开云体育

To start this thread you should really start reading the posts here as this is where it started and provides the context:

/g/machine-tool-rebuilding/message/13



A horizontal force on the gib will produce a force normal to the surface and a downward force. I don't see an issue with the gib being in contact with the base of the male dovetail. That's a precision surface.

The problem is that if the gib can't preload itself against the gib screws using the downward force, you don't know where the screws will seat in the gib bores. Yes, you can put points on the gib screws but this is self-defeating as it creates high stress at the screw seats so they deform quickly with use (confirmed, I tried it). Well then you say, use a ball and socket shape for the screw and bore, or make the screw tip and bore exactly the same diameter, or use a conical tip going into an undersized bore so that finally, we will have a gib-screw interface which seats correctly every time and never deforms. But no, you still have a problem because the screws themselves have play in the threads and move around easily under stress. So maybe you fix that somehow, but also did you consider that the screws tips may not be perfectly concentric, and so adjusting them always introduces errors? So then you find a way to fix that. Then what you're left with is a very finicky mechanical system where numerous small things all have to be working perfectly and is basically impossible to maintain in optimal condition and is always degraded after a repair.

The form taken by this error is also particularly bad. What happens is that when horizontal force pushes the gib against the dovetail, defects in the system flip a coin to determine whether the slide is deflected down against the slide or up away from it (or both alternately during a cut like a toggle switch). The latter is bad because the compound suddenly becomes very compliant. This means that if you clamp the gib screws in an attempt to rigidize the compound, you may actually end up pushing the slides slightly apart, and then you lose accuracy as well as rigidity.

I considered this stuff when I wrote my webpage. In my solution the preload between the gib and screws serves not only to rigidize the gib but to stabilize the screws and screw seats. Preload here provides us with room for error, room for wear in, and a system that doesn't need to be too reliant on everything going well.

BTW I found confirmation that there should be clearance between the gib and the upper surface of the female dovetail in a short article about making gibs in one of Guy Lautard's "Machinist's Bedside Reader" volumes. On reflection, that makes sense. You would not want the upper face of the gib to bind against the clearance face of the female dovetail. That requires that the lower face of the gib be forced into contact with the horizontal surface of the male dovetail

I'm interested in this article. A solution would be to shim the gib enough that the upper left corner is below the male dovetail corner. Another solution would be to make a groove in the left face of the gib to provide clearance around the dovetail corner using a Dremel mounted something like a router.

Anthony, as you have done a good bit of work in this area before, would you mind posting a bit about measuring the deflections of a dovetail way in various positions? The real focus here should be, "How do you determine what the problem is? How do you measure it? How do you fix it." With links to projects-in-metal for how to make things you don't have.

You determined that the gib was the problem, I simply accepted your premise as I had experienced it. I posted some pictures of my measurement setup. I wrote a webpage explaining how to fix it. I didn't have to build anything to do the measurements, I just used a random pipe, spring scale and tenths indicator. Depending on the state of your gibs, fixing it might be as simple as adding a shim.

I was mainly concerned about sideways forces on the cutter, either from facing or trepanning or from the cutter being extended past the edge of the cross slide so it acts as a lever. I quantified this by using a tenths indicator to measure the movement between the slides close to the gib. I put a pipe over the toolpost mounting bolt and used a spring scale to apply a known force and extrapolated the force on the gib according to the ratio of lengths from the gib and scale to the fulcrum which is the left side of the cross slide. You can see my setup in this picture and I also included the force deflection chart.

To start this thread you should really start reading the posts here as this is where it started and provides the context:

/g/machine-tool-rebuilding/message/13



A horizontal force on the gib will produce a force normal to the surface and a downward force. I don't see an issue with the gib being in contact with the base of the male dovetail. That's a precision surface.

The problem is that if the gib can't preload itself against the gib screws using the downward force, you don't know where the screws will seat in the gib bores. Yes, you can put points on the gib screws but this is self-defeating as it creates high stress at the screw seats so they deform quickly with use (confirmed, I tried it). Well then you say, use a ball and socket shape for the screw and bore, or make the screw tip and bore exactly the same diameter, or use a conical tip going into an undersized bore so that finally, we will have a gib-screw interface which seats correctly every time and never deforms. But no, you still have a problem because the screws themselves have play in the threads and move around easily under stress. So maybe you fix that somehow, but also did you consider that the screws tips may not be perfectly concentric, and so adjusting them always introduces errors? So then you find a way to fix that. Then what you're left with is a very finicky mechanical system where numerous small things all have to be working perfectly and is basically impossible to maintain in optimal condition and is always degraded after a repair.

The form taken by this error is also particularly bad. What happens is that when horizontal force pushes the gib against the dovetail, defects in the system flip a coin to determine whether the slide is deflected down against the slide or up away from it (or both alternately during a cut like a toggle switch). The latter is bad because the compound suddenly becomes very compliant. This means that if you clamp the gib screws in an attempt to rigidize the compound, you may actually end up pushing the slides slightly apart, and then you lose accuracy as well as rigidity.

I considered this stuff when I wrote my webpage. In my solution the preload between the gib and screws serves not only to rigidize the gib but to stabilize the screws and screw seats. Preload here provides us with room for error, room for wear in, and a system that doesn't need to be too reliant on everything going well.

BTW I found confirmation that there should be clearance between the gib and the upper surface of the female dovetail in a short article about making gibs in one of Guy Lautard's "Machinist's Bedside Reader" volumes. On reflection, that makes sense. You would not want the upper face of the gib to bind against the clearance face of the female dovetail. That requires that the lower face of the gib be forced into contact with the horizontal surface of the male dovetail

I'm interested in this article. A solution would be to shim the gib enough that the upper left corner is below the male dovetail corner. Another solution would be to make a groove in the left face of the gib to provide clearance around the dovetail corner using a Dremel mounted something like a router.

Anthony, as you have done a good bit of work in this area before, would you mind posting a bit about measuring the deflections of a dovetail way in various positions? The real focus here should be, "How do you determine what the problem is? How do you measure it? How do you fix it." With links to projects-in-metal for how to make things you don't have.

You determined that the gib was the problem, I simply accepted your premise as I had experienced it. I posted some pictures of my measurement setup. I wrote a webpage explaining how to fix it. I didn't have to build anything to do the measurements, I just used a random pipe, spring scale and tenths indicator. Depending on the state of your gibs, fixing it might be as simple as adding a shim.

I was mainly concerned about sideways forces on the cutter, either from facing or trepanning or from the cutter being extended past the edge of the cross slide so it acts as a lever. I quantified this by using a tenths indicator to measure the movement between the slides close to the gib. I put a pipe over the toolpost mounting bolt and used a spring scale to apply a known force and extrapolated the force on the gib according to the ratio of lengths from the gib and scale to the fulcrum which is the left side of the cross slide. You can see my setup in this picture and I also included the force deflection chart.

To start this thread you should really start reading the posts here as this is where it started and provides the context:

/g/machine-tool-rebuilding/message/13



A horizontal force on the gib will produce a force normal to the surface and a downward force. I don't see an issue with the gib being in contact with the base of the male dovetail. That's a precision surface.

The problem is that if the gib can't preload itself against the gib screws using the downward force, you don't know where the screws will seat in the gib bores. Yes, you can put points on the gib screws but this is self-defeating as it creates high stress at the screw seats so they deform quickly with use (confirmed, I tried it). Well then you say, use a ball and socket shape for the screw and bore, or make the screw tip and bore exactly the same diameter, or use a conical tip going into an undersized bore so that finally, we will have a gib-screw interface which seats correctly every time and never deforms. But no, you still have a problem because the screws themselves have play in the threads and move around easily under stress. So maybe you fix that somehow, but also did you consider that the screws tips may not be perfectly concentric, and so adjusting them always introduces errors? So then you find a way to fix that. Then what you're left with is a very finicky mechanical system where numerous small things all have to be working perfectly and is basically impossible to maintain in optimal condition and is always degraded after a repair.

The form taken by this error is also particularly bad. What happens is that when horizontal force pushes the gib against the dovetail, defects in the system flip a coin to determine whether the slide is deflected down against the slide or up away from it (or both alternately during a cut like a toggle switch). The latter is bad because the compound suddenly becomes very compliant. This means that if you clamp the gib screws in an attempt to rigidize the compound, you may actually end up pushing the slides slightly apart, and then you lose accuracy as well as rigidity.

I considered this stuff when I wrote my webpage. In my solution the preload between the gib and screws serves not only to rigidize the gib but to stabilize the screws and screw seats. Preload here provides us with room for error, room for wear in, and a system that doesn't need to be too reliant on everything going well.

BTW I found confirmation that there should be clearance between the gib and the upper surface of the female dovetail in a short article about making gibs in one of Guy Lautard's "Machinist's Bedside Reader" volumes. On reflection, that makes sense. You would not want the upper face of the gib to bind against the clearance face of the female dovetail. That requires that the lower face of the gib be forced into contact with the horizontal surface of the male dovetail

I'm interested in this article. A solution would be to shim the gib enough that the upper left corner is below the male dovetail corner. Another solution would be to make a groove in the left face of the gib to provide clearance around the dovetail corner using a Dremel mounted something like a router.

Anthony, as you have done a good bit of work in this area before, would you mind posting a bit about measuring the deflections of a dovetail way in various positions? The real focus here should be, "How do you determine what the problem is? How do you measure it? How do you fix it." With links to projects-in-metal for how to make things you don't have.

You determined that the gib was the problem, I simply accepted your premise as I had experienced it. I posted some pictures of my measurement setup. I wrote a webpage explaining how to fix it. I didn't have to build anything to do the measurements, I just used a random pipe, spring scale and tenths indicator. Depending on the state of your gibs, fixing it might be as simple as adding a shim.

I was mainly concerned about sideways forces on the cutter, either from facing or trepanning or from the cutter being extended past the edge of the cross slide so it acts as a lever. I quantified this by using a tenths indicator to measure the movement between the slides close to the gib. I put a pipe over the toolpost mounting bolt and used a spring scale to apply a known force and extrapolated the force on the gib according to the ratio of lengths from the gib and scale to the fulcrum which is the left side of the cross slide. You can see my setup in this picture and I also included the force deflection chart.

?

The copper tube I used is 8.5" long. The scale is a 0-2Kg spring scale. The indicator is a tenths indicator. There is some question of which part of the tube the hook actually engages, so I may have just used 8".

?

The force at the bottom of the chart is not the force as read on the scale but the upward force on the gib which I extrapolated from the lever relationship of the tube to the gib using the left cross slide edge as a fulcrum. So the force on the gib is higher than what is being read on the scale since the tube is a lever.

?

If the distance from left edge (fulcrum) to gib is 2" and the distance from left edge to scale hook is 8". Then according to the lever relationship your gib force will be 8"/2"=4x the scale force.

?

I did this not to confuse anyone, but so that if one wanted to run the numbers with known cutting forces, they could apply the same lever relationship back from the gib to the tool. So if the fulcrum to tool distance is the same as the fulcrum to gib distance, which is?in the ballpark of the average use case for the mini lathe, then the force given on the graph will be about equal to the force on the cutting tool (also, the gib deflection will be about the same as it's contribution to tool deflection).

?

On Monday, December 2, 2019, 04:21:09 PM CST, keantoken via Groups.Io <keantoken@...> wrote:

?

?

?

I've been hoping someone would eventually have their own figures to compare, even if it shows my results are inferior. The main thing I wanted to show was the high rigidity preload area, which I think I did. I think I posted that information on the 7x12 minilathe group, but will have to go find it later. Right now I have to stuff a moving van.

On Monday, December 2, 2019, 03:18:38 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

What’s the length of the pipe you have on the toolpost stud and what kinds of loads were you applying?

I’m curious to compare figures from my home made, scraped in cross slide, which has the gib pinned and closely fit to the also home made half dog point screws but not preloaded, but am unsure from the description how you reached the figures in your chart.

On Wed, Jul 17, 2019 at 07:14 PM, keantoken wrote:

To start this thread you should really start reading the posts here as this is where it started and provides the context:

/g/machine-tool-rebuilding/message/13



A horizontal force on the gib will produce a force normal to the surface and a downward force. I don't see an issue with the gib being in contact with the base of the male dovetail. That's a precision surface.

The problem is that if the gib can't preload itself against the gib screws using the downward force, you don't know where the screws will seat in the gib bores. Yes, you can put points on the gib screws but this is self-defeating as it creates high stress at the screw seats so they deform quickly with use (confirmed, I tried it). Well then you say, use a ball and socket shape for the screw and bore, or make the screw tip and bore exactly the same diameter, or use a conical tip going into an undersized bore so that finally, we will have a gib-screw interface which seats correctly every time and never deforms. But no, you still have a problem because the screws themselves have play in the threads and move around easily under stress. So maybe you fix that somehow, but also did you consider that the screws tips may not be perfectly concentric, and so adjusting them always introduces errors? So then you find a way to fix that. Then what you're left with is a very finicky mechanical system where numerous small things all have to be working perfectly and is basically impossible to maintain in optimal condition and is always degraded after a repair.

The form taken by this error is also particularly bad. What happens is that when horizontal force pushes the gib against the dovetail, defects in the system flip a coin to determine whether the slide is deflected down against the slide or up away from it (or both alternately during a cut like a toggle switch). The latter is bad because the compound suddenly becomes very compliant. This means that if you clamp the gib screws in an attempt to rigidize the compound, you may actually end up pushing the slides slightly apart, and then you lose accuracy as well as rigidity.

I considered this stuff when I wrote my webpage. In my solution the preload between the gib and screws serves not only to rigidize the gib but to stabilize the screws and screw seats. Preload here provides us with room for error, room for wear in, and a system that doesn't need to be too reliant on everything going well.

BTW I found confirmation that there should be clearance between the gib and the upper surface of the female dovetail in a short article about making gibs in one of Guy Lautard's "Machinist's Bedside Reader" volumes. On reflection, that makes sense. You would not want the upper face of the gib to bind against the clearance face of the female dovetail. That requires that the lower face of the gib be forced into contact with the horizontal surface of the male dovetail

I'm interested in this article. A solution would be to shim the gib enough that the upper left corner is below the male dovetail corner. Another solution would be to make a groove in the left face of the gib to provide clearance around the dovetail corner using a Dremel mounted something like a router.

Anthony, as you have done a good bit of work in this area before, would you mind posting a bit about measuring the deflections of a dovetail way in various positions? The real focus here should be, "How do you determine what the problem is? How do you measure it? How do you fix it." With links to projects-in-metal for how to make things you don't have.

You determined that the gib was the problem, I simply accepted your premise as I had experienced it. I posted some pictures of my measurement setup. I wrote a webpage explaining how to fix it. I didn't have to build anything to do the measurements, I just used a random pipe, spring scale and tenths indicator. Depending on the state of your gibs, fixing it might be as simple as adding a shim.

I was mainly concerned about sideways forces on the cutter, either from facing or trepanning or from the cutter being extended past the edge of the cross slide so it acts as a lever. I quantified this by using a tenths indicator to measure the movement between the slides close to the gib. I put a pipe over the toolpost mounting bolt and used a spring scale to apply a known force and extrapolated the force on the gib according to the ratio of lengths from the gib and scale to the fulcrum which is the left side of the cross slide. You can see my setup in this picture and I also included the force deflection chart.

?

?

The copper tube I used is 8.5" long. The scale is a 0-2Kg spring scale. The indicator is a tenths indicator. There is some question of which part of the tube the hook actually engages, so I may have just used 8".

?

The force at the bottom of the chart is not the force as read on the scale but the upward force on the gib which I extrapolated from the lever relationship of the tube to the gib using the left cross slide edge as a fulcrum. So the force on the gib is higher than what is being read on the scale since the tube is a lever.

?

If the distance from left edge (fulcrum) to gib is 2" and the distance from left edge to scale hook is 8". Then according to the lever relationship your gib force will be 8"/2"=4x the scale force.

?

I did this not to confuse anyone, but so that if one wanted to run the numbers with known cutting forces, they could apply the same lever relationship back from the gib to the tool. So if the fulcrum to tool distance is the same as the fulcrum to gib distance, which is?in the ballpark of the average use case for the mini lathe, then the force given on the graph will be about equal to the force on the cutting tool (also, the gib deflection will be about the same as it's contribution to tool deflection).

?

On Monday, December 2, 2019, 04:21:09 PM CST, keantoken via Groups.Io <keantoken@...> wrote:

?

?

?

I've been hoping someone would eventually have their own figures to compare, even if it shows my results are inferior. The main thing I wanted to show was the high rigidity preload area, which I think I did. I think I posted that information on the 7x12 minilathe group, but will have to go find it later. Right now I have to stuff a moving van.

On Monday, December 2, 2019, 03:18:38 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

What’s the length of the pipe you have on the toolpost stud and what kinds of loads were you applying?

I’m curious to compare figures from my home made, scraped in cross slide, which has the gib pinned and closely fit to the also home made half dog point screws but not preloaded, but am unsure from the description how you reached the figures in your chart.

On Wed, Jul 17, 2019 at 07:14 PM, keantoken wrote:

To start this thread you should really start reading the posts here as this is where it started and provides the context:

/g/machine-tool-rebuilding/message/13



A horizontal force on the gib will produce a force normal to the surface and a downward force. I don't see an issue with the gib being in contact with the base of the male dovetail. That's a precision surface.

The problem is that if the gib can't preload itself against the gib screws using the downward force, you don't know where the screws will seat in the gib bores. Yes, you can put points on the gib screws but this is self-defeating as it creates high stress at the screw seats so they deform quickly with use (confirmed, I tried it). Well then you say, use a ball and socket shape for the screw and bore, or make the screw tip and bore exactly the same diameter, or use a conical tip going into an undersized bore so that finally, we will have a gib-screw interface which seats correctly every time and never deforms. But no, you still have a problem because the screws themselves have play in the threads and move around easily under stress. So maybe you fix that somehow, but also did you consider that the screws tips may not be perfectly concentric, and so adjusting them always introduces errors? So then you find a way to fix that. Then what you're left with is a very finicky mechanical system where numerous small things all have to be working perfectly and is basically impossible to maintain in optimal condition and is always degraded after a repair.

The form taken by this error is also particularly bad. What happens is that when horizontal force pushes the gib against the dovetail, defects in the system flip a coin to determine whether the slide is deflected down against the slide or up away from it (or both alternately during a cut like a toggle switch). The latter is bad because the compound suddenly becomes very compliant. This means that if you clamp the gib screws in an attempt to rigidize the compound, you may actually end up pushing the slides slightly apart, and then you lose accuracy as well as rigidity.

I considered this stuff when I wrote my webpage. In my solution the preload between the gib and screws serves not only to rigidize the gib but to stabilize the screws and screw seats. Preload here provides us with room for error, room for wear in, and a system that doesn't need to be too reliant on everything going well.

BTW I found confirmation that there should be clearance between the gib and the upper surface of the female dovetail in a short article about making gibs in one of Guy Lautard's "Machinist's Bedside Reader" volumes. On reflection, that makes sense. You would not want the upper face of the gib to bind against the clearance face of the female dovetail. That requires that the lower face of the gib be forced into contact with the horizontal surface of the male dovetail

I'm interested in this article. A solution would be to shim the gib enough that the upper left corner is below the male dovetail corner. Another solution would be to make a groove in the left face of the gib to provide clearance around the dovetail corner using a Dremel mounted something like a router.

Anthony, as you have done a good bit of work in this area before, would you mind posting a bit about measuring the deflections of a dovetail way in various positions? The real focus here should be, "How do you determine what the problem is? How do you measure it? How do you fix it." With links to projects-in-metal for how to make things you don't have.

You determined that the gib was the problem, I simply accepted your premise as I had experienced it. I posted some pictures of my measurement setup. I wrote a webpage explaining how to fix it. I didn't have to build anything to do the measurements, I just used a random pipe, spring scale and tenths indicator. Depending on the state of your gibs, fixing it might be as simple as adding a shim.

I was mainly concerned about sideways forces on the cutter, either from facing or trepanning or from the cutter being extended past the edge of the cross slide so it acts as a lever. I quantified this by using a tenths indicator to measure the movement between the slides close to the gib. I put a pipe over the toolpost mounting bolt and used a spring scale to apply a known force and extrapolated the force on the gib according to the ratio of lengths from the gib and scale to the fulcrum which is the left side of the cross slide. You can see my setup in this picture and I also included the force deflection chart.

?

?

Yes, that is how I did it, using more accurate final measurements I took since the compound rest contributes to the lever length.? I also made sure not to do any measurement twice in order to get consistent results, as well as never exceeding the measurement force I need for a given data point. There is a lot of friction so any awkward or double measurements or over measurements will introduce hysteresis stick/slip error (which will be some fraction of the difference between rising force and falling force measurements). I wanted to have some sort of winch or clamp setup to pull the scale so I didn't have those confounders but I wasn't sure how to cobble that together. I was able to do okay by hand but I had to be very careful. Having both rising force and falling force curves gives a good visual check against hysteresis errors.

?

Also notice that I took measurements at each tenths point rather than at fixed force values. This is because reading in between tenths on the indicator is too hard and inaccurate while pulling on the scale, and also the tenths indicator is what you need to pay close attention to to monitor whether the measurement is excessive. Notice that at 12-13 tenths on the increasing force curve the result jumps up. I would guess this is due to an over measurement, resulting in excessive settling. But it is also possible that is just a natural slip point due to sliding surface roughness.

?

So the process goes something like this.

?

1: Work the slide so it is in a resting state and is fully settled with gravity. Zero the indicator. Pull on the slide to get an idea of the play and make this your first datapoint. This can be somewhat subjective if there is no firm endpoint to the play.

2: Pull the scale while watching the indicator until it reaches the next tenths mark, but no more. If it goes over, re-seat the slide with a hammer or something and repeat. Ideally never let this happen. Note down the tenths mark and the force at which it was reached.

3: Repeat 1 and 2 until your arm is tired, the scale is maxed out or your measurement limit is reached.

4: Start the falling force measurement by starting at your highest measurement or higher, and reducing the force, taking note at each tenths mark until you reach zero force. Any failed measurements need to be reset by maxing out the force again, but still we don't want any failed measurements to introduce noise into the data.

On Wednesday, December 4, 2019, 04:15:20 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

[Edited Message Follows]

Thanks! I just finally did move and set up my two mini lathes again and picked up a fishing scale to test my plodding upgrades with some real figures. I guess it would’ve been more useful if I did some “before” measurements, too - what I’m really interested in is any effect from the braces I put under the backside of the bed under the headstock, but they’re in kind of permanently.

?

Just to make sure, for example with a 6” pipe and the DTI probe 2” from the front, headstock edge of the bottom of the cross slide, should get forces like yours multiplying the scale readings by 3. (*Edit: 5” pipe, 6” from the scale hook to the cs edge)

?

?

For fun, here’s the start of one of the next upgrades I’ll make to my 7x16 - an external back gear to go under the bench with the motor.



On Tue, Dec 3, 2019 at 03:07 AM, keantoken wrote:

?

The copper tube I used is 8.5" long. The scale is a 0-2Kg spring scale. The indicator is a tenths indicator. There is some question of which part of the tube the hook actually engages, so I may have just used 8".

?

The force at the bottom of the chart is not the force as read on the scale but the upward force on the gib which I extrapolated from the lever relationship of the tube to the gib using the left cross slide edge as a fulcrum. So the force on the gib is higher than what is being read on the scale since the tube is a lever.

?

If the distance from left edge (fulcrum) to gib is 2" and the distance from left edge to scale hook is 8". Then according to the lever relationship your gib force will be 8"/2"=4x the scale force.

?

I did this not to confuse anyone, but so that if one wanted to run the numbers with known cutting forces, they could apply the same lever relationship back from the gib to the tool. So if the fulcrum to tool distance is the same as the fulcrum to gib distance, which is?in the ballpark of the average use case for the mini lathe, then the force given on the graph will be about equal to the force on the cutting tool (also, the gib deflection will be about the same as it's contribution to tool deflection).

?

On Monday, December 2, 2019, 04:21:09 PM CST, keantoken via Groups.Io <keantoken@...> wrote:

?

?

?

I've been hoping someone would eventually have their own figures to compare, even if it shows my results are inferior. The main thing I wanted to show was the high rigidity preload area, which I think I did. I think I posted that information on the 7x12 minilathe group, but will have to go find it later. Right now I have to stuff a moving van.

On Monday, December 2, 2019, 03:18:38 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

What’s the length of the pipe you have on the toolpost stud and what kinds of loads were you applying?

I’m curious to compare figures from my home made, scraped in cross slide, which has the gib pinned and closely fit to the also home made half dog point screws but not preloaded, but am unsure from the description how you reached the figures in your chart.

On Wed, Jul 17, 2019 at 07:14 PM, keantoken wrote:

To start this thread you should really start reading the posts here as this is where it started and provides the context:

/g/machine-tool-rebuilding/message/13



A horizontal force on the gib will produce a force normal to the surface and a downward force. I don't see an issue with the gib being in contact with the base of the male dovetail. That's a precision surface.

The problem is that if the gib can't preload itself against the gib screws using the downward force, you don't know where the screws will seat in the gib bores. Yes, you can put points on the gib screws but this is self-defeating as it creates high stress at the screw seats so they deform quickly with use (confirmed, I tried it). Well then you say, use a ball and socket shape for the screw and bore, or make the screw tip and bore exactly the same diameter, or use a conical tip going into an undersized bore so that finally, we will have a gib-screw interface which seats correctly every time and never deforms. But no, you still have a problem because the screws themselves have play in the threads and move around easily under stress. So maybe you fix that somehow, but also did you consider that the screws tips may not be perfectly concentric, and so adjusting them always introduces errors? So then you find a way to fix that. Then what you're left with is a very finicky mechanical system where numerous small things all have to be working perfectly and is basically impossible to maintain in optimal condition and is always degraded after a repair.

The form taken by this error is also particularly bad. What happens is that when horizontal force pushes the gib against the dovetail, defects in the system flip a coin to determine whether the slide is deflected down against the slide or up away from it (or both alternately during a cut like a toggle switch). The latter is bad because the compound suddenly becomes very compliant. This means that if you clamp the gib screws in an attempt to rigidize the compound, you may actually end up pushing the slides slightly apart, and then you lose accuracy as well as rigidity.

I considered this stuff when I wrote my webpage. In my solution the preload between the gib and screws serves not only to rigidize the gib but to stabilize the screws and screw seats. Preload here provides us with room for error, room for wear in, and a system that doesn't need to be too reliant on everything going well.

BTW I found confirmation that there should be clearance between the gib and the upper surface of the female dovetail in a short article about making gibs in one of Guy Lautard's "Machinist's Bedside Reader" volumes. On reflection, that makes sense. You would not want the upper face of the gib to bind against the clearance face of the female dovetail. That requires that the lower face of the gib be forced into contact with the horizontal surface of the male dovetail

I'm interested in this article. A solution would be to shim the gib enough that the upper left corner is below the male dovetail corner. Another solution would be to make a groove in the left face of the gib to provide clearance around the dovetail corner using a Dremel mounted something like a router.

Anthony, as you have done a good bit of work in this area before, would you mind posting a bit about measuring the deflections of a dovetail way in various positions? The real focus here should be, "How do you determine what the problem is? How do you measure it? How do you fix it." With links to projects-in-metal for how to make things you don't have.

You determined that the gib was the problem, I simply accepted your premise as I had experienced it. I posted some pictures of my measurement setup. I wrote a webpage explaining how to fix it. I didn't have to build anything to do the measurements, I just used a random pipe, spring scale and tenths indicator. Depending on the state of your gibs, fixing it might be as simple as adding a shim.

I was mainly concerned about sideways forces on the cutter, either from facing or trepanning or from the cutter being extended past the edge of the cross slide so it acts as a lever. I quantified this by using a tenths indicator to measure the movement between the slides close to the gib. I put a pipe over the toolpost mounting bolt and used a spring scale to apply a known force and extrapolated the force on the gib according to the ratio of lengths from the gib and scale to the fulcrum which is the left side of the cross slide. You can see my setup in this picture and I also included the force deflection chart.

?

?

Yes, that is how I did it, using more accurate final measurements I took since the compound rest contributes to the lever length.? I also made sure not to do any measurement twice in order to get consistent results, as well as never exceeding the measurement force I need for a given data point. There is a lot of friction so any awkward or double measurements or over measurements will introduce hysteresis stick/slip error (which will be some fraction of the difference between rising force and falling force measurements). I wanted to have some sort of winch or clamp setup to pull the scale so I didn't have those confounders but I wasn't sure how to cobble that together. I was able to do okay by hand but I had to be very careful. Having both rising force and falling force curves gives a good visual check against hysteresis errors.

?

Also notice that I took measurements at each tenths point rather than at fixed force values. This is because reading in between tenths on the indicator is too hard and inaccurate while pulling on the scale, and also the tenths indicator is what you need to pay close attention to to monitor whether the measurement is excessive. Notice that at 12-13 tenths on the increasing force curve the result jumps up. I would guess this is due to an over measurement, resulting in excessive settling. But it is also possible that is just a natural slip point due to sliding surface roughness.

?

So the process goes something like this.

?

1: Work the slide so it is in a resting state and is fully settled with gravity. Zero the indicator. Pull on the slide to get an idea of the play and make this your first datapoint. This can be somewhat subjective if there is no firm endpoint to the play.

2: Pull the scale while watching the indicator until it reaches the next tenths mark, but no more. If it goes over, re-seat the slide with a hammer or something and repeat. Ideally never let this happen. Note down the tenths mark and the force at which it was reached.

3: Repeat 1 and 2 until your arm is tired, the scale is maxed out or your measurement limit is reached.

4: Start the falling force measurement by starting at your highest measurement or higher, and reducing the force, taking note at each tenths mark until you reach zero force. Any failed measurements need to be reset by maxing out the force again, but still we don't want any failed measurements to introduce noise into the data.

On Wednesday, December 4, 2019, 04:15:20 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

[Edited Message Follows]

Thanks! I just finally did move and set up my two mini lathes again and picked up a fishing scale to test my plodding upgrades with some real figures. I guess it would’ve been more useful if I did some “before” measurements, too - what I’m really interested in is any effect from the braces I put under the backside of the bed under the headstock, but they’re in kind of permanently.

?

Just to make sure, for example with a 6” pipe and the DTI probe 2” from the front, headstock edge of the bottom of the cross slide, should get forces like yours multiplying the scale readings by 3. (*Edit: 5” pipe, 6” from the scale hook to the cs edge)

?

?

For fun, here’s the start of one of the next upgrades I’ll make to my 7x16 - an external back gear to go under the bench with the motor.



On Tue, Dec 3, 2019 at 03:07 AM, keantoken wrote:

?

The copper tube I used is 8.5" long. The scale is a 0-2Kg spring scale. The indicator is a tenths indicator. There is some question of which part of the tube the hook actually engages, so I may have just used 8".

?

The force at the bottom of the chart is not the force as read on the scale but the upward force on the gib which I extrapolated from the lever relationship of the tube to the gib using the left cross slide edge as a fulcrum. So the force on the gib is higher than what is being read on the scale since the tube is a lever.

?

If the distance from left edge (fulcrum) to gib is 2" and the distance from left edge to scale hook is 8". Then according to the lever relationship your gib force will be 8"/2"=4x the scale force.

?

I did this not to confuse anyone, but so that if one wanted to run the numbers with known cutting forces, they could apply the same lever relationship back from the gib to the tool. So if the fulcrum to tool distance is the same as the fulcrum to gib distance, which is?in the ballpark of the average use case for the mini lathe, then the force given on the graph will be about equal to the force on the cutting tool (also, the gib deflection will be about the same as it's contribution to tool deflection).

?

On Monday, December 2, 2019, 04:21:09 PM CST, keantoken via Groups.Io <keantoken@...> wrote:

?

?

?

I've been hoping someone would eventually have their own figures to compare, even if it shows my results are inferior. The main thing I wanted to show was the high rigidity preload area, which I think I did. I think I posted that information on the 7x12 minilathe group, but will have to go find it later. Right now I have to stuff a moving van.

On Monday, December 2, 2019, 03:18:38 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

What’s the length of the pipe you have on the toolpost stud and what kinds of loads were you applying?

I’m curious to compare figures from my home made, scraped in cross slide, which has the gib pinned and closely fit to the also home made half dog point screws but not preloaded, but am unsure from the description how you reached the figures in your chart.

On Wed, Jul 17, 2019 at 07:14 PM, keantoken wrote:

To start this thread you should really start reading the posts here as this is where it started and provides the context:

/g/machine-tool-rebuilding/message/13



A horizontal force on the gib will produce a force normal to the surface and a downward force. I don't see an issue with the gib being in contact with the base of the male dovetail. That's a precision surface.

The problem is that if the gib can't preload itself against the gib screws using the downward force, you don't know where the screws will seat in the gib bores. Yes, you can put points on the gib screws but this is self-defeating as it creates high stress at the screw seats so they deform quickly with use (confirmed, I tried it). Well then you say, use a ball and socket shape for the screw and bore, or make the screw tip and bore exactly the same diameter, or use a conical tip going into an undersized bore so that finally, we will have a gib-screw interface which seats correctly every time and never deforms. But no, you still have a problem because the screws themselves have play in the threads and move around easily under stress. So maybe you fix that somehow, but also did you consider that the screws tips may not be perfectly concentric, and so adjusting them always introduces errors? So then you find a way to fix that. Then what you're left with is a very finicky mechanical system where numerous small things all have to be working perfectly and is basically impossible to maintain in optimal condition and is always degraded after a repair.

The form taken by this error is also particularly bad. What happens is that when horizontal force pushes the gib against the dovetail, defects in the system flip a coin to determine whether the slide is deflected down against the slide or up away from it (or both alternately during a cut like a toggle switch). The latter is bad because the compound suddenly becomes very compliant. This means that if you clamp the gib screws in an attempt to rigidize the compound, you may actually end up pushing the slides slightly apart, and then you lose accuracy as well as rigidity.

I considered this stuff when I wrote my webpage. In my solution the preload between the gib and screws serves not only to rigidize the gib but to stabilize the screws and screw seats. Preload here provides us with room for error, room for wear in, and a system that doesn't need to be too reliant on everything going well.

BTW I found confirmation that there should be clearance between the gib and the upper surface of the female dovetail in a short article about making gibs in one of Guy Lautard's "Machinist's Bedside Reader" volumes. On reflection, that makes sense. You would not want the upper face of the gib to bind against the clearance face of the female dovetail. That requires that the lower face of the gib be forced into contact with the horizontal surface of the male dovetail

I'm interested in this article. A solution would be to shim the gib enough that the upper left corner is below the male dovetail corner. Another solution would be to make a groove in the left face of the gib to provide clearance around the dovetail corner using a Dremel mounted something like a router.

Anthony, as you have done a good bit of work in this area before, would you mind posting a bit about measuring the deflections of a dovetail way in various positions? The real focus here should be, "How do you determine what the problem is? How do you measure it? How do you fix it." With links to projects-in-metal for how to make things you don't have.

You determined that the gib was the problem, I simply accepted your premise as I had experienced it. I posted some pictures of my measurement setup. I wrote a webpage explaining how to fix it. I didn't have to build anything to do the measurements, I just used a random pipe, spring scale and tenths indicator. Depending on the state of your gibs, fixing it might be as simple as adding a shim.

I was mainly concerned about sideways forces on the cutter, either from facing or trepanning or from the cutter being extended past the edge of the cross slide so it acts as a lever. I quantified this by using a tenths indicator to measure the movement between the slides close to the gib. I put a pipe over the toolpost mounting bolt and used a spring scale to apply a known force and extrapolated the force on the gib according to the ratio of lengths from the gib and scale to the fulcrum which is the left side of the cross slide. You can see my setup in this picture and I also included the force deflection chart.

?

?

Yes, that is how I did it, using more accurate final measurements I took since the compound rest contributes to the lever length.? I also made sure not to do any measurement twice in order to get consistent results, as well as never exceeding the measurement force I need for a given data point. There is a lot of friction so any awkward or double measurements or over measurements will introduce hysteresis stick/slip error (which will be some fraction of the difference between rising force and falling force measurements). I wanted to have some sort of winch or clamp setup to pull the scale so I didn't have those confounders but I wasn't sure how to cobble that together. I was able to do okay by hand but I had to be very careful. Having both rising force and falling force curves gives a good visual check against hysteresis errors.

?

Also notice that I took measurements at each tenths point rather than at fixed force values. This is because reading in between tenths on the indicator is too hard and inaccurate while pulling on the scale, and also the tenths indicator is what you need to pay close attention to to monitor whether the measurement is excessive. Notice that at 12-13 tenths on the increasing force curve the result jumps up. I would guess this is due to an over measurement, resulting in excessive settling. But it is also possible that is just a natural slip point due to sliding surface roughness.

?

So the process goes something like this.

?

1: Work the slide so it is in a resting state and is fully settled with gravity. Zero the indicator. Pull on the slide to get an idea of the play and make this your first datapoint. This can be somewhat subjective if there is no firm endpoint to the play.

2: Pull the scale while watching the indicator until it reaches the next tenths mark, but no more. If it goes over, re-seat the slide with a hammer or something and repeat. Ideally never let this happen. Note down the tenths mark and the force at which it was reached.

3: Repeat 1 and 2 until your arm is tired, the scale is maxed out or your measurement limit is reached.

4: Start the falling force measurement by starting at your highest measurement or higher, and reducing the force, taking note at each tenths mark until you reach zero force. Any failed measurements need to be reset by maxing out the force again, but still we don't want any failed measurements to introduce noise into the data.

On Wednesday, December 4, 2019, 04:15:20 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

[Edited Message Follows]

Thanks! I just finally did move and set up my two mini lathes again and picked up a fishing scale to test my plodding upgrades with some real figures. I guess it would’ve been more useful if I did some “before” measurements, too - what I’m really interested in is any effect from the braces I put under the backside of the bed under the headstock, but they’re in kind of permanently.

?

Just to make sure, for example with a 6” pipe and the DTI probe 2” from the front, headstock edge of the bottom of the cross slide, should get forces like yours multiplying the scale readings by 3. (*Edit: 5” pipe, 6” from the scale hook to the cs edge)

?

?

For fun, here’s the start of one of the next upgrades I’ll make to my 7x16 - an external back gear to go under the bench with the motor.



On Tue, Dec 3, 2019 at 03:07 AM, keantoken wrote:

?

The copper tube I used is 8.5" long. The scale is a 0-2Kg spring scale. The indicator is a tenths indicator. There is some question of which part of the tube the hook actually engages, so I may have just used 8".

?

The force at the bottom of the chart is not the force as read on the scale but the upward force on the gib which I extrapolated from the lever relationship of the tube to the gib using the left cross slide edge as a fulcrum. So the force on the gib is higher than what is being read on the scale since the tube is a lever.

?

If the distance from left edge (fulcrum) to gib is 2" and the distance from left edge to scale hook is 8". Then according to the lever relationship your gib force will be 8"/2"=4x the scale force.

?

I did this not to confuse anyone, but so that if one wanted to run the numbers with known cutting forces, they could apply the same lever relationship back from the gib to the tool. So if the fulcrum to tool distance is the same as the fulcrum to gib distance, which is?in the ballpark of the average use case for the mini lathe, then the force given on the graph will be about equal to the force on the cutting tool (also, the gib deflection will be about the same as it's contribution to tool deflection).

?

On Monday, December 2, 2019, 04:21:09 PM CST, keantoken via Groups.Io <keantoken@...> wrote:

?

?

?

I've been hoping someone would eventually have their own figures to compare, even if it shows my results are inferior. The main thing I wanted to show was the high rigidity preload area, which I think I did. I think I posted that information on the 7x12 minilathe group, but will have to go find it later. Right now I have to stuff a moving van.

On Monday, December 2, 2019, 03:18:38 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

What’s the length of the pipe you have on the toolpost stud and what kinds of loads were you applying?

I’m curious to compare figures from my home made, scraped in cross slide, which has the gib pinned and closely fit to the also home made half dog point screws but not preloaded, but am unsure from the description how you reached the figures in your chart.

On Wed, Jul 17, 2019 at 07:14 PM, keantoken wrote:

To start this thread you should really start reading the posts here as this is where it started and provides the context:

/g/machine-tool-rebuilding/message/13



A horizontal force on the gib will produce a force normal to the surface and a downward force. I don't see an issue with the gib being in contact with the base of the male dovetail. That's a precision surface.

The problem is that if the gib can't preload itself against the gib screws using the downward force, you don't know where the screws will seat in the gib bores. Yes, you can put points on the gib screws but this is self-defeating as it creates high stress at the screw seats so they deform quickly with use (confirmed, I tried it). Well then you say, use a ball and socket shape for the screw and bore, or make the screw tip and bore exactly the same diameter, or use a conical tip going into an undersized bore so that finally, we will have a gib-screw interface which seats correctly every time and never deforms. But no, you still have a problem because the screws themselves have play in the threads and move around easily under stress. So maybe you fix that somehow, but also did you consider that the screws tips may not be perfectly concentric, and so adjusting them always introduces errors? So then you find a way to fix that. Then what you're left with is a very finicky mechanical system where numerous small things all have to be working perfectly and is basically impossible to maintain in optimal condition and is always degraded after a repair.

The form taken by this error is also particularly bad. What happens is that when horizontal force pushes the gib against the dovetail, defects in the system flip a coin to determine whether the slide is deflected down against the slide or up away from it (or both alternately during a cut like a toggle switch). The latter is bad because the compound suddenly becomes very compliant. This means that if you clamp the gib screws in an attempt to rigidize the compound, you may actually end up pushing the slides slightly apart, and then you lose accuracy as well as rigidity.

I considered this stuff when I wrote my webpage. In my solution the preload between the gib and screws serves not only to rigidize the gib but to stabilize the screws and screw seats. Preload here provides us with room for error, room for wear in, and a system that doesn't need to be too reliant on everything going well.

BTW I found confirmation that there should be clearance between the gib and the upper surface of the female dovetail in a short article about making gibs in one of Guy Lautard's "Machinist's Bedside Reader" volumes. On reflection, that makes sense. You would not want the upper face of the gib to bind against the clearance face of the female dovetail. That requires that the lower face of the gib be forced into contact with the horizontal surface of the male dovetail

I'm interested in this article. A solution would be to shim the gib enough that the upper left corner is below the male dovetail corner. Another solution would be to make a groove in the left face of the gib to provide clearance around the dovetail corner using a Dremel mounted something like a router.

Anthony, as you have done a good bit of work in this area before, would you mind posting a bit about measuring the deflections of a dovetail way in various positions? The real focus here should be, "How do you determine what the problem is? How do you measure it? How do you fix it." With links to projects-in-metal for how to make things you don't have.

You determined that the gib was the problem, I simply accepted your premise as I had experienced it. I posted some pictures of my measurement setup. I wrote a webpage explaining how to fix it. I didn't have to build anything to do the measurements, I just used a random pipe, spring scale and tenths indicator. Depending on the state of your gibs, fixing it might be as simple as adding a shim.

I was mainly concerned about sideways forces on the cutter, either from facing or trepanning or from the cutter being extended past the edge of the cross slide so it acts as a lever. I quantified this by using a tenths indicator to measure the movement between the slides close to the gib. I put a pipe over the toolpost mounting bolt and used a spring scale to apply a known force and extrapolated the force on the gib according to the ratio of lengths from the gib and scale to the fulcrum which is the left side of the cross slide. You can see my setup in this picture and I also included the force deflection chart.

?

It looks like you're using a mil indicator rather than a tenths indicator. The face on your indicator looks like a mil indicator. If not, then your results are about 10x better than mine. If you want to double check the indicator, try slipping a feeler gauge under the slide next to the indicator probe.

?

If you're using a mil indicator, your results are in the right ballpark for a stock minilathe.

On Thursday, December 5, 2019, 03:31:12 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

Got it.?

?

Maybe I did something wrong? My numbers are so different from yours and I don’t know what kind of readings a good lathe will get. I’ll check the scale when I get a chance.

?

The post puts the center of the hook 6.50”, and the DTI probe sets 3.24” from the front, headstock side lower edge of the cross slide. Using the headstock for leverage I could comfortably and steadily pull up to 11.4kg producing a 0.032mm rise.?

?

I forgot to do the falling force.

?

On Wed, Dec 4, 2019 at 08:21 PM, keantoken wrote:

?

Yes, that is how I did it, using more accurate final measurements I took since the compound rest contributes to the lever length.? I also made sure not to do any measurement twice in order to get consistent results, as well as never exceeding the measurement force I need for a given data point. There is a lot of friction so any awkward or double measurements or over measurements will introduce hysteresis stick/slip error (which will be some fraction of the difference between rising force and falling force measurements). I wanted to have some sort of winch or clamp setup to pull the scale so I didn't have those confounders but I wasn't sure how to cobble that together. I was able to do okay by hand but I had to be very careful. Having both rising force and falling force curves gives a good visual check against hysteresis errors.

?

Also notice that I took measurements at each tenths point rather than at fixed force values. This is because reading in between tenths on the indicator is too hard and inaccurate while pulling on the scale, and also the tenths indicator is what you need to pay close attention to to monitor whether the measurement is excessive. Notice that at 12-13 tenths on the increasing force curve the result jumps up. I would guess this is due to an over measurement, resulting in excessive settling. But it is also possible that is just a natural slip point due to sliding surface roughness.

?

So the process goes something like this.

?

1: Work the slide so it is in a resting state and is fully settled with gravity. Zero the indicator. Pull on the slide to get an idea of the play and make this your first datapoint. This can be somewhat subjective if there is no firm endpoint to the play.

2: Pull the scale while watching the indicator until it reaches the next tenths mark, but no more. If it goes over, re-seat the slide with a hammer or something and repeat. Ideally never let this happen. Note down the tenths mark and the force at which it was reached.

3: Repeat 1 and 2 until your arm is tired, the scale is maxed out or your measurement limit is reached.

4: Start the falling force measurement by starting at your highest measurement or higher, and reducing the force, taking note at each tenths mark until you reach zero force. Any failed measurements need to be reset by maxing out the force again, but still we don't want any failed measurements to introduce noise into the data.

On Wednesday, December 4, 2019, 04:15:20 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

[Edited Message Follows]

Thanks! I just finally did move and set up my two mini lathes again and picked up a fishing scale to test my plodding upgrades with some real figures. I guess it would’ve been more useful if I did some “before” measurements, too - what I’m really interested in is any effect from the braces I put under the backside of the bed under the headstock, but they’re in kind of permanently.

?

Just to make sure, for example with a 6” pipe and the DTI probe 2” from the front, headstock edge of the bottom of the cross slide, should get forces like yours multiplying the scale readings by 3. (*Edit: 5” pipe, 6” from the scale hook to the cs edge)

?

?

For fun, here’s the start of one of the next upgrades I’ll make to my 7x16 - an external back gear to go under the bench with the motor.



On Tue, Dec 3, 2019 at 03:07 AM, keantoken wrote:

?

The copper tube I used is 8.5" long. The scale is a 0-2Kg spring scale. The indicator is a tenths indicator. There is some question of which part of the tube the hook actually engages, so I may have just used 8".

?

The force at the bottom of the chart is not the force as read on the scale but the upward force on the gib which I extrapolated from the lever relationship of the tube to the gib using the left cross slide edge as a fulcrum. So the force on the gib is higher than what is being read on the scale since the tube is a lever.

?

If the distance from left edge (fulcrum) to gib is 2" and the distance from left edge to scale hook is 8". Then according to the lever relationship your gib force will be 8"/2"=4x the scale force.

?

I did this not to confuse anyone, but so that if one wanted to run the numbers with known cutting forces, they could apply the same lever relationship back from the gib to the tool. So if the fulcrum to tool distance is the same as the fulcrum to gib distance, which is?in the ballpark of the average use case for the mini lathe, then the force given on the graph will be about equal to the force on the cutting tool (also, the gib deflection will be about the same as it's contribution to tool deflection).

?

On Monday, December 2, 2019, 04:21:09 PM CST, keantoken via Groups.Io <keantoken@...> wrote:

?

?

?

I've been hoping someone would eventually have their own figures to compare, even if it shows my results are inferior. The main thing I wanted to show was the high rigidity preload area, which I think I did. I think I posted that information on the 7x12 minilathe group, but will have to go find it later. Right now I have to stuff a moving van.

On Monday, December 2, 2019, 03:18:38 PM CST, Clark Panaccione via Groups.Io <threesixesinarow@...> wrote:

?

?

What’s the length of the pipe you have on the toolpost stud and what kinds of loads were you applying?

I’m curious to compare figures from my home made, scraped in cross slide, which has the gib pinned and closely fit to the also home made half dog point screws but not preloaded, but am unsure from the description how you reached the figures in your chart.

On Wed, Jul 17, 2019 at 07:14 PM, keantoken wrote:

To start this thread you should really start reading the posts here as this is where it started and provides the context:

/g/machine-tool-rebuilding/message/13



A horizontal force on the gib will produce a force normal to the surface and a downward force. I don't see an issue with the gib being in contact with the base of the male dovetail. That's a precision surface.

The problem is that if the gib can't preload itself against the gib screws using the downward force, you don't know where the screws will seat in the gib bores. Yes, you can put points on the gib screws but this is self-defeating as it creates high stress at the screw seats so they deform quickly with use (confirmed, I tried it). Well then you say, use a ball and socket shape for the screw and bore, or make the screw tip and bore exactly the same diameter, or use a conical tip going into an undersized bore so that finally, we will have a gib-screw interface which seats correctly every time and never deforms. But no, you still have a problem because the screws themselves have play in the threads and move around easily under stress. So maybe you fix that somehow, but also did you consider that the screws tips may not be perfectly concentric, and so adjusting them always introduces errors? So then you find a way to fix that. Then what you're left with is a very finicky mechanical system where numerous small things all have to be working perfectly and is basically impossible to maintain in optimal condition and is always degraded after a repair.

The form taken by this error is also particularly bad. What happens is that when horizontal force pushes the gib against the dovetail, defects in the system flip a coin to determine whether the slide is deflected down against the slide or up away from it (or both alternately during a cut like a toggle switch). The latter is bad because the compound suddenly becomes very compliant. This means that if you clamp the gib screws in an attempt to rigidize the compound, you may actually end up pushing the slides slightly apart, and then you lose accuracy as well as rigidity.

I considered this stuff when I wrote my webpage. In my solution the preload between the gib and screws serves not only to rigidize the gib but to stabilize the screws and screw seats. Preload here provides us with room for error, room for wear in, and a system that doesn't need to be too reliant on everything going well.

BTW I found confirmation that there should be clearance between the gib and the upper surface of the female dovetail in a short article about making gibs in one of Guy Lautard's "Machinist's Bedside Reader" volumes. On reflection, that makes sense. You would not want the upper face of the gib to bind against the clearance face of the female dovetail. That requires that the lower face of the gib be forced into contact with the horizontal surface of the male dovetail

I'm interested in this article. A solution would be to shim the gib enough that the upper left corner is below the male dovetail corner. Another solution would be to make a groove in the left face of the gib to provide clearance around the dovetail corner using a Dremel mounted something like a router.

Anthony, as you have done a good bit of work in this area before, would you mind posting a bit about measuring the deflections of a dovetail way in various positions? The real focus here should be, "How do you determine what the problem is? How do you measure it? How do you fix it." With links to projects-in-metal for how to make things you don't have.

You determined that the gib was the problem, I simply accepted your premise as I had experienced it. I posted some pictures of my measurement setup. I wrote a webpage explaining how to fix it. I didn't have to build anything to do the measurements, I just used a random pipe, spring scale and tenths indicator. Depending on the state of your gibs, fixing it might be as simple as adding a shim.

I was mainly concerned about sideways forces on the cutter, either from facing or trepanning or from the cutter being extended past the edge of the cross slide so it acts as a lever. I quantified this by using a tenths indicator to measure the movement between the slides close to the gib. I put a pipe over the toolpost mounting bolt and used a spring scale to apply a known force and extrapolated the force on the gib according to the ratio of lengths from the gib and scale to the fulcrum which is the left side of the cross slide. You can see my setup in this picture and I also included the force deflection chart.

?

My gibs use the stock screws and simply add a preload. So the deflection will be greater than a pinned gib after the preload has been overcome. The hope is that we have enough preload to cover most cuts. I've added your chart to mine to compare them more easily. It does seem that the preload may have had an effect, but has no advantage over your gib after 5500g. So the preload force should be increased if possible. If you could do a falling force curve that would help to sanity check these ideas.

My gibs use the stock screws and simply add a preload. So the deflection will be greater than a pinned gib after the preload has been overcome. The hope is that we have enough preload to cover most cuts. I've added your chart to mine to compare them more easily. It does seem that the preload may have had an effect, but has no advantage over your gib after 5500g. So the preload force should be increased if possible. If you could do a falling force curve that would help to sanity check these ideas.

My gibs use the stock screws and simply add a preload. So the deflection will be greater than a pinned gib after the preload has been overcome. The hope is that we have enough preload to cover most cuts. I've added your chart to mine to compare them more easily. It does seem that the preload may have had an effect, but has no advantage over your gib after 5500g. So the preload force should be increased if possible. If you could do a falling force curve that would help to sanity check these ideas.