And most likely not perfectly round either (usually they are ground S355 + hard chroming or 42CrNiMo with a hard chroming). And when you chuck something 30" long for a couple of inches of gripping range, no matter what type of chuck you have, you will most definately get a runout.
Originally Posted by .RC.
Originally Posted by lazlo
And I'm "pretty sure", after actually reading it years ago, that you'd do about as well by just oiling the machine and starting to use it.....
if you try to follow RDM to actually make a basic alinement *from scratch*, and not to just "tweak" an existing "known to be reasonable" alinement, you'll likely twist the bed into a pretzel chasing the numbers... That applies DOUBLE to the usual "practitioner", who is new to all things machinery, and reads about the "magic secret knowledge" known as "Rollie's Dad's method".
Such an individual is not likely to understand when the adjustments are getting ridiculous and it is time to stop and do something different.
Plus, I challenge ANYONE to explain why and how making dozens of delicate measurements with an indicator (every newbie has lots of THOSE, and knows all about using them, right?) is simpler and easier than making TWO simple micrometer measurements per trial with the "TCT" (two collars test).
Especially when "TCT" actually tests the results you DO get, not "the potential error you might get if these delicate measurements are ALL exactly correct and everything works out just exactly right".
Rollie may be a great guy, and Rollie's dad may have been a smart cookie, but it STILL looks like the very longest way and roughest road by which to get to the result.
I will say that RDM does not depend on the test rod being straight.... but it DOES depend on the rod roundness being better than the error you can tolerate.
Last edited by J Tiers; 05-21-2012 at 09:24 AM.
Originally Posted by J Tiers
I didn't realize your criticisms were theoretical and based on reading RDM rather than actually trying it. I'd be interested to hear what the dozens of measurements you mention are and what they're used for.
I've actually used RDM and had good results from it. By that I mean that I reduced the taper cut by my lathes to acceptable, which was my original goal.
There seem to be a few versions of the "two collars test" = TCT. The descriptions I've found are:
http://www.wswells.com/data/howto/H-3.pdf South Bend
http://www.wrathall.com/Interests/ma...20accuracy.pdf J. Latta
http://www.chaski.org/homemachinist/...it=RDM#p101250 J Tiers
The J Tiers link is to a thread from several years ago where you raised a number of issues with RDM and I learned more about it. When I raised questions about use of the TCT per your description you didn't address them but instead said that lathe manufacturers used the TCT so it shouldn't be questioned., i.e the Argument from Authority fallacy.
It's interesting that in Forrest's description of the TCT he indicates that making the cuts on the collars is much more difficult than your description linked above would lead one to expect. In fact, he suggests using a piece of wood to damp chatter and polishing the resulting collar area to reduce roughness. Converting theory to practice can be messy
My take on RDM is that it allows one to measure the deviation between the spindle axis of rotation and the ways. These may be non-parallel because the headstock is mis-aligned and/or the bed is twisted. In Forrest's description of TCT he says this is the complicated part... but then doesn't say how to separate these causes. My thought is that RDM allows me to get information equivalent to that provided by the TCT but more quickly and easily.
I'm still learning about RDM so detailed info about WHY it doesn't work rather than broad generalities would be helpful. I'm trying to figure out, in detail, what I've misunderstood about RDM since my experience is quite different from that reflected in this thread.
Originally Posted by GadgetBuilder
The comment was not that it "should not be questioned"...... as you suggest...
In contrast to your erroneous remark, the idea ACTUALLY was that since that is how the manufacturer tested, it makes sense to use it again... you are more likely to get the same results in the same conditions.....
RDM is aimed at the novice, the one least able to get consistent results from it, as I think most would agree... That is my objection..... and why I bother to spend the time typing... I don't have to, and I don't give a rip if you "listen" or if you think I am an ass..... Doesn't hurt me a bit, I've got thick skin.
Of course, as far as the so called "authority argument" is concerned, since you bring it up.....if the manufacturers thought RDM was better, they'd use it.
They could use ANY method they want, and after taking a lot of trouble to make a very accurate lathe, they could afford any test that was best, most representative, etc. It seems significant that the industry accepted test is TCT..... not because "they say so", but because they have settled on it as the best and most effective they can devise.... and because there are precious few arguments AGAINST it, (or in favor of RDM).
Dozens of measurements? I freely BRAG that I exaggerated....
It might only take 6 or 8 tries per measurement, to be sure the reading is the same..... and that you have actually reached the accurate max or min, etc, etc.
And .........you have a high and low to measure, and then mathematically average...........
I never said I hadn't tried it, you never asked..... I did try it, and rapidly decided to go back to TCT, which is far easier, and involves actually USING the lathe.
I have REPEATEDLY said that it "works" for the exact same measurement as TCT...... if you want to bother.
if you love it to death, go ahead and use it.... but do not make it seem like the "secret method that cures all ills, eliminates levels, that has been suppressed by the level industry"..... it ain't, and that has been proven.
I found it fiddly, and in the end inaccurate.
Rule #1 of metrology....... The best method is nearly always the one that involves the least measurements, and the measurements that are easiest to make accurately.
Compare TCT and RDM on that basis...... if you STILL prefer RDM, that's your privilege, and your own problem....
Not professing RDM, (which does not require a straight rod) just posted the runout after "roughly" leveling the lathe with a .0005"/in level. It improved by .002".
Originally Posted by .RC.
I plan on doing both RDM and TCM and comparing the results. At present I don't have any pipe to turn to do TCM.
I did put the "ram" in the tailstock however, and ran the indicator down the length of the side with each jaw of the chuck in the 9 oclock position. Deflection was less than .001" in each of the three positions.
It's as round as I need it. Less than .001" run out at the chuck. Chucked 3 times and measured 3 times.
Originally Posted by Jaakko Fagerlund
I can't resist....... was it still too short?
Originally Posted by customcutter
Yes, but only by .0001" on the last try.
Originally Posted by J Tiers
so, from experience, how much can you "twist the bed"? the alternative being to align the headstock.
for example, if i am getting a conus of 1 : 0.0002 (double that for difference in diameters), do i work on the jack screws? if i get 1 : 0.002, do i move the headstock?
Much depends on the exact construction of the machine, mainly its weight and weight distribution and if it is bolted solidly or just standing on its feet.
Originally Posted by dian
If the machine is very heavy and solidly bolted, you can twist it until it cracks in half from somewhere. If it is just standing on the legs, the weight pretty much determines how much it can twist/bend. But I'm sure you can easily achieve a great twist when trying hard