Deflection of tool vs lateral forces

[PaulRowntree]

Greetings All;
I am trying to be quantitative about finding and reducing the flex in my home built 4'x2' chain drive system. I am asking for your help in determining what is 'normal' for working machines of this size.

I am measuring the lateral deflection of the router tool using a dial indicator mounted to the working surface, while applying the lateral force using an electronic fish scale, also connected to the tool. I expect that a measured weight running over a pulley would work just as well. All electronics are on, with the stepper driver holding the motors still.

My machine now gives 0.010" of deflection with a transverse 'force' of 10 lbs, and is roughly the same in the X and Y directions. This is 0.25 mm for a force of 45 N for those so inclined.

I would greatly appreciate any input from the forum that would help me define 'typical' machine properties.
Cheers!

[louieatienza]

Greetings All;
I am trying to be quantitative about finding and reducing the flex in my home built 4'x2' chain drive system. I am asking for your help in determining what is 'normal' for working machines of this size.

I am measuring the lateral deflection of the router tool using a dial indicator mounted to the working surface, while applying the lateral force using an electronic fish scale, also connected to the tool. I expect that a measured weight running over a pulley would work just as well. All electronics are on, with the stepper driver holding the motors still.

My machine now gives 0.010" of deflection with a transverse 'force' of 10 lbs, and is roughly the same in the X and Y directions. This is 0.25 mm for a force of 45 N for those so inclined.

I would greatly appreciate any input from the forum that would help me define 'typical' machine properties.
Cheers!

Pictures of your machine for reference? I haven't made any measurements, but with my machine I cannot get it to stall or visibly deflect if I leaned on the gantry or carriage while it's moving. That's 425 in oz steppers direct-driving 8-start leadscrews (1 tpi)...

[mlabruyere]

My gantry is a 3030 65" beam of extruded aluminum with a piece of cold rolled steel plate at .25"x4"x57". The steel plate is attached with 40 5/18" bolts. The beam is supported at 7" and 58" (sits on top of 2 3030 supports). At 10 lbs of force with the Z-Axis at the center of the beam I can't measure any deflection.

On paper a bare 3030 beam with 51" unsupported should deflect .002" at center with 10lbs of force. Since mine has the laminated steel I'm sure that has added a lot of rigidity to the system.

[PaulRowntree]

My gantry is a 3030 65" beam of extruded aluminum with a piece of cold rolled steel plate at .25"x4"x57". The steel plate is attached with 40 5/18" bolts. The beam is supported at 7" and 58" (sits on top of 2 3030 supports). At 10 lbs of force with the Z-Axis at the center of the beam I can't measure any deflection.

On paper a bare 3030 beam with 51" unsupported should deflect .002" at center with 10lbs of force. Since mine has the laminated steel I'm sure that has added a lot of rigidity to the system.

Thanks for the info. To be certain that we are discussing the same thing, are these forces applied to the tool, and are you measuring displacment of the tool relative to the table? My gantry is not as strong or large as yours, but it also shows no deflection if I apply the force to the beam itself.
Cheers!

[Treischl]

You mentioned you were measuring the deflection at the "tool", which I will take to mean you mounted something like a dowel pin in the collet of the router? If you did, hats off to you!

Most people just think in terms of how much the gantry beam is going to deflect. But there is a lot more to it than that.

First, you have the bearings in the router itself. The Z axis has bearings and also can have a bit of deflection. The Z is riding on the beam with bearings, more clearance. Of course the beam has deflection. Then the gantry is riding on bearings. It may or may not be on risers which can deflect.

You also have some slop in the drive mechanisms, whether it be R&P or screw driven. If you ever notice, they are termed "anti" backlash as a rule, not "Zero" backlash. (Zero backlash is available but is usually only available in expensive industrial grade equipment).

Then we have the tool itself. Tools deflect. Even on industrial milling machines they deflect when cutting. It is one of the reasons why machinists and toolmakers take roughing passes and finish cuts.

You have to have all of these clearances or else the machine would not move at all. If you "tighten" up the machine, then you will encounter greater wear and it will require more power to move the machine. It is a balancing act.

So, the idea is to rough out your material, then if you are doing fussy work, take a finishing pass so that you are not putting a lot of pressure on the tool, which gets translated to your machine, to get a quality cut.

An interesting side observation: A couple of machines that I know of that do not transfer tool pressure to the machine are lasers and water jets. However, they do have "tool pressure" in the sense that the beam or jet will "bend" as the feed rate is increased, if increased too much, they will "break" and stop cutting.

[PaulRowntree]

You mentioned you were measuring the deflection at the "tool", which I will take to mean you mounted something like a dowel pin in the collet of the router? If you did, hats off to you!

Most people just think in terms of how much the gantry beam is going to deflect. But there is a lot more to it than that.

First, you have the bearings in the router itself. The Z axis has bearings and also can have a bit of deflection. The Z is riding on the beam with bearings, more clearance. Of course the beam has deflection. Then the gantry is riding on bearings. It may or may not be on risers which can deflect.

You also have some slop in the drive mechanisms, whether it be R&P or screw driven. If you ever notice, they are termed "anti" backlash as a rule, not "Zero" backlash. (Zero backlash is available but is usually only available in expensive industrial grade equipment).

Then we have the tool itself. Tools deflect. Even on industrial milling machines they deflect when cutting. It is one of the reasons why machinists and toolmakers take roughing passes and finish cuts.

You have to have all of these clearances or else the machine would not move at all. If you "tighten" up the machine, then you will encounter greater wear and it will require more power to move the machine. It is a balancing act.

Hi Treischl;
Yes, I have a pin in the collet, and I am trying to do exactly as you suggest : walking back through the machine to find what is moving (and by how much) and to reduce the slop starting with the major flex points. Trouble is that I don't have a good feel for what normal cutter forces are and how much typical homebuilt machines deflect under these forces. ger21 estimated cutting force values of 30-50 lbs some years ago in another thread; if this is typical, my machine is weak beyond my wildest fears.

[jsheerin]

What you are really measuring (and interested in) is the stiffness of the machine. When I measured my machine the same way a few months ago, it was around 2000 lb / in which correlated pretty well to a finite element model I made (after the fact - I'm getting ready to work on redesigning it). Just divide your deflection by the applied force, so your machine is 1000 lb/in, or about half as stiff as mine.

In terms of what is a good stiffness value, from my research, for cutting steel you want around 50k to >200k lb/in, for aluminum you'd want 20k to >50k, and for wood around 2.8k to 12k. So I'd say your machine is a bit too flexible for really good performance in wood. I'd assume you'd agree if you're trying to stiffen it up, which is interesting to me as it gives me another data point that tells me the above numbers are probably pretty reasonable.

[mlabruyere]

Thanks for the info. To be certain that we are discussing the same thing, are these forces applied to the tool, and are you measuring displacment of the tool relative to the table? My gantry is not as strong or large as yours, but it also shows no deflection if I apply the force to the beam itself.
Cheers!

Yes at the tool, which is why I stated that the Z-Axis was at the center of my beam to ensure the weakest part of the span was measured.

[PaulRowntree]

What you are really measuring (and interested in) is the stiffness of the machine. When I measured my machine the same way a few months ago, it was around 2000 lb / in which correlated pretty well to a finite element model I made (after the fact - I'm getting ready to work on redesigning it). Just divide your deflection by the applied force, so your machine is 1000 lb/in, or about half as stiff as mine.

In terms of what is a good stiffness value, from my research, for cutting steel you want around 50k to >200k lb/in, for aluminum you'd want 20k to >50k, and for wood around 2.8k to 12k. So I'd say your machine is a bit too flexible for really good performance in wood. I'd assume you'd agree if you're trying to stiffen it up, which is interesting to me as it gives me another data point that tells me the above numbers are probably pretty reasonable.

That is exactly the sort of info and comparison that I needed. Thank you! And yes, it was less than acceptable so I am trying to map out the source of the flex. It looks like I am after a factor of 3 improvement, minimum.
Cheers!

[ger21]

ger21 estimated cutting force values of 30-50 lbs some years ago in another thread; if this is typical, my machine is weak beyond my wildest fears.

Think of these numbers as the absolute max you'd probably encounter. With a 1/4" tool, cutting wood, it would probably more likely be in the 5-20lb range. Lots of variables come into play when determining force.

[jsheerin]

If you get a factor of 3, that will be really good. I think my machine is reasonable at 2k for hobby use. I cut aluminum with it, but get a less than perfect finish on side milling and can't take really deep or fast cuts. Attached are two pics of my machine for your reference. The pic with the milling vise attached is how it was setup when I measured stiffness. When I ran FEA on it, I was getting a lot of deflection in the bed of the machine, although I don't have the results sitting right in front of me. I can try to find them this weekend if you're interested.

[jsheerin]

Here are some fea results for my router (or an approximation of my router). One is for a load in the Z direction, and one in X. The block on the plate is fixed, but the base is not supported or constrained. I converted all the beams into hollow steel beams with the same moment of inertia that they have as aluminum beams in order to reduce the memory required for the model. The bearings have a contact stiffness specified between the rollers and the rails, but they are free to slide along the rails. However the screws are rigidly fixed. These are stress plots - anywhere the colors get bright, there is a lot of stress, and reinforcing those areas would reduce deflection and increase stiffness. Anywhere the color is grey, the stress is above the upper limit I specified, so those are the highest stress areas. Some of the areas under the most stress are the cross beams of the bed, the plate the vise is on, the Z axis screw, and some parts of the gantry beam, especially where the Y axis bearings are and where the beam attaches to the uprights. So some of the things I plan on changing on my rebuild of this machine are a beefier Z axis, a larger steel gantry beam, and a bed that is stiffer in all directions and is better connected to the X rails. I'll also be shortening the distance between the x rails and the gantry cross bar, changing to THK bearings, a water cooled spindle and separate drilling head, and swapping the 1/2-10 acme rods for rack and pinion drive and larger diameter ball screws, all of which should further help improve stiffness.