Haven't posted an update in quite a while. Neither the family nor the few people still paying me money occasionally seem content to let me spend all day every day in the shop. Even the Giants keep insisting on extra inning walk-offs. Such problems!
Anyway, some progress since the last post: I've only (!!) scraped in the outer ways, headstock, and saddle. The headstock is now aligned to [kisses fingers] near perfection:
There was a whole lot of learning involved for me to get it to this point.
1. Attitude
The first thing was to "get mad" as Richard likes to say. I finally got tired of seeing my lathe in pieces, and stopped repeating check after check after check before actually
doing anything.
I think it's a natural tendency for beginners like myself to pussyfoot around for fear of breaking things. Watching my dad try to use a computer used to drive me insane, for example. I spent my entire adult and professional life with computers and have
zero fear of hurting the darned things. With the lathe I'm terrified of making a mistake that can't be corrected. Most mistakes just mean more scraping. Occasionally, they require gluing on some Rulon or phenolic and then more scraping.
I'm still being careful and trying to think things through, but mistakes are unavoidable the first time you do anything in life. So I quit messing about with the inner ways and tailstock, and moved onto the important bits. That definitely paid off, as I learned quite a bit more.
2. Vee Way Angle
Before the change in attitude, I was quite worried about accidentally changing the angle of the inverted vee ways (90 degree included on a Logan). Then I remembered mentioning it to Richard who just shrugged and said to remember that you scrape the saddle in to the bed anyway. A slight change in angle is no big deal.
I think that as long as each side of the vee is
FLAT, it doesn't matter if one side is, say 47 degrees off of vertical and the other 51 degrees.
I think what matters most is that the vee doesn't get wider or narrower at the tailstock or headstock end of the way. In other words, what matters is the center of rotation for each plane. As long as those two lines are parallel with each other, it doesn't matter if one or rotates more than the other.
Put yet another way: Imagine a tiny little man wearing a pair of skis, and resting each one on the outside of the vee. It doesn't matter how much the man's knees spread apart or come together. What matters is that he doesn't go pigeon toed or duck footed — his feet just need to be parallel no matter how bowlegged or knock-kneed he gets.
Keeping the angle of each side exactly 45 degrees off of vertical is
hard (nearly impossible) because each side is only about half an inch wide. As long as both sides are flat, though, it's relatively easy to tell if one end or the other is narrower. One way is simply by feel and by sound: move the (roughly scraped in) saddle to each end test for play. A slightly more precise way is to blue up the ways, slide the saddle to the two ends and examine the resulting marks on the ways (each end should see roughly the same amount of streaked/wiped away ink).
The job of an inverted vee is to keep the saddle from rotating on the horizontal plane. As long as the saddle has full bearing, the precise included angle just doesn't matter (unless someone convinces me otherwise).
3. Headstock alignment
EVERYTHING on a lathe references from the center of rotation of the spindle in the headstock (more specifically, the center of rotation of the Morse taper at the chuck end of the spindle). Unfortunately, it's a purely imaginary line.
You can't indicate an imaginary line! Nor can you throw a level on it.
3.1 Test bars
The way you find the center of rotation is interesting. It requires a test bar: a precision ground bar about 12" long, with centers bored at each end, a morse taper at one end, and a precision ground cylinder for most of the length.
I bought a surprisingly inexpensive pair of test bars shortly after taking Richard's class for the first time (from India on ebay, I think). One MT3 and one MT2 to fit the headstock and tailstock tapers of my lathe.
The right way to validate a test bar is to put it between centers and check for runout. Since I, ahem, currently lack a fully assembled and precisely aligned lathe, I had to use a matched pair of vee blocks on my granite plate. Testing the cylinder portion was easy: just find top dead center with a tenths indicator, and carefully rotate the cylinder. Repeat at each end and in the middle of the bar.
I was shocked at how accurately ground the cylinder portion was. The needle on my tenths indicator
barely moved. Total indicator runout of like 0.00005". Amazing.
You basically test for runout on the Morse taper portion the same way. Unfortunately, because I couldn't put it it between centers it was much harder to test. Even a
tiny amount of movement axially while rotating the test bar will move the needle on a tenths indicator significantly. I tried sandwiching a small ball bearing between the center in the far end and a heavy block to push against while rotating. Using this method I measured about 0.0003" to 0.0004" TIR, but I'm not confident in the result — it may actually be better than this. Tenths indicators are unbelievably finicky beasts.
The last test for the test bar is for the taper itself. You ink a stripe down the length of the taper, insert it into a matching taper in the spindle, twist, then analyze the resulting smear.
The taper in my spindle was pretty grody from decades of misuse. There is a bit of surface rust, quite a few scores, and lots of "character" from use since 1947.
I cleaned it as best I could, then took a MT3 hand reamer (one of the most expensive cutters I've ever purchased) and
very carefully removed any remaining burrs and crud, twisting the reamer purely by hand. I did
NOT insert the reamer fully and turn it with a wrench to cut a new taper, of course, as there would be no (easy) way to precisely align the taper with the spindle axis. I just removed the burrs.
I then inserted the test bar with a stripe of spotting ink, and couldn't have been happier with the result:
That's an excellent fit. And a very well made test bar. The wider unstreaked areas were due to worn grooves in the spindle taper, not anything with the test bar.
3.2 Testing the spindle
Next, I needed to ensure the spindle and spindle bearings themselves were reasonably precise.
First I indicated for runout on the spindle itself. A little less than 0.0002" TIR. Honestly, not as good as I was hoping for (hey, even I can scrape to a few tenths) but probably as little as could be expected from a very old hobbyist lathe:
Logan's use a screw-on chuck. I also tested if the surface the chuck references against had any cam action going on. Again, the surface quality is far from perfect, but I'm still quite satisfied with the results (and some careful stoning should eliminate the one little bump you see):
3.3 Finding lines on the same vertical/horizontal planes as the center of rotation
Finally we get to the thing I found most interesting in the whole process.
I've now validated that the test bar is astonishingly accurate, and that the spindle itself is in reasonably good shape. Time to actually use the test bar for its intended purpose.
There is nothing like a tenths indicator and a 10-12" lever arm to reveal the
tiniest little bits of dirt, inconsistent pressure, etc.
This is the very best result I saw with the test bar inserted yesterday:
That's about 0.0001" TIR at the chuck end, and 0.0009" at the tailstock end.
That's the best result, that I was never able to repeat. Results vary every time you remove and re-insert the test bar. Sometimes it would move as much as 0.005" TIR at the tailstock end!
No matter what, the spindle and test bar tapers aren't perfect. Minute specs of dust will invariably kick things around a bit. It doesn't matter how carefully you re-insert the test bar, the end is pretty much guaranteed to show some runout, and the bar will basically circumscribe a
cone.
Here's the thing, though: it really doesn't matter too much!
With one little trick, that cone can still reveal a line in the same plane as the axis of rotation.
Here is possibly my favorite passage in all of Edward F. Connelly's fantastic book, Machine Tool Reconditioning:
"To nullify the effect of eccentricity error, the mean position is located at the vertical diameter for this portion of the test."
That is some
seriously opaque technical writing, especially for something so important to understand. It's impossible to read that without your eyes glossing over! I must have read it two dozen times without understanding until the penny finally dropped.
As usual, the underlying concept is critically important (and actually kind of cool):
- Place the indicator tip at roughly top dead center at the end of the test bar.
- Rotate the spindle to find a high point and a low point.
- Set a zero at the low point, then rotate to find the high point (which should be 180 degrees away). Let's say the high point reads 0.0016".
- Now rotate the spindle to find the halfway point precisely between those two values (rotate until the indicator shows 0.0008").
Since the test bar circumscribed a cone, the line formed at the top of the test bar is now on
the same horizontal plane as the axis of rotation.
That is, the line formed between a point at top dead center at the headstock end, and one at the tailstock end will be twisted fore or aft around the vertical axis, but it doesn't tip up or down at the headstock or tailstock end — it's precisely aligned with the axis of rotation.
If you now sweep an indicator mounted on the carriage along the length of the test bar, with the tip at the top of the bar, you can measure exactly how much the spindle axis is tipped up or down.
The test then needs to be repeated once with the vertical plan, and once with the horizontal plane (once with the indicator on top, and once with it on the side of the test bar). In both cases you just rotate to the halfway point to find the true axis of rotation.
That is what I'm showing in the first video of this massive post: the outer ways of the bed are guiding the carriage along a path precisely aligned with the axis of rotation of the spindle.
Whew! I hope someone finds this as interesting as I do. (laugh)
