Accuracy of Machinist Levels

[ChandlerJPerry]

What is everyone's preferred minimum accuracy for a machinist level? I see the Starrett 98 mentioned a lot which I believe is .005" per graduation on the vial. I know quality measuring equipment isn't cheap, but my budget is unfortunately tight (laid off due to covid related slowdowns), so I was wondering if anyone has any experience with other levels and what accuracy is really useful for the home shop. Something tells me a .0005 level might just lead to ripping hair out in the garage!

I was looking on eBay and found a 4" Stanley No. 34 for a reasonable price, it looks very similar to a Starrett 98 but I can't find any published statistics on the accuracy of that model. There are also plenty of Stanley No. 36, 37, and 237 on eBay which have adjustable precision ground glasses, but are marketed more towards high precision carpentry and plumbing, etc. in vintage Stanley catalogs. So what levels are everyone using for their machines?

Just for showing off's sake, here's my newest level that I purchased on a whim before realizing it wasn't nearly precise enough.
It's a Stanley No. 39 1/2, 6" long, marketed as a machinist level in their catalog but definitely more suited to general leveling tasks, it may be more precise than a similar carpenter's level, as the glass seems significantly longer to me and it is enclosed in a heavy cast iron body. However it lacks adjustability and fine graduations for precision leveling, as well as the glass not being precision ground. It is very pretty though! The logo dates it as being from 1922-1935.

 

[GunsOfNavarone]

Personal experience, and I wolud go look at the thread on this subject by @Aukai , I wouldn't even recommend the SUPER high res levels. It's just not necessary for what we do in the home shop and it will frustrate the heck out of you.

 

[benmychree]

A level with .005 graduation is nearly worthless for machine leveling, it will only get you close enough to go to the .0005 grad. level, Having said that, a precision level is not entirely necessary for small lathe (alignment), the two collar method will be adequate for that chore.

 

[Papa Charlie]

I purchased a Starrett 98-6 on Ebay for around $90. Thought I got a good deal. In excellent condition in original box. There is also the Polish made ones on Ebay. From what I have heard, they are very good quality, but just less known. But I am no expert on these.

 

[Larry$]

20+- years ago I bought a precision level made in Poland. I was setting up CNC equipment and this one was recommended by a service tech. I'm pretty sure you can still buy them new or find a used one that hasn't been abused. Heavy cast iron V bottom. Plastic finger holds so the heat of your hand doesn't throw it off, a small cross vial. Very nice BUT it can drive you nuts because it is so sensitive that a sheet of paper under one end (it's 10" long) will move the bubble quite a lot.

I don't know what needs to be that level in a hobby workshop. A lathe can be out of level by an inch and not affect much of anything.

 

[GunsOfNavarone]

Yes, the most frustrating time of my life in the last 10 years was trying to calibrate my machinists level, A h3ll I would only wish on my worst enemy. I still have nightmares and wake in a cold sweat....Sorry, I meant to say it wasn't that bad.

 

[RJSakowski]

Levels are one of the few self proving metrology instruments. It is capable of being calibrated using itself and a relatively flat aqnd level surface.

A level has only two important requirements; sensitivity and a flat bearing surface. Garden variety levels typically are fairly low sensitivity. A half bubble off level is around 1/4"/ft. That is .250"/ft. This includes carpenter's and plumber's levels. Contrast that with the Starrett 98 series levels which have a sensitivity of .005"/ft. That's fifty times as sensitive. A step further is the Starrett 199 series which has a sensitivity of .0005"/ft or ten times more sensitive yet.For those not willing to shell out for a 199, there are levels on eBay made after the Polish level design for well under $100. The one that I bought is an 8" level and has a sensitivity of .0005"/10" or .0006"/ft. So far, I have not had any problems with it.

 

[pstemari]

(redundant)

 

[ChandlerJPerry]

Thank you everyone for the advice! Are these the inexpensive levels on eBay that everyone is speaking of? I can find what looks like the same 8" level from any number of sellers for around $60. I am concerned that I won't be able to level my lathe very precisely, I'm bolting it to a 2" thick industrial countertop made of very heavy particle board, which I think is rigid. However the legs are 4x4 posts that are braced with 2x6, I am worried that with the tendency of wood to expand and contract that this could introduce twist. Will using such a precise level with my chosen mounting method be a hopeless endeavor?

 

[mikey]

Chandler, perhaps a better way to discuss precision levels is to tell you what the overall procedure is, so that you can understand the role a good level plays.

Leveling a lathe simply means we are trying to get the ways of the lathe to be co-planar so that there is no twist across the bed. The reason we do this is because an un-level lathe will turn a taper into your work piece. The way we remove twist is to use a level across the ways at the headstock end and at the tailstock end and use leveling adjusters under the lathe, under the stand or both. When both ends of the lathe are level that indicates that both ways, front and rear, are in the same plane. Typically, using a level like this will get your lathe very close to level but it may/probably will still produce tapered cuts unless you fine tune the leveling even further. The way we do this is with a 2-collar test that allows you to really fine tune your levelers to ideally remove all evidence of a taper in the work piece. Go look up the 2-collar test to see how this is done. When both collars are the same diameter after test cuts, the lathe is level. Know that all lathes move and a preliminary leveling must be repeated in a week or two and then checked every so often to make sure nothing moves.

Now, with that background, the role of a precision level is the same as a less precise level. For example, a Starrett 98-8 is sensitive to 0.005"/foot, whereas the level you showed above is sensitive to 0.0002"/foot. As noted by the other guys, a 98-8 is not all that sensitive but it will still get you close enough and you can use a 2-collar test to get you the rest of the way. The advantage to the more sensitive precision level is that it will get you closer a lot faster, so that you will do far fewer trial cuts with your 2-collar test. This may not seem like a big enough difference to justify the cost of a precision level but it can mean hours of saved time, literally.

I have both a Starrett 98-6 and a Kinex precision level sensitive to 0.02mm/Meter and the difference in time that the better level saves can be significant. I use the Starrett to get me close, then switch to the Kinex to get me closer. When the Kinex says I'm level, I am maybe one or possibly two cuts away in a 2-collar test to dead level, and we're talking about resolution in the low tenths to zero over a 10-12" distance.

So, think of the need for a precision level in terms of time and frustration. The more sensitive the level, the less time you will spend doing test cuts.

As for your bench, as long as it is solid and the whole top provides a stable platform, it should be fine. Just make sure you have some way to level the lathe and the stand so you can make the adjustments you need.