiding labor!

Anyway, the analysis done so far has given me a deep understanding of the issue and my confidence is growing. More to come.

There is one thing that I would like to ask you and others.

It would be great if more folks will take a little time out to measure the flex of their machines and share it. This will give everyone some idea of the level of improvement they can expect given a certain level of effort and access to shop equipment. I, for example, only have a mini lathe, mini mill and a small band saw. The measurement method I documented here is easy and you don't need to take apart the machine.

[sillythings]

Plan of Attack A

Now we are get into the exciting stuff!

Please note that not only will I share my successes, I will also share my failure. This way, readers of this thread can read about my failure and will not need to repeat my mistakes firsthand. As I am writing this plan, I don't know if this plan will work. Please stay tuned


Before I begin, here is some additional commentary on the source of flex analysis that I've posted previously. Here are the numbers:

Attachment 250176

flex at angle bracket base (green box): 0.0015"
flex at joint area (blue box): 0.0007"

I'd like to emphasize the numbers are much worse than they look. There are 2 reasons for this:

• the column bends both backward and forward (i.e. multiplies by 2 the flex reading)
• the position of spindle is higher than where I measure with my indicator dial (i.e. multiple by a length multiplier)

This is why my original base line measurement (first post) of flex is approximately +-0.004 (or 0.008 inches total) at the spindle. This is pathetic.


Plan A's Principal of Operation

This plan focus on improving the angle bracket to base attachment. The flex here accounts for ~68% of total flex. Therefore, it is the biggest fish to fry first!

To begin with, the following illustrates, with exaggeration, how the flexing works at the angle bracket base.

column bending backward:


Attachment 250178

column bending forward:

Attachment 250180

The problem is the width of the angle bracket base is too small (only 1.6" on my mill). The angle bracket base can be thought of as a lever that resists the flex of the Z column. As physics dictates, a short lever is weak!

How do we strengthen this situation? Archimedes would have answered "go get a long lever!"


Attachment 250182


Here is how. Add 2 side supporting blocks (the big grey block in the picture) bolted to the base of the mill and to the side of the Z column. Once the whole assembly is tighten down, we will have practically (I know it is not exactly true) triple the length of the lever arm! I am particularly pleased that the new total lever arm length is considerably longer than the solid column kit from LMS!

The key to the improved rigidity is the fact that the side supporting blocks are directly tightened against the base (no air gaps!). With this change, the Z column cannot lift up and cannot press down. It should be obvious that this one enhancement addresses flexes in both the X and Y axis.

This enhancement will also reduce the flex in the joint area (blue box in first picture). Regardless, I do have another plan shimmering in my mind for that area.

To visualize how the forces are at work, take a look at the previous 2 image of the column bending backward and forward.



Construction Notes:

1. order of tightening matters
   
   1. first bolt the supporting block to the base by tighten the long screw
   2. second tighten the 4 screws to the side of the Z column
   
   
2. ​the green line in the picture are shims made of household aluminum foil. It is a simple way to trim the Z column in both X and Y axis
   
   1. ​add equal number of shims to both left and right side will bend the column forward, remove the same number of shims will bend the column back
   2. use unequal number of shims on the left and right side of the mill will bend the Z column left or right
   3. this should be much easier than the old way of trimming the mini mill.
   
   
3. ​the strength of the cast iron base is a concern. Not sure if there is a danger of cracking the base when tightening the long vertical bolt. When I build this enhancement, I plan to explore ways to add support to the underside of the base. Please let know if you have any clever ideas to strengthen the cast iron base

That is it for this post. I am itching to build this but my day job is keeping me busy. To be honest, I am really happy with the progress thus far.

Next post I'll report back on my experience building this and quantify the result.

[GITERDUN]

Seems like 10,000.00 worth of time to improve a 799.00 mill. How about you just bolt the top of the head to the wall behind. If you want that angle feature, flame an arc in a steel plate bolted to the wall.

[sillythings]

Seems like 10,000.00 worth of time to improve a 799.00 mill. How about you just bolt the top of the head to the wall behind. If you want that angle feature, flame an arc in a steel plate bolted to the wall.

Any idiot can spend money. Many good and decent folks across the world spend untold hours to build little silly things for its own pleasure. My thread is for these people.

Really don't worry your little head, you don't need to "get it".

[GITERDUN]

Silly,
If the exercise is to find the point of flex, anyone could see the problem at a glance. No need for any complicated calculations or measurements. If you are looking for a solution, it's a simple matter of locking the top of the column in place. Most people have 3 main assets- time, energy and money. When we are young we have lots of time and energy, but as the years go by those assets diminish and if we plan carefully, they are replaced by money. I would guess by this post that you are very, very young.

[Eldon_Joh]

if you like your tiltable mill you can keep your tiltable mill
http://www.hossmachine.info/images/b...s%20column.jpg

if you want even more rigidity you need to put a closed box beam around the back corner and connect it to the front of the Y axis casting too. (the open y axis casting doesn't have any rigidity for torsion loads, which are the bulk of the X axis forces

[sillythings]

if you like your tiltable mill you can keep your tiltable mill
http://www.hossmachine.info/images/b...s%20column.jpg

if you want even more rigidity you need to put a closed box beam around the back corner and connect it to the front of the Y axis casting too. (the open y axis casting doesn't have any rigidity for torsion loads, which are the bulk of the X axis forces

Thank you for the feed back. What do you mean by "open y axis casting"? Do you mean the underside of the mill's base is hollow and therefore not rigid?

Anyway, I think the mill's existing base can provide much better support to the Z column with some simple change. Soon I will build out my plan and I will compare the result with my baseline. The numbers will tell me if I need to do more for X and/or Y axis.

[sillythings]

Silly,
If the exercise is to find the point of flex, anyone could see the problem at a glance. No need for any complicated calculations or measurements. If you are looking for a solution, it's a simple matter of locking the top of the column in place. Most people have 3 main assets- time, energy and money. When we are young we have lots of time and energy, but as the years go by those assets diminish and if we plan carefully, they are replaced by money. I would guess by this post that you are very, very young.

Giterdun, you sounded like a miserable old thing and you are desperately clanking the last few coins in your pocket to the whole world as loudly as you can!

Go on miserable old thing! Spend your coins while you can! After all, you can't exactly buy your way into heaven now can you?!

Tick tock! Tick tock!

Oh heck, I'll let you in on a little secret. Many good people enjoy spending our time doing silly things that give us pleasure. It is a blast!

Sorry to hear you lived a miserable life time and still clueless nearing your last gasp.

Go on miserable old thing! Spend your coins like an idiot! After all, you can't exactly pay off the devil now can you?!

Tick tock! Tick tock!

Giterdun, ugly is as ugly does!

Tick tock! Tick tock!

You poor miserable old thing.

[steve123]

Thanks for doing this! Few people actually measure mill flex, even on DIY machines! The scientific method. What a concept! It looks like you will end up with a simple retrofit plan for this kind of mill. Many will benefit. The retrofit parts could be the first ones people make on their new mill.

Others should do the same work with different mill models. I will be doing this kind of testing as I build my mill.

P.S.
Some grumble about you doing this. IT'S A HOBBY! That means it can take up lots of time. IT'S SUPPOSED TO DO THAT. You don't measure this time with $$$. It's one of the ways you enjoy life. If you don't enjoy it, then DO SOMETHING ELSE.

[arvidj]

I also want to say Thanks!! and hope you continue to publish what you find.

[Gregj888]

Silly,

Great work, I too want to thank you for putting numbers on the this.

I have a $420 HF variation that has been scraped in to about the 80% level. One thing I did was to scrape in the contact points in he are area in question. No measurements, but a very notable difference. What was interesting to me was how little contact area there was between may of the components when I started. This helps with "rock" not flex of course, I still need to add stiffners. Make sure you pick a solution that allows column alignment in both X and Y (IMHO).

Thank you again, numbers are always better than hand waving,

[sillythings]

Hi,

Thank you all for the kind words and encouragement.

My day job has been keeping me really busy for the past few weeks. I finally have some updates for implementing plan A. Please see previous post for details.

Support Mounts

Attachment 251486


Attachment 251488


Making the left and right support mounts are straightforward. However, once I've taken the mill a part, I had no tools to drill accurate screw holes in the Z column and the base. I managed to get access to a very crappy drill press with a spindle that flops all over the place. I had to resort to using a round file to manually fit the pieces so that the screws can be tighten without too much side binding! What a real pain! I don't recommend using a hand drill to drill screw holes into the base and the z column.

Also notice I used 2 screws for the base and 2 screws for the z column. This has made the alignment of all screw holes that much more critical! This added complexity has turn out to be unnecessary. I'll explain below.


Measurements

The following are the preliminary measurements. Please keep in mind that since the machine has been disassembled these numbers should be compared with my earlier post on the source of flex and not the baseline numbers. Later on, I will put the whole machine back together and redo all measurements for the baseline comparison.


Attachment 251490

The angle bracket base flex is now 0.0005". The original flex is 0.0015" for a 67% reduction.


Attachment 251492

The joint area flex is 0.0005". The original flex is 0.0007" for a 29% reduction.

Overall, the reinforcement mounts provided a nice improvement! Granted, I know there are some errors in these numbers due to the precision limit of my dial indicator.

In a later post, I will compare and contrast the nature of the movement of the angle bracket and the joint area. The reinforcement support mount above is targeted at the angle bracket. I will explore ways to reinforce the joint area. There is probably more gain to be had.


Simplified Support Mount

Since I can't use my mini mill to drill my mini mill, I had a hard time with drilling the z column and the base accurately. I therefore wondered if I can use fewer screws and still get similar improvement.

Attachment 251496

Notice on the support mount, I used only 1 screw for the z column and 1 screw for the base on each side. With this simplify configuration, I measured the angle base flex and it is the same 0.0005". I should have went this route from the start! I originally did worried about cracking the cast iron base and the z column. On hindsight, I think that concern is unwarranted.

[sillythings]

Today, I was able to improve the rigidity further by enhancing yesterday's implementation. However, I am also a bit stuck and can use some suggestions. I will describe the issue further below.


Anyway, here is the improvement to yesterday's post. As I mentioned in my post from yesterday, because of the cheap drill press I used, the screw holes drilled into the mill base and the z column were not accurately positioned and were also not perfectly perpendicular. As a result, when tightening the screws there were a lot of binding of the screw with the side wall of the holes in the side support mounts. Therefore, the support mounts were not as tightly screwed down as they could be.

Today, I've reassembled my mini mill. Using the mini mill, I enlarged the screw holes in the support mounting blocks so that its much more tolerant of inaccuracy of the screw positions on the base and the z column. There are no more binding of the screws when tightening.

Here is a new picture of the support mounts fully tightened.

Attachment 251624




Since I have the mini mill reassembled, I decided to measure the improvements and compare them to the baseline.



Attachment 251626

New front column flex 0.0015". Previous baseline 0.004". Reduction 63%



Attachment 251628

New back column flex 0.0005"! Previous baseline 0.0035". Reduction 86%

The side support mounts have significantly stiffen the forward and backward flex of the z column.



Attachment 251630

New left column flex 0.0005. Previous baseline was also 0.0005

Notice the 20LB weight is out of focus behind the dial indicator. Basically, the left side of the z column has always been very rigid.



Attachment 251632


New right column flex 0.003". Previous baseline was 0.0035". Reduction of only 14%



The right side is the problem direction that now needs the most attention. I really don't understand why the left side is so rigid while the right side is so much worse (7X!). I also don't understand why the side support mounts didn't help the right side flex more?

I have one wild guess. The tightening of the large nut in the back of the mill requires the nut be turned in a clockwise direction. Could the tightening of the large nut somehow preloaded the Z column in the left direction? Because of this "preloading", further bending of the Z column to the left will require much more force?


I'd like to fix the right column flex next since it is biggest problem. However, I am stuck. Please share your thoughts:

1. Why the large difference in rigidity between the left and right side?

2. Ideas on making the right side more rigid


In summary, I am quite happy with the overall progress. Three of the four flex directions are now pretty good.

[sillythings]

From my previous post, I was puzzled by the remaining large right column flex. I've now figured out why! However, to engage my reader in a fun puzzle, I will hold off disclosing my finding until my next post! Please take a look at the following pictures and share your theories.


Picture A:

Attachment 252186

Dial shows right column flex 0.0025

Side note: I've improve the contact surface of the support mount with the z-column and the base. This further reduced the right flex to 0.0025". Anyway, this thread is not about the 0.0005" improvement from my previous post.


Picture B:

Attachment 252188

Dial shows right column flex 0.001" !

The only difference between picture A and picture B is the position of the mill on the wood bench. In picture B, I turned the mill ~45 degree on the bench.

The question is why the huge 60% improvement?


I've tested my theory and have verified it. I am really glad that I have a reliable way to measure flex. Without a systematic way of measuring the flex, I'd never have guess the problem! Anyway, now I know and I have an idea for fixing it!

[rs4race]

Nice work on this study. Someone has to do the work, so why pay someone to do what you enjoy?

My guess would be that your benchtop isn't level or the mill base doesnt sit level. That leaves the mill tilted a bit and gives the weights a longer lever arm to act on the mill on the right side, on the other side the lever arm is shorter so less deflection. I'm guessing its more benchtop than the mill base since rotating 45 degrees helped. Did you try rotating it 180 degree from original and check deflection?

[sillythings]

Another Source of Flex

In my previous post, I've shown that just by rotating the mill on my wood bench by 45 deg, the right column flex was reduced by 60%.

This was the smoking gun that identified another source of flex in the mill. The mill base was twisting along the Y axis when a side load was applied! Since we are measuring movements in the thousands of an inch, the wood bench, wood being wood, will deform unevenly when uneven load is applied. The key is the mill base has to be rigid. As such, I got a large 3/8 inch thick aluminum plate from the local metal recycler Sims Metal for $2 a pound or $10 total to fix the twisting base. I'll explain why I didn't get a thicker plate further down.


I've put the mill on the aluminum plate and I bolted down only the front 2 screws but not the 2 screws in the back. This is so that the rear of the mill base can rotate relative to the front of the mill base.

As you can see in the picture below, the rear screws are not screw in.
.
Attachment 253670

The right column flex is 0.0023" which is very close to my previous post.

Now I tighten both rear screws and remeasure right column flex.

Attachment 253672

The new right column flex is 0.0007"!

I definitely didn't expect this when I started on this project. The cast iron base seemed quite beefy from casual observation!


Principal of Operation

The cast iron base is hollow at the bottom. Therefore I had high hope that if I close the bottom, the mill base will become a stressed skin structure (see google) and the rigidity of base should improve a lot. I am aware that the top of the mill base also has a narrower opening and therefore not the ideal stressed skin structure.

The reason why I didn't get a thicker plate or a steel plate for that matter is a stressed skin structure doesn't need a lot of material to achieve high rigidity (see airplane design). Besides, I always had a fine backup plan that rely on brute force which is to simply use a stack of plates if I am not happy with just one!

I've also cut another corner. To maximize the stressed skin structure, I probably should use a couple of more screws, in addition to the 4 that I used, to bolt the plate to the base to more evenly distribute the load. I am too lazy!


Here a shot of the base plate all bolted down.
Attachment 253642


Comparison to Baseline

I've remeasure all flex directions and compared with my baseline. Please see my first post in this thread for the baseline number.

front column flex

Attachment 253648

current : 0.0007"
baseline: 0.004"
reduction: 82.5%


back column flex
Attachment 253674

current: 0.001"
baseline: 0.0035"
reduction: 71%


left column flex

Attachment 253650
current: 0.0005"
baseline: 0.0005"
reduction: no change, has always been pretty good.


right column flex
Attachment 253672
current: 0.0007"
baseline: 0.0035"
reduction: 80%


Conclusion


At this point, I am quite pleased with the improvements! I plan to work with the mill a bit first before trying more improvements. There are couple of areas still on my mind. For example, the twist in the mill base reminded me that the Z column may also be susceptible to twist. In an earlier post, I've verfied z column bending isn't an issue. However, I didn't check z column twist. Another area that may need more attention is the joint area (i.e. the large bolt and nut). On the other hand, for further improvement, now it is even more critical to measure.

I'll probably write one more post to share my thoughts on other attempts found on the web for improving the rigidity of the mini mill and why I like my solution the best :-) Heheh, vanity!

[sillythings]

Nice work on this study. Someone has to do the work, so why pay someone to do what you enjoy?

My guess would be that your benchtop isn't level or the mill base doesnt sit level. That leaves the mill tilted a bit and gives the weights a longer lever arm to act on the mill on the right side, on the other side the lever arm is shorter so less deflection. I'm guessing its more benchtop than the mill base since rotating 45 degrees helped. Did you try rotating it 180 degree from original and check deflection?

rs4race, I think you are right. I didn't check but most likely my mill's back left corner is slightly lower ( a few thousands of an inch) than the back right corner. Because the back left corner is longer, it provided better support against left column flex. Similarly because the back right corner is slightly higher, there is less support for the base and therefore the base twists. Thus the bad right column flex. This is like a 4 leg table with one leg slightly shorter.

In summary, there are 2 problems here:

1. the back left and back right of the mill base is not perfectly level
2. the rigidity of mill base, because of the open bottom, is not very strong and allows the base to twisting in the Y axis

This combination of issues explains why the left column flex is so much more rigid than right column flex. It makes a lot of sense to me! Also, just to be clear, this slight unevenness of the base cannot be measured on a wood bench since wood will compress a few thousands if you just put your thumb on it.

Thanks rs4race for the feedback!

[jdrew1]

Great write up Silly. I've been thinking about doing blocks like that in addition to my back plate for a few months now. I never measured my deflection and I don't have a tenths indicator yet anyway, but I definitely saw the quality of my cuts improve with the plate. I'm looking forward to your conclusions on column twist.

[sillythings]

rs4race,

You mentioned in your post that either the wood bench or the mill base is not level and that your bet is the table. However, my previous reply is that I think it is the mill base. After thinking this over, I think my previous reply is incorrect.

The mill base, if it is not level to begin with, it's probably only a few thousandth of an inch off. However, the wood bench is wood! The flatness of the top of the wood bench probably varies 10 times that of the cast iron mill base! Doh! Attachment 254434

Furthermore, I notice the left side of my mill is close to the left end of the bench (i.e. near the bench legs) and the right side of my mill is closer toward the center of the bench. Most likely the wood bench bows downward in the middle due to the weight of the mill.

For folks planning to measure your mill's flex, please always position your mill in the same location on the table for repeatable results.

[sillythings]

Source of Flex Episode III: Twist Again!


In my previous post, I demonstrated the mill base is susceptible to twisting. I really had hoped that the z column would not have this issue since the z column bending was not an issue. Unfortunately that was not to be. The following are the baseline measurements of the z-column twist .


Counter Clockwise Twist
Attachment 254448

Measure from: mill base
Twist amount 0.0015"


Clockwise Twist
Attachment 254450

Measure from: mill base
Twist amount: 0.0015"

The amount of twist is the same for both clockwise and counter-clockwise.

Also please note that the mill beds are locked to the mill base. This is why I say the measurement is from the mill base even though the magnetic base is on the mill bed.


Identifying The Source of Twists

To identify the source of the twist, I've made multiple measurements from different locations

Attachment 254452

Measure from right angle bracket
Twist amount: 0.0012"
Notice, this is closer to the top of the spindle then the previous measurement (mill base).

Attachment 254454

Measure from lower z column
Twist amount 0.001"
Notice this is slightly closer to the spindle then the previous measurement (right angle bracket).

Attachment 254456

Measure from mid way on z column
Twist amount 0.0005"
Notice this is half way closer to the spindle the the previous measurement (lower z column)


The z column can twist +-0.0015" in both clockwise and counter-clockwise direction (or 0.003" end to end). The above series of measurements show that the amount of twist increases linearly with the distance of the magnetic base to the spindle. These measurements show that the twist is primary caused by the z column. Importantly, the silver lining, the joint area is not a significant factor.

I like to improve the Z column twist.


Plan of Attack

There are couple of options.

1. Bolt a rectangle pipe to the back of the z column
2. Create a torque tube from a thck plumbing pipe and use JB weld to fit it inside the Z column. I haven't figure out exactly how just yet.
3. Fill the z column with engine block filler

Attachment 254460


#3 is the most rigid and probably the easiest to implement. However, I do need to drill holes and modify the z column in the future. Does anyone know if engine block filler can be drilled and machined after it has cured?