Checking the Runout and Squareness of my PM-940V

[ptrotter]

I have had my PM-940V mill for some time now and have made several mods to it as I have shown in previous posts here. It seems to work well enough for my purposes, but I decided it was time to measure the run out and squareness. I made a video of the process and posted it to YouTube. You can look at it here:

.

There are 2 things in particular that I need to address. First the head tram is off by .0035 over 5 inches in the Y direction, which is quite a bit. This is greater than the runout of column to the table, so the head is also not square to the column. I'll play with the gib a little and then I'll probably re-shim the column more accurately and add a shim between the head and the Z saddle if I have to.

The second item I need to address is some cyclical variation in the runout that occurs primarily on the Y table movement when the table is near the front of its travel. This is quite visible in the video where I am checking the squareness of X and Y axes. This must have something to do with the ball screw as that is the only thing I can think of that could have this kind of effect. Maybe the ball screw is slightly bent or misaligned. This will take some investigation. I might be able to dampen this effect by tightening the gib, but I need to be careful not to overtighten.

I might be able to improve some of the other areas by working with the gibs, we'll see. Other than that, it is pretty much what I expected from a Chinese mill. I think it will work fine for my needs, after all, I am just a hobbyist.

Please take a look at the video. I would be interested in any ideas or comments you might have.

Making this video has given me a great appreciation for the YouTube creators that have posted so many videos I have learned so much from. These videos are not easy to make and are quite time consuming. I am obviously not a great video maker.

 

[davidpbest]

Paul, good video. Your methods of checking alignment are sound and follow the techniques I use and recommend. I know you’re doing a CNC conversion, and thus probably don’t care whether the quill travel is aligned to the head travel on the column, but it would be interesting to see what that shows on your machine just for reference by others. The term ”runout” I generally relate to rotational variation rather than linear, but I get your point. “Linear alignment” might be a better choice of terms. Anyway . . . . Here are some reactions to what I see from your video.

I wouldn’t trust the back edge (or the T-slots) of the XY table to be aligned with the X-axis travel of the table - that’s asking a lot of a mill of this caliber. I know the back edge is ground, but I wouldn’t rely on that as a reference surface for alignment. Same for the front edge.

The differences in linear variations you observe when moving the XY table in opposite directions suggest to me there is sloppiness in the gib-to-dovetail interface between the XY table and the saddle, and between the saddle and the machine base.

The source of the cyclical variations you see in the Y-axis travel are probably related to the ball screw. However, if the gibs were fitted and adjusted properly, I would not expect to see as much of the ball screw eccentricity to be telegraphed into the table - I would expect the gibs to do a better job of constraining the linear variations even if the ball screw is bent slightly. Consider the loads generated by an end mill cutting into something like 4041, and you’d expect the table deflection to be more tightly constrained than the cyclical variations you observe. For this reason, I would encourage you to remove the gibs, stone them for burrs, and check them for flatness. Then I would test their fit against the dovetails by bluing up one of the surfaces and checking the length of contact between the dovetails and the gib.

Here are a couple of video resources that might be helpful. First one is Keith Rucker illustrating how he checks for gib fitting. That video is here. The second are some videos I produced that illustrate one method of checking gib adjustments - these are based on the only published empirical method I’ve seen documented, taken from the Wells-Index mill user manual. Those videos are here. Obviously, the knee gib adjustments don’t apply, but the head-to-column gib adjustment is probably even more critical since you’re depending on that to be tightly controlled under cutting loads.

The tramming gauge (dual indicator setup) indicates to me that the spindle is not aligned to the XY table in the Y-axis. I would recommend the following steps (in sequence) to deal with this aspect. First, get the head-to-column gib properly fitted (bluing it up and checking the amount of surface contact) and adjusted (checking the amount of head displacement using an indicator when side loads are applied - one crude example test is shown here at about 9:00 in the timeline). Next I would shim the column as required to get the Z-axis head travel dialed in to under 0.001” over 12” travel (I was able to do that with my RF-45, although it was tedious, but only have to do it once). Then I would dial in the spindle-to-XYTable tram shimming (or scraping) the head-to-column interface as required.

Hope this helps. I’ll be interested to follow your progress. Your PM-940 saddle appears to be well machined in terms of X and Y travels being perpendicular to each other which should be a relief.

 

[ptrotter]

David,

Thank you for you comments, I appreciate your insight. I knew that "runout" was not the correct term in some cases but I couldn't think of the proper one, but I figured people would understand. Thank you for the better term.

I also know that running the indicator along the back of the table is not very useful as any work piece or work holding device is going to be indicated to the spindle anyway, but I find that the more information, the better and it only took a couple of minutes. I thought it might give me a little sense of how well they machined the parts. It did demonstrate t the variation of linear alignment moving in different directions, so I will play with my X gib some to see if it will go away. I watched your videos and I will make those checks.

I have watched Keith Ruckers videos on making a new gib and they were very interesting. Someday I might get a surface grinder and can try it myself. Gibs seem to be a general issue with these mills. They must not spend too much time fitting them properly. I have spoken at length with Dave L. (B2) about the Z axis gib problems he is having with his PM-940CNC and I recall a recent thread from someone whose gibs drop down several inches below the Z saddle which would make it impossible to tighten properly. After watching Keith's videos, I have a much better understanding of gibs, so I will be looking at mine much closer. I'll probably put some dye on them and see how they are fitting.

I have a couple of things to mill which have a small work envelope so any misalignment won't be an issue, but after I am done with those, I will re-tram the column to the table and the head to the table to try and get them much much closer. It is tedious but not difficult.

Fortunately I don't seem to have the issue that Clough42 ran into on has PM-940. I am not exactly sure what his issue was as he never completely explained it, but PM tells me that they think it was a one off issue, so I am not too concerned.

By the way, I cannot check the quill travel due to the PDB fittings and the pulley I have attached to the top of the spindle for my spindle encoder. I can't lower the quill at all. I may remove the quill down feed assembly completely to clean up head a little.

Again, thanks for your comments. I have learned a lot from your posts and the videos of your mill and lathe rebuilds.

 

[davidpbest]

Paul, based on a few conversations with James about his 935, and have come to believe his 940 was an exception - a saddle that was essentially machined improperly and not something all these mills suffer from. It's not unrealistic to expect the X and Y axis to move perpendicular to each other. That said, I still regard these Asian machines built to a price point for the hobbyist as "kits" with lots of potential, screaming out for enhancement - even the ones made in Taiwan. Someone buying a Chinese-built machine on the basis of "lowest cost" (even considering options) and expecting performance of a Haas or Hardinge is beyond the pale IMO. I had been considering a CNC conversion like yours last year and had my eye on the 940 as the target platform. Frankly I wish it was available as a "castings only" kit. But in the end, I decided that I'd attempt to quell my fascination with CNC buy buying a 3D printer, then evaluate if I really "need" the experience of building a subtraction-based technology. Even then I would be tempted to do a granite platform from scratch. Looking forward to following your project.

 

[ptrotter]

David, I had no illusions on what I was getting in a Chinese mill. I made a conscious decision to go that way. I had considered the 833 but wanted the larger X travel of the 940. As much as I like to measure things in microns, I knew that very little of what I might want to do needs that level of precision or accuracy. I figured that over time I could tweak the best I could out of it and use the difference in cost for more tooling. I will tram it better and probably improve things by adjusting the gibs and that will meet my needs. I tend to spend so much time trying to make tools perfect that I never get to use them. Now if I can just find a surface grinder...

 

[paradox_pete]

I wouldn’t trust the back edge (or the T-slots) of the XY table to be aligned with the X-axis travel of the table - that’s asking a lot of a mill of this caliber

Knowing how square the edge of the table is to the X axis travel can be very useful. I use a "Toolmakers" vise on my mill, with all surfaces ground square and parallel to pretty close tolerance. Having verified that the back edge of the table is also pretty true to the x-axis, I can align the vise very closely with a machinists square. This is very quick and easy, and means I don't hesitate to remove the vise when it's desirable.
-Pete

 

[davidpbest]

Knowing how square the edge of the table is to the X axis travel can be very useful. I use a "Toolmakers" vise on my mill, with all surfaces ground square and parallel to pretty close tolerance. Having verified that the back edge of the table is also pretty true to the x-axis, I can align the vise very closely with a machinists square. This is very quick and easy, and means I don't hesitate to remove the vise when it's desirable.
-Pete

If the table edges are not parallel to the x-axis travel, they aren’t good references surfaces for alignment. If you know they are off by say 0.004” over 8” of travel, I’m not sure how helpful that knowledge is other than to not use the table edges for alignment.

 

[RJSakowski]

I have had my PM-940V mill for some time now and have made several mods to it as I have shown in previous posts here. It seems to work well enough for my purposes, but I decided it was time to measure the run out and squareness. I made a video of the process and posted it to YouTube. You can look at it here:

.

There are 2 things in particular that I need to address. First the head tram is off by .0035 over 5 inches in the Y direction, which is quite a bit. This is greater than the runout of column to the table, so the head is also not square to the column. I'll play with the gib a little and then I'll probably re-shim the column more accurately and add a shim between the head and the Z saddle if I have to.

The second item I need to address is some cyclical variation in the runout that occurs primarily on the Y table movement when the table is near the front of its travel. This is quite visible in the video where I am checking the squareness of X and Y axes. This must have something to do with the ball screw as that is the only thing I can think of that could have this kind of effect. Maybe the ball screw is slightly bent or misaligned. This will take some investigation. I might be able to dampen this effect by tightening the gib, but I need to be careful not to overtighten.

I might be able to improve some of the other areas by working with the gibs, we'll see. Other than that, it is pretty much what I expected from a Chinese mill. I think it will work fine for my needs, after all, I am just a hobbyist.

Please take a look at the video. I would be interested in any ideas or comments you might have.

Making this video has given me a great appreciation for the YouTube creators that have posted so many videos I have learned so much from. These videos are not easy to make and are quite time consuming. I am obviously not a great video maker.

Paul, I have done similar checks and adjustments on my Tormach PCNC 770 mill. The Tormach mill is similar to your mill in configuration except it doesn't have quill travel to deal with.

As I see it, the conditions for your mill alignment are table parallel to x and y ways, The x ways perpendicular to y ways, the z ways perpendicular to the x and y ways, the quill travel parallel to the z ways and the spindle axis parallel to the quill travel. Spindle runout is another concern.

One thing that I found to be a significant factor is gib adjustment. The adjustment procedure recommended by Tormach is based on monitoring an increase in lost motion due to friction increase as the gib is tightened. The adjustment suggested by Tormach doesn't account for table lift or side play and it results in loose gibs.

The Tormach 770 has a fault in this regard, IMO. The stepper drive is located in a casting on the left side of the table. This additional weight places the center of gravity beyond the support of the ways when the table is moved to the left of center which results in the table tilting and raising the table with respect to the spindle to the extent that gib play allows. I see this as a design flaw. The x ways could have been made longer and the saddle could have been made wider. But it is what it is.

I have posted multiple times on HM on subject of mill alignment. Here is one post that discusses some of the issues.https://www.hobby-machinist.com/threads/g0704-accuracy-question.60622/#post-499447
Regarding head tram, I first ascertained that my column ways were perpendicular to x and y ways. The next task was to adjust the z axis gibs. Determining the amount of play in the z ways is more difficult as there is a little over a 100 lb. preload due to the weight of the head. To address that, I used a dial indicator to measure head deflection as I measured force exerted on a scale. I found that about 130 lbs. of force, I began to lift the head. I adjusted the gibs to minimize the free play. I estimated that the de facto pivot point was about 6" from my measurement point.

With gib adjustment optimized, I then checked tram. I was able to make slight adjustments to tram in the x direction by rotating the head. The y adjustment was accomplished by use of shims. https://www.hobby-machinist.com/threads/improved-tramming-for-the-pm-833t.97549/

I have a white paper started regarding mill adjustment. It is srill a work in progress but if you're interested, I will share it.

 

[RJSakowski]

If the table edges are not parallel to the x-axis travel, they aren’t good references surfaces for alignment. If you know they are off by say 0.004” over 8” of travel, I’m not sure how helpful that knowledge is other than to not use the table edges for alignment.

There is no specification published for parallelness of the back edge of my Tormach 770 table to the x axis ways. I use it as a reference for rough alignment but I always check by sweeping an edge for critical work.

An assumption made by many is that machined features are perfect when in truth there is no such thing. What is true is that the feature are machined to an acceptable tolerance to meet the needs of the feature.

A good example is a pin gage. The diameter of the pin is specified to a few tenths and roundness is implied by that specification but nothing is stated about cylindricity of the pin. It is probably fairly good but without certification, shouldn't be relied upon. There is definitely nothing stated about length and while my pins are about 2" long, they are not precise.

Another example is parallels. while they are specified as parallel in the width direction and have specified tolerances for those widths as well as matching pair tolerance, nothing is said about the thickness or length of the parallels.

Wise words to work by are trust but verify.

 

[ptrotter]

@RJSakowski Thank you for your response, it is very informative. I would definitely be interested in your white paper.

I’m going to be away from the shop for a few weeks but when I get back I am going to spend some time adjusting everything to see how accurate I can make it. I am also going to pull the gibs and inspect them carefully. I will probably put some dye on them and see if I can determine how well they fit. I do not think gibs are made very well on these machines.