When we last left Chapter 4, Kent said:
>It's worth underscoring that you now know more about the angular accuracy of your instrument than its manufacturer probably does.
>Keep in mind that while instrument centering was set to 0 for the purposes of this test, in ordinary work there would be a centering error to be assigned. From your test, a standard error of 0.001 ft. for instrument centering would be realistic.
If one (as in, me) were to run this same test with a different total station, would it be necessary to include the instrument centering targets, given that my technique for centering with the new instrument would likely be the same?
> When we last left Chapter 4, Kent said:
>
> >It's worth underscoring that you now know more about the angular accuracy of your instrument than its manufacturer probably does.
>
> >Keep in mind that while instrument centering was set to 0 for the purposes of this test, in ordinary work there would be a centering error to be assigned. From your test, a standard error of 0.001 ft. for instrument centering would be realistic.
>
>
> If one (as in, me) were to run this same test with a different total station, would it be necessary to include the instrument centering targets, given that my technique for centering with the new instrument would likely be the same?
If the targets are going to be close range targets as you used in a prior test, you pretty much do have to evaluate the instrument centering errors as was done.
If you use targets 150 ft. or more distant from the instrument, I think it would be reasonable to assume an instrument centering standard error of +/-0.25mm as long as you've verified that the optical plummet is in good adjustment.
The same program of rotating the instrument between sets and recentering applies, of course.
Kent,
If I recall correctly in a thread like this one you made a comment along the lines that someone's total station encoder sensors only read one part of the circle and hence the need to plunge the telescope and read another face to correct for eccentricity of the circle. Apologies if this is incorrect.
The Leica total station that I use has a diametrical encoder reading system where apparently opposite sides of the circle are sampled at once and an average derived from here. I this were the case would you accept that multiple re-pointings would be acceptable procedure in a well adjusted total station rather than plunging?
Apologies to the OP for the hijack.
> If I recall correctly in a thread like this one you made a comment along the lines that someone's total station encoder sensors only read one part of the circle and hence the need to plunge the telescope and read another face to correct for eccentricity of the circle. Apologies if this is incorrect.
>
> The Leica total station that I use has a diametrical encoder reading system where apparently opposite sides of the circle are sampled at once and an average derived from here. I this were the case would you accept that multiple re-pointings would be acceptable procedure in a well adjusted total station rather than plunging?
Basically, what you're asking is how to evaluate the standard error of a direction taken with a total station that has two or more circle scanning points and that has dual axis compensation?
Taking directions on both faces cancels both horizontal collimation and tilting axis errors as well as canceling errors from circle eccentricity. For control measurements, I'd think it would pretty much always be best practice to take directions on both faces of the instrument. If a surveyor wanted to estimate the standard error of a direction on just one face, I'd think that the problem would be broken down into just the uncertainties of a direction taken at a zenith angle near 90° and the uncertainties resulting from the tilting axis correction.
You could determine the s.e. of a direction measured on just one face of the instrument by a test procedure similar to that I've previously outlined, i.e. four sets of directions, Face Lt and Face Rt, to five targets, but after each set, before rotating the instrument and circle for the next set, taking a set of just Face Lt directions to compute the residuals in relation to the means of Face Lt and Face Rt.
The analysis of variance would need to take into account the variance of the means of Face Lt and Face Rt, but aside from the random errors in the means those values would be acceptable as unbiased estimates.
Shooting both faces doesn't cancel tilt errors.
> Shooting both faces doesn't cancel tilt errors.
It certainly does cancel errors of the tilting axis, aka vertical collimation errors resulting from the tilting axis deviating from perpendicular with the turn axis.
I think you meant levelment errors, didn't you? That's something quite different.
Yeah
Well I run the instrument through it's 'check and adjust' routine as often as suggested by the manufacturer. Through repeated FL/FR observation of roughly horizontal and also steep targets the instrument calculates values for Hz and V collimation, tilting axis error, and level compensator errors. I do this until the reported standard deviation of the calculated corrections is no more than 1"
My personal procedure for control surveys is to do FL/FR but it is probably just out of habit, as the instrument should be a perfectly accurate 1-face instrument with an accurate set of calibration values on-board. Agree/disagree?
> My personal procedure for control surveys is to do FL/FR but it is probably just out of habit, as the instrument should be a perfectly accurate 1-face instrument with an accurate set of calibration values on-board. Agree/disagree?
In my opinion, best practice still would be to take F Lt and F Rt directions for important angles where the uncertainty needs to be well characterized.
I'd be interested in knowing what sort of transient collimation effects are present in an instrument set up in sunlight that the software corrections don't capture. Naturally, one doesn't have to worry about that in the means of F Lt and F Rt.
I cut my teeth on an assortment of theodolites (and even a couple of transits) that got shuffled around from crew to crew, so I got in the habit of turning direct/inverted for anything important early on. These days I still do that, not so much to improve accuracy, but rather to demonstrate to myself that nothing has gone out of whack that would introduce an unacceptably large error in a one-face-only measurement. It's a QC procedure that tests key instrument adjustment values at every setup throughout the field day, since I almost always begin each setup by checking into one or more existing control points and/or setting one or more new ones.
Call me pedantic, but, my best practice is to arrive at a station, set up the instrument and tripod and set the total station to rotate continuously for 10-15 minutes (it's robotic) while I do something else. This is in order to 'bake' the entire instrument evenly in the sun and have the legs expand and settle at their heated position. I then do a final level and centre before I do the critical measurements all at once. I have seen what I think were FL/FR differences caused by uneven heating of the instrument so this is my attempt to eliminate that.
> Call me pedantic, but, my best practice is to arrive at a station, set up the instrument and tripod and set the total station to rotate continuously for 10-15 minutes (it's robotic) while I do something else. This is in order to 'bake' the entire instrument evenly in the sun and have the legs expand and settle at their heated position. I then do a final level and centre before I do the critical measurements all at once. I have seen what I think were FL/FR differences caused by uneven heating of the instrument so this is my attempt to eliminate that.
That's certainly a much better practice than just setting up the instrument and sallying forth into the world of measurement.
What would concern me (after the instrument is at ambient air temperature) would be differential heating of the instrument in bright sunlight with one side of the instrument shaded. In my experience, changing face tends to "barbecue" both sides more or less equally and to minimize the collimation drift from differential heating. It may well be that robotic instruments are heavier and not as susceptible to these effects as lighter instruments seem to be.
While we're on the subject of "Check and adjust"
> Well I run the instrument through it's 'check and adjust' routine as often as suggested by the manufacturer.
In the back of the manual for my instrument, it lays out routines to check and adjust the following:
Instrument Constant
Optical Axis
Circular Level
Vertical Cross Hair
Collimation
Optical Plummet
Vertical Angle datum
Compensation of Systematic Error of Instrument.
The last of these seems to adjust out the FL/FR errors. The question is: Does doing this adjustment periodically, eliminate the need to take FL and FR measurements?
The second question is: Should one even dink with all the others or even learn how to do them? Isn't that what dealers do for $200 or so?
it seems that 0.001 is VERY optimistic. 20+ years ago we, while working at a Federal Laboratory, we tested our field technicians on instrument centering. We found that 2 mm was more realistic.
> it seems that 0.001 is VERY optimistic. 20+ years ago we, while working at a Federal Laboratory, we tested our field technicians on instrument centering. We found that 2 mm was more realistic.
With rotatable optical plummets and distinct ground marks, at normal tripod height, a standard error of centering of +/-0.25mm (which is nominally +/-0.001 ft.) is not unrealistic at all. I've tested my total station in the past and have gotten that result and various other posters did the same. All values clustered in that neighborhood. The test is not difficult at all.
Now, a fixed optical plummet that can't be rotated is another story.
