Thread: precision level?

Originally Posted by Dave J

No need to be level other than for coolant flow. What your doing with the level is taking the twist out of the bed by comparing each end with the level.

Dave

Why not a straight edge?

So what if the headstock end was level and the tailstock end was level but they were 10 thou out of flat?

A level just seems to be the wrong tool for the job.

Originally Posted by Greg Q

It just so happens that "level" is a quality that is very easily verified with a bubble level. There is no other affordable instrument that so directly and easily can tell you how each part of the machine is aligned to an external master reference(in this case: the earth). The real goal is "untwisted", and the precision level is the tool used to verify that condition. While a level is used, the ultimate goal isn't to level the lathe, that's just a byproduct of the process.

Greg

Yeah!

That makes sense.

Thank you.

Originally Posted by Stustoys

Hi Avery,
You're mostly right. It doesnt have to be level, just on the same plain(spelling?).
Its just that the easiest way to be sure its on the same plain is to set it up level.
Being level can also help setup sometimes.
Stuart

Yep!

that makes sense

Thanks.

I have a precision level but I hardly ever use it. A lathe does not need to be level. I have my lathe set with a slight fall to the right and to the back. This slight fall ensures that the coolant will drain to the drain hole and into the coolant tank.
The most important thing when setting up your lathe is try and set it so there is no twist in the bed. If your lathe is twisted, you will never produce a parallel shaft and the further out from the chuck, the worse it becomes.
To set your lathe up, find a length of precision ground bar. Set it up in the four jaw with a dial indicator keeping close to the chuck for the initial truing up. Get it to run as true as you can. Now move the indicator to the other end of the bar. The bar needs to be at least half the length or more of the lathe, the longer it is, the better. Rotate the chuck and note how far the bar is running out. Find the highest point of the run-out and turn the chuck until this point is at the very top of the bar. This should only require a 90 degree turn.
Run the carriage up and down the bed and note how far out the dial indicator reads. This amount shown is the amount of twist in the bed. Place shims under one corner of the "feet" and tighten the holding down bolts. Check the twist as before and adjust the packing until the dial indicator reads "Zero". Now rotate the chuck 180 degrees and check it again. If an error shows up, set the packing so the error is halved. This will be as good as it gets using this method. With an accurate chuck and a ground bar, you should be able to set the lathe up to within 0.001" over 24 inches along the bed or even better.
My lathe is mounted on a welded frame made of 50x50x3 mm RHS. I made it with adjustable feet that can be screwed up or down and locked with a lock nut. There is just enough "spring" in the frame (even though it's very stiff and solid) to correct any twist in the bed. I have checked the bed with my precision level after doing this and found that the technique produced the same accurate results as if I had only used the level by itself.
If you have a precision level, place it on the bed at the tailstock end and then lean on the tailstock like so many of us do. Check the level when you're leaning on the lathe and you will find that it will actually move. This is why some of the die-hard machinists get all excited if someone leans on the lathe when they are machining. Leaning on the lathe can cause a substantial error in the finished job.

Originally Posted by Stustoys

Hi Avery,
You're mostly right. It doesnt have to be level, just on the same plain(spelling?).
Stuart

Close Stuart, actually that's the boring one and it's plane

As Stuart and others have mentioned, it's not necessary to either have a precision level, nor level it in order for it to cut accurately. However it helps. I have one and rarely use it, but then it's like a lot of tools, it's handy when it's there and they're not stupidly expensive.

At the end of the day, the aim is to have the lathe cutting as accurately as possible. To do so all the components need to be in the same plane, and the easiest plane to determine is horizontal (as determined by a level). However I can just about guarantee the people who owned lathes like this a generation ago wouldn't have used precision levels, as they were relatively rare and quite expensive (in a home environment). So they would have been set up using a builder's type level and then determined that they weren't cutting a taper. The method Kody described is an excellent way to relatively quickly set up a lathe, however the traditional way is the so called "2 collar" method. As described by others, it simply involves turning 2 collars, one at the chuck end and one at the far end of a 300 mm or so length of bar and confirming they're the same size (ie the lathe isn't turning a taper). However the only reason collars are used is to speed up the process, as only 20 mm or so at each end need to be turned accurately. If you're patient however it's just a matter of turning a bar down the whole length and confirming it has no taper. That's it! The problem is, taking "dust" off a bar for the whole 300 mm takes some time, especially if you're doing adjustments, but that's all there is to it.

Pete

I have always preferred not to machine two collars on the 300 mm shaft. The problem is because of the length of shaft protruding from the chuck. It is all but impossible to machine a 25 mm shaft of this length without at the same time, stopping any chatter. It is important to note that the end of the shaft must not be supported by a center in the tailstock. If the tailstock is used, the bar will simply flex to run true from the center supporting it. This will never show the true twist that may or may not be in the lathe bed. This flexing will occur if you center drill the bar after carefully aligning it in the four jaw chuck with only about 25 mm protruding out.
It is so much better to run the dial gauge along the shaft being held in the chuck and why the chuck is rotated 180 degrees to obtain an average reading. There is one more "slight" error that can be a problem. It's the initial inherent accuracy of the alignment of the axis of the headstock to the axis of the bed itself. This should never have to be questioned but on some (cheaper imported) lathes, there is a possibility for an error to have crept in via transport, dropping the lathe etc. Most lathes are very precisely built and designed to prevent any misalignment from ever happening. I mention the headstock alignment because I have seen cases where the alignment is out of spec.
However, by rotating the chuck as described in my previous post, you can reduce any error to almost zero. This also reduces any errors from the chuck itself and we all know about chuck errors which is why I prefer to use the four jaw chuck for this setup. Hope I have written this so you can all understand it and it is a help to you all.

Kody, as mentioned, the method you use is an excellent one, and I simply present an alternative. A couple of things however, you're correct the tailstock should not be used, and if it is you will measure the tailstock setover. In other words the precise same method is used when it comes time to measure just that, only this time the tailstock supports the work.

With regards to machining such a long piece, if you're getting "chatter" I'd respectfully suggest you're taking FAR to big cuts. You should never get chatter with cuts this light unless the lathe has major problems. The difficulty with machining a part so far from the chuck jaws is in fact deflection. The amount of material being take off is literally "dust" and requires a very sharp tool to be barely touching the work. I have done this many times and it works just fine, unremarkably as it is the traditional way to finally test a lathe.

At the end of the day, regardless of the method used, the final goal is to have the lathe turning without a taper, hence why it is a good idea to do this as a final check regardless of what other method you use to arrive at that goal. Better to find out the lathe is cutting a taper on some scrap piece than discovering it by accident some time later!

Pete

here are two screenshots from the Hercus ToT. They only use 3 1/2" between the collars. This is how I intend to set my lathe up. If it's good enough for Hercus it's good enough for a hack like me.

Jack, yes that is the so called "2 collar" alignment method. As I mentioned, there is actually no need to machine collars as such, only to speed up the process. It is equally as valid to simply machine a bar of the same length and measure the taper (hopefully none) at each end. The longer the bar used, the more sensitive the test however by the same token the greater the potential deflection. A sharp tool and extremely light cuts are both essential.

Pete

Here's a couple of pages from my lathe's manual (Dutch made AI Hembrug, hence the three languages. Another funny thing is the Dutch and German instructions quote metric values, imperial for the English. Circa 1965)

Greg

When I first set up my lathe I needed a precision level but didn't really want to fork out money for something I thought I would only use once. I ended up buying a secondhand starett from US for a good price. Now that I have it, I find I use it probably 2 or 3 times a year to verify that the bed is flat. It's amazing how much a concrete slab floor can move throughout the seasons! Also, since I recently built my shed I have moved my lathe several times before I was happy with it's location (no different to my wife with the couch when we first moved in!) and everytime required it to be levelled.

Of course the proof of the pudding is in the turning and so the true test is to see if it will turn parrallel regardless of how level it is. That's when the turning between centres test can be performed. Not that I'm an expert on that.....

Cheers,

Simon

Originally Posted by simonl

Of course the proof of the pudding is in the turning and so the true test is to see if it will turn parrallel regardless of how level it is. That's when the turning between centres test can be performed. Not that I'm an expert on that.....

Cheers,

Simon

Simon from what you wrote I suspect like many people you are mistaking the test between centres with the "2 collar" test.

The first test is to turn WITHOUT a centre on the tailstock, and is commonly referred to as the 2 collar test. The work is held by one end only in some form of collet or chuck. The accuracy of the holding method is immaterial, just that it be held by one end only. This test will establish that the lathe can turn without a taper.

The next test is THEN between centres. While it's true that it also determines the lathe turns without taper, it is to establish zero setover on the tailstock. The two are commonly confused but measure completely different factors of the lathe's functioning.

Hopefully that helps clarify things a little.

Pete

Hi All,
In measuring if the Head is aligned and if there is any twist in the Bed by the methods posted and as kody has said you are relying on the Chuck to be true.
Would a good way to check is to use if you have them Collets in the Spindle Bore holding a Precision Ground Rod of suitable size, this would take away any error that the Chuck may have.
Any one want to comment on this method if it a good way or not.

Regards,
Keith.

keith, if you are cutting and subsequently measuring it does not require the chuck to be true.

Pete