Rollie's Dad's Method?

[daved20319b]

So this post is mostly thinking out loud, but if someone has some insight, I'd like to hear it.

So in my ongoing effort to get my older Grizzly 12" x 24" dialed in, I decided to put the cart before the horse and try to align the headstock to the ways. I say cart before the horse because I haven't really done a thorough job of leveling this thing, there's enough twist in the bed that a precision level soon has me chasing my tail. But that's fodder for another post.

Anyway, RDM seems to be a fairly popular and respected alternative to the usual method of dialing in lathe alignment, so I decided to give it a try. For a test bar, I chucked up a piece of 1.25" aluminum heavy wall tubing in my 3 jaw, I figured that would be stiffer than a comparable solid bar, and it was a good length. And right away, I was having issues with the instructions. They say to record the high and low numbers and average them. But why/how are they getting high and low numbers in the first place? Why not just zero the indicator on the low point, then record total runout? Anyway, that's what I did, and I was getting about .002" at the chuck, and almost .060" at 16" from the chuck. Which seemed excessive, if I'm understanding the instructions, that would seem to indicate my alignment is out by ~.030" at the 16" mark. But that's about double what I'm seeing for taper when I turn something. For no particular reason, I decided to swap ends on my test bar, and I got totally different numbers, about half my previous measurements. So then I switched to a 7/8" 416 SS round bar, and got a whole new set of numbers nowhere near the previous checks, much lower and closer together end to end. So in 3 tries, I came up with an alignment error of ~ .030", .015", and .003". Not exactly useful info, what am I supposed to do with that, average all three readings?

My other issue with this method is correcting the alignment by shimming the feet. Seems like this is assuming the lathe is twisted, but if it's leveled when you start, as it should be, then why not go to the source and adjust the headstock directly? I realize not all lathes allow adjusting the headstock, but mine does, and once (if) I ever get mine level and straight, I'm sure not going to mess it up by twisting it again.

Maybe I'm totally missing something here, but it seems to me that if this is going to produce useable data, I should have gotten at least similar numbers from all three attempts. I was careful to make each measurement at the same location on the bars, and I made all the measurements multiple times. I've decided to set this aside for now and concentrate on trying to get my lathe properly leveled, but at this point, I won't be trying RDM again, I'll go with the more traditional method.

And BTW, yes, I'm working on the level/twist again, only this time I'm going to see if time and gravity can accomplish what I couldn't with brute force. More later.

Dave

 

[Pevehouse]

Keep us posted on what you are doing. I have a 12 x 36 encore that I need to level and align also. I don’t really have any answers for you for I have never done the RDM.

 

[daved20319b]

Keep us posted on what you are doing. I have a 12 x 36 encore that I need to level and align also. I don’t really have any answers for you for I have never done the RDM.

Will do. Getting this thing leveled has been an ongoing PITA, but that's probably at least partly my fault, I wanted it perfect and I wanted it NOW. My lathe has 4 adjustment feet, but there's several thou twist in the bed from end to end, and nothing I'd tried seemed to do anything. Well, other than raise my blood pressure from chasing my tail around. This time, I put the TS end on a machinist's jack, so just a center pivot, leveled the HS end front to back, then the bed from end to end. From there, I checked the TS end, and as usual, the front was high. I'd zeroed all the feet when I started, so this time I lowered the back foot till it was just touching, then pulled out the machinist's jack and left the front foot floating. Obviously, that threw off the HS end a little bit, but I just positioned the jack to bring it back level. That was yesterday, when I checked this morning, the bubble has definitely moved, so I'm just going to try and exercise a little patience and leave it all weekend. If this doesn't work, it'll be time to either bolt it to the floor, or build a whole new stand, and still bolt it to the shop floor. Later.

Dave

 

[homebrewed]

While RDM copes with a test rod that is slightly bent, I don't think it will work if the rod isn't perfectly round. I very much doubt that your extruded aluminum tube is uniformly round, and that out-of-roundness could easily vary along the length of the tube. I would be more inclined to believe the results of your SS rod.

For an experiment, chuck up that aluminum tube so you can turn its OD perfectly round, right at the chuck, at a number of places along the rod -- and perform RDM only at those locations. You don't want to make the tube too thin so take off just enough.

I don't know how much, if at all, the measurements will be affected if you use enough clamp pressure on the tube to deform it. So I have an "out" if all you get is more confused by the results. My motto is, the more information you get, the more confused you get . That is, until the light bulb turns on......hopefully.

 

[mksj]

I think one is missing the point on the method, one is only looking at the TIR, but that the swing of the TIR in a +/- direction is the same when the head is aligned. The roundness or trueness of the rod should not be a factor as one is just looking that the magnitude of the +/- is the same at each test point away from the chuck. On that respect you do not need to account for deviations due to chuck axial skew or that the rod needs to be high precision. Things get more complicated when you have bed wear but with a new lathe one levels it, there is no bed wear, and then one can proceed to a 2 ring test or a test bar with no chuck. There still will be slight taper difference depending on the chuck used and cutting dynamics.

This is an ER40 Set-Tru type chuck that I made, it somewhat shows that as one gets further away from the chuck that there is a small increase in total TIR but the +/- magnitude remains pretty much the same. If I recall I used a 9" precision ground rod with 6" extension from the collet.

 

[RJSakowski]

When using the RDM method, it is essential that the bar be perfectly round at both test points.. It isn't necessary that the two points are the same diameter but if they differ, that has to be taken into account so it's simpler if the diameters are the same.. When averaging the readings at each point, it is essential that it be done algebraically; i.e if you read a maximum of +.005 and a minimum of -.001, the algebraic average is (.005 +(-.001))/2 = +.002. It is also essential that the dial indicator not be disturbed as you move from one position to the next.

The theory behind the RDM method is quite simple. When reading minimum and maximum at any position, the algebraic average. is located at the spindle axis. What we are trying to do when removing taper is to align the carriage so that the cutting tool is at the same distance from the spindle axis. When the two RDM averages are equal, we know that a cutting tool mounted on the carriage is equidistant from the spindle axis.

Personally, I find the RDM method to be reliable and quite useful in multiple ways. In addition to running the traditional test, it can be used to check alignment in the vertical plane.. It can be modified to align the tailstock for turning on centers, eliminating the need for a test bar.. I also use it for aligning the headstock on my CNC mill to the z axis ways. The method is nondestructive and quick to implement, allowing multiple successive adjustments and verification checks.

The bottom line, though is to use whatever method you are comfortable with. If you don't have confidence in a test method, you won't be comfortable with it.

 

[daved20319b]

While RDM copes with a test rod that is slightly bent, I don't think it will work if the rod isn't perfectly round. I very much doubt that your extruded aluminum tube is uniformly round, and that out-of-roundness could easily vary along the length of the tube. I would be more inclined to believe the results of your SS rod.

For an experiment, chuck up that aluminum tube so you can turn its OD perfectly round, right at the chuck, at a number of places along the rod -- and perform RDM only at those locations. You don't want to make the tube too thin so take off just enough.

I don't know how much, if at all, the measurements will be affected if you use enough clamp pressure on the tube to deform it. So I have an "out" if all you get is more confused by the results. My motto is, the more information you get, the more confused you get . That is, until the light bulb turns on......hopefully.

I disagree to some extent. One of the claims for this method is that you DON'T need a test bar that's perfect. I also think that turning it at several places along the length is kind of defeating the purpose of the test, as soon as you turn it down, it's going to be concentric to the chuck at that point. I'll also add that the tube I'm using is drawn, not extruded, it rolls very smoothly on a smooth surface, and is very consistent in OD along it's full length, and at various spots along it's circumference. And by consistent, I mean less than a thou variation, although it IS undersize. I misstated the size, it's actually supposed to 1.125", not 1.25", with .125" walls. I'll grant that it's possible I could have deformed it slightly with chuck pressure, but I wasn't trying to crush it, since I knew it wasn't going to turning under power. What really got me going wast the wildly different numbers I was getting. Hey, if a test isn't repeatable, it's not a valid test, and my numbers were anything but repeatable. Are there possible variables in my methodology? Sure, but I just don't believe that I was that sloppy, certainly not sloppy enough for the pretty much random results I got.

So all that said, I'll try again, but with my 4 jaw chuck and dialing in my test bar(s). That's not supposed to be necessary with the RDM, but I'm curious if that will get me more consistent results. Because it's entirely possible that my 3 jaw chuck has some significant wear, it's original to the lathe, so dates back to the early 80's. More later, and thanks for your response.

Dave

 

[daved20319b]

When using the RDM method, it is essential that the bar be perfectly round at both test points.. It isn't necessary that the two points are the same diameter but if they differ, that has to be taken into account so it's simpler if the diameters are the same.. When averaging the readings at each point, it is essential that it be done algebraically; i.e if you read a maximum of +.005 and a minimum of -.001, the algebraic average is (.005 +(-.001))/2 = +.002. It is also essential that the dial indicator not be disturbed as you move from one position to the next.

The theory behind the RDM method is quite simple. When reading minimum and maximum at any position, the algebraic average. is located at the spindle axis. What we are trying to do when removing taper is to align the carriage so that the cutting tool is at the same distance from the spindle axis. When the two RDM averages are equal, we know that a cutting tool mounted on the carriage is equidistant from the spindle axis.

Personally, I find the RDM method to be reliable and quite useful in multiple ways. In addition to running the traditional test, it can be used to check alignment in the vertical plane.. It can be modified to align the tailstock for turning on centers, eliminating the need for a test bar.. I also use it for aligning the headstock on my CNC mill to the z axis ways. The method is nondestructive and quick to implement, allowing multiple successive adjustments and verification checks.

The bottom line, though is to use whatever method you are comfortable with. If you don't have confidence in a test method, you won't be comfortable with it.

Not being argumentative, I appreciate your input, but I have a couple of questions. How round is perfect? I just checked my test bar with a dial caliper, one I know is accurate, and like I said above, it varies less than a thou over the full length and circumference.

My other question has to do with the constant reference to plus and minus dimensions. Where and how do you determine that? Plus and minus what? And how does that differ from taking the total runout and dividing it by 2? What are you using for a zero? Is it just some random point that your indicator happened to land on? I can see that you and I would have ended up with different numbers, I'd have zeroed my dial indicator at the -.001", thus ending up at .006" TIR, divide by 2 and I'd have gotten .003" of alignment error. .001" isn't much, but multiply that by 10 and you're talking significant deviation in results.

Again, I'm really not trying to argue, but it looks like I'm missing something basic to the method. It sounds like you're very familiar with it and trust it, if you can get me straightened out, I'd sure appreciate it. Later.

Dave

PS Quick update, I pulled out a micrometer and checked, this tube is VERY consistent, I didn't bother checking 10ths. but I was getting 1.128" along the length and at 90° at several places around the circumference. So not an issue with my test bar, I think.

 

[RJSakowski]

I check roundness by measuring at multiple clockings with a micrometer. The measurements have to agree to .0001" before I use a bar for the RDM method. There are better ways of doing this but it's the best I've got. Anyway, they work for this use which is what is important. The bars that I use are salvaged from old office.printers or photocopiers and range from .5" to 1.25" in diameter.

As to the second question, the actual value doesn't matter. What you are looking for is the average of one position relative to the second position. for convenience sake, I will adjust my headstock reading so the are equal plus and minus so the average is zero. Then any non zero average at the tailstock is an indication of taper.

 

[RJSakowski]

I would not adjust my headstock in lieu of level or shimming to remove twist. I made that mistake years ago. I was able to remove the taper while turning but I now cut a dish when facing. One has to assume that when a lathe is assembled, that the headstock is aligned correctly.. It is more likely that the ways are twisted, either through rough handling or bolting it to a non flat surface.

My lathe alignment procedure consists. of first leveling the lathe to remove any twist that I may have introduced. Next I check the headstock alignment to the bed, I do this by mounting a bolt on my faceplate. The bolt has a polished head and I place an X with a Sharpie so that I can make measurements on the same point. It is position so that I can touch that point with my dial indicator in a position at the front of the lathe and at the rear by moving only the cross slide and rotating the spindle. I lock the carriage and with the bolt to the front, I adjust the dial indicator to contact the center of the X and zero it. Then I rotate the faceplate to the bolt is at the rear and move the cross slide to contact the X,, without disturbing the indicator. Any non zero indicator reading is due to headstock misalignment and I adjust to correct.

When I am satisfied with the headstock alignment, I then move to the alignment of the ways using the RDM method and shim as necessary to correct. Next, I run the two collar test to verify alignment . The next step is to align the tailstock. I do this again with the RDM method. I mount a bar in the lathe, but instead of running free on the tailstock end,it is run on a center. I made a special split ball for a 1" bar so I can mount the bar in my 3 jaw chuck without introducing any strain on the tailstock center. From there is it straightforward. Adjust the tailstock so the difference in the averages at either end of the bar are equal to the desired taper.

There is one more measurement to be made and that is for vertical alignment of the spindle axis to the ways. This is not normally done as there is no easy way to correct a misalignment and it has a relatively small influence on use of the lathe.