The Chinese precision level teardown

[graham-xrf]

@RJSakowski :
We agree on the method, though maybe use some terminology differently.
If the plate underneath is level, then the bubble will stay in the same (wrong) place, regardless how you rotate it.
Thus for Case 2: Reversing the level, in my mind, does not change the direction of the bubble, but that of course, is relative to the level.
Relative to the rest of the room, I suppose it does look as if it has changed. The fix is obvious.

Getting to Case 3: Yes indeed. Adjusting to have the level read the same amount off when the level is reversed has fixed the calibration. All it means is you have calibrated correctly, even though using using a line through the surface that was not level. We don't have to use horizontal as the calibration point. We can use something a bit off, because reversing allows us to get there anyway. Having calibrated, most folk will do the obvious, and seek the proper rotation, to see the bubble in the middle, and then reverse it, and expect to see the bubble again end up in the middle.

I suppose most of us have had to tangle with calibrating levels at some point. I put down a 123 block against it, to be able to set the level down again exactly reversed. I will obviously have to be doing more of this, but first, I wanted to show folks what it looked like inside, and how much it cost. At some stage, lacking an autocollimator, I will be walking a level in equal increments along a lathe bed, plotting the ups and downs.

Your point about the little cross level needing to also be correct is interesting. I thought that in a tilted plane, there was only one line through that could be level. The line at 90° to it can hardly also be so. To get both vial to read OK, the surface under had better be set up level beforehand.

 

[graham-xrf]

Can you kindly explain why you suggest a calibration on each use? Thanks

It's so very easy to do. Put a level down on something - see it be some amount out of level.
Reverse the level. What you see had better be the very same reading of amount out of level.
If you see a different amount out of level when you reverse it, then something is wrong with the level, not with the thing you are measuring!

 

[WobblyHand]

@RJSakowski :
We agree on the method, though maybe use some terminology differently.
If the plate underneath is level, then the bubble will stay in the same (wrong) place, regardless how you rotate it.
Thus for Case 2: Reversing the level, in my mind, does not change the direction of the bubble, but that of course, is relative to the level.
Relative to the rest of the room, I suppose it does look as if it has changed. The fix is obvious.

Getting to Case 3: Yes indeed. Adjusting to have the level read the same amount off when the level is reversed has fixed the calibration. All it means is you have calibratedcorrectly ,even though using using a line through the surface that was not level. We don't have to use horizontal as the calibration point. We can use something a bit off, because reversing allows us to get there anyway. Having calibrated, most folk will do the obvious, and seek the proper rotation, to see the bubble in the middle, and then reverse it, and expect to see the bubble again end up in the middle.

I suppose most of us have had to tangle with calibrating levels at some point. I put down a 123 block against it, to be able to set the level down again exactly reversed. I will obviously have to be doing more of this, but first, I wanted to show folks what it looked like inside, and how much it cost. At some stage, lacking an autocollimator, I will be walking a level in equal increments along a lathe bed, plotting the ups and downs.

Your point about the little cross level needing to also be correct is interesting. I thought that in a tilted plane, there was only one line through that could be level. The line at 90° to it can hardly also be so. To get both vial to read OK, the surface under had better be set up level beforehand.

I'm really happy this is being all put in print. Thanks everyone for participating!

When I went through this, I was really floundering, not quite knowing what to do next. Was even lacking the vocabulary to discuss it properly. The proverbial not knowing enough how to ask the question. Hardest part for me was understanding I needed to find a line that was level enough to be within the range of the level. When your level was not just uncalibrated, but de-calibrated ( yes, it was me fiddling with the controls ) it was not easy to determine the next step.

When my flat was plausibly level, the cross vial was pretty good. I think if the cross vial was off, you'd have to be within a fraction of a degree of true to have a remote chance, whereas if it is level, the region around the true line is broader. When my cross vial was off, I couldn't find the magic line, and I tried for hours. After I made the tri-legged table for the flat and had repeatable adjustments, I rapidly converged.

Can't say my level is 100% calibrated, but it's within a division. Really difficult to adjust without wildly overshooting, at least on mine. Considering a single division on my level is a tilt of 20 parts in a million, that's good enough for me.

 

[graham-xrf]

I'm really happy this is being all put in print. Thanks everyone for participating!

It started out as just me showing some pictures of the insides of a low-cost level. I am real pleased you found it useful.

Can't say my level is 100% calibrated, but it's within a division. Really difficult to adjust without wildly overshooting, at least on mine. Considering a single division on my level is a tilt of 20 parts in a million, that's good enough for me.

Definitely good enough! We don't absolutely have to have things level, but finding out where stuff is located on a machine relative to other parts of itself, and how straight something is along it's length, is a whole lot easier when we can reach for the free horizontal reference the Earth provided us.

I reckon a 0.0005"/10" level which is 10.3 arc-seconds is likely to be just as useful, and would settle a lot faster. Many levels were made 0.0005" per foot, which is slightly more sensitive at 8.6 arc-seconds.

 

[graham-xrf]

Interesting mechanism. Thanks for doing the tear down. Quite interesting to see the mechanism. Mine seems to be a 150 mm cast iron body and uses a different adjusting mechanism. Instead of a screw point and spring, my cheap level uses a pair of opposing pin spanner nuts and what are called spherical washers that sandwich the bubble arm. Haven't taken it apart yet, but looks more like star washers. I had to find and watch a video in Mandarin that showed someone adjusting the level since the instruction sheet was incomprehensible. Thank goodness for subtitles.
View attachment 392181
My level was a handful to calibrate. Unlike you, I didn't have a known good level to find a level enough spot to do the calibration. I had a 25 cent bubble (with no lines) that I roughly used to level my toolmaker's flat. It was a vial that was encased in acrylic that I had to turn 4 times to find the side that seemed to be closest to "level". I made a cheap 3 legged adjustable stand for my flat and adjusted the flat so at least the cheap level seemed to be centered in both axes. I made the legs with some fine pitch thread so I could coax the flat to be level. Even with that

My desk and workbench were nowhere level enough to find the magic level line. By that I mean there was a line, but it was almost impossible to find. Once I had leveled my flat a bit, it was far easier to find that magic spot, or at least somewhere near that spot. Once near the magic line, then it was a lot easier to proceed. A that point, at least there was some sort of indication in which way to adjust the mechanism!

When you examine things in the 0.02mm/M domain, not many things are solid or stable. (Kind of like measuring vise jaw lift!) My floor influenced the levelness of my stand. This is understandable, since it was a wooden floor. Had to place myself perpendicular to the magic line to avoid the influence of me sagging the floor, which was translated to the countertop where the adjustable stand was. My steel frame, 2" thick wooden top bench (which is sturdy and seemingly rigid to the touch) was just as bad.

Thank you for your posting, and describing your trials with levels.
I do get it that most stuff shifts and sways around is, and not much really stays put.
For me, getting levels right has had to be taken to extremes in my job before I retired. Sure, when building my place, I have used a home-made Egyptian water level to get within 3mm over 40m (that's way better than the surveyor did) , and I have one of those 3-axis green laser levels for construction stuff, but ultimate is using digital clinometers for satellite tracking servos. Machinists levels can be super-accurate, and frustratingly slow, but we are using them for very different types of measurement.

Your level adjuster looks to be one like where you use a C-spanner hooked into those holes in the adjuster nuts, similar to those used for calibrating micrometers.

 

[WobblyHand]

Your level adjuster looks to be one like where you use a C-spanner hooked into those holes in the adjuster nuts, similar to those used for calibrating micrometers.

A micrometer C-spanner would be preferable to the tool that was supplied. The one I have has the pin on the handle side of the C rather than at the far end of the C. The preferred orientation for turning is opposite to that of an ordinary micrometer c-spanner. Tripped me up several times.

The experience base of HM members is really quite astounding, it spans lots of disciplines and talents. It's great to listen in and glean tidbits of knowledge that were hard earned over a lifetime. It's rare not to learn something interesting here.