Just a quick thought M.
I can't come even remotely close to those tolerances on measurements - for 2 reasons.
-  1, no dial indicator
- and 2, even if I had one ( yes I know they can be quite inexpensive ) my single top plate rocks some - quite a lot actually - when you're talking discrepancies of less than a third of a millimeter. I could probably apply slight pressure to any side and have it dip by 2mm without any real vertical motion. You've got a lever arm amplifying those minute measurements too. Any possible way to mount the dial so it's reading directly off the top of that weight?

The variations I mentioned earlier were more in the water levels. Start with measured differentials in two layers - allow it to rise a little / sink a little. Then try to get all 4 levels back to the exact same spots. Couldn't quite seem to get back there.

Maybe more tomorrow
Dale

@Mondrasek: Your very precise results are interesting. To get an idea of the hysteresis, if any, we really need a series of data points from an identical set of test trials. Do you have the patience and time to do 20 runs, with a short interval between runs, say a minute, for relaxation and settling? Make a blank data table, try to use the exact same procedure each run, fill in your data table objectively without thinking or analysis, just the facts ma'am, then when you have measurements from a set of 20 runs, I'll run, or you can run, basic statistics on them, and then we will be able to see if what you are seeing is genuine hysteresis or is noise in the data. In other words, we will determine means and standard deviations of the data points, which will enable us to evaluate the _actual_ degree of precision you are attaining with your eyeballs and dial indicator (without a dial....) and tell us what the level of noise is in your measurements and apparatus. If the hysteresis is larger than the standard error, then it's probably real. But since its value seems small and you are measuring to such high precision.... we need to tease out the accuracy level from all that precise data and see just how _accurate_ it all is. Precision and accuracy are very different, but related, critters. So what if you can put five rounds into a circle the size of a quarter at 100 yards, if that circle is down on the corner of the target rather than in the center bullseye that you were aiming at. Precision, and accuracy. For best results you need both.

Quote from: TinselKoala on October 13, 2012, 01:23:04 AM
@Mondrasek: Your very precise results are interesting. To get an idea of the hysteresis, if any, we really need a series of data points from an identical set of test trials. Do you have the patience and time to do 20 runs, with a short interval between runs, say a minute, for relaxation and settling?

I see no problem with my patience for completing that test run.  But I have some concerns with relieving the load weight on the ZED system.  I can add weight repeatably.  But "lifting" by a repeatable (negative) value is a challenge.  I guess I can try to pick up a pulley so I can set up a simple counterweight system. 

Maybe these tests do not require the use of a constant mass value for the added and subtracted weight between readings?  But, then again, it's the only way to be sure.

M.

Hysteresis test data.

@Mondrasek: thanks for doing that. I'll run the numbers shortly, but meanwhile can you give some description that correlates the parameters in your column titles with your verbal description in the earlier post? I don't get these "max" and "min" terms.....