Hello People,
Check out this graph:
I have collected a lot of observations in order to try and map the angular 'errors' present in our total stations. There is enough there, and at sub-second quality to see very small effects when processed and averaged. This one I can only guess is tripod or tribrach wind-up.
What you are looking at is a the average of 10 or so sets of observations taken to 6 targets in a forward and back direction. The order is t1, t1, t2, t2, t3, t3, t4, t4, t5, t5, t6, t6, t6, t6, t5, t5, t4, t4, t3, t3, t2, t2, t1, t1. The Y-axis is in decimal degrees.
The first sight to each target is assigned zero status and the differences are with respect to the first sighting. I decided to check the numbers because I seemed to get good angular comparisons between the first and last observations (on target 1), but the spreads on targets 3 and 4 seemed to always be larger than the spreads on the other targets. The graph bears this out. Interesting is the dip in the first re-pointing of target 1. I put this down to me always turning the instrument anti-clockwise to line up target 1 to begin the set quickly by hand. I suspect things bounce back somewhat quickly between the initial target 1 sighting and the next re-pointing after this large torque applied by gripping the instrument by the standards and twisting.
The instrument is motored between targets on a low speed, and I would have thought the torque required to do this would be absolutely a minimum. The tripod is an iso-heavy, brand new Leica GST20. There is no slack in it that I can find. The tribrach is a professional series tribrach rated at less than 1" hysteresis.
I am quite pleased at the smooth and logical shape of the graph considering the minute scale of it. The angles seem to be affected by an average maximum of around 0.8" to the middle targets. I suspect the effect would be larger when turning sets robotically at full speed, or when manually turning the instrument by hand between the targets as this instrument will want to drag the tribrach and tripod around with it due to the friction braking.
Comments welcomed as always.
>tribrach rated at less than 1" hysteresis.
And if I understand your Y axis, that is about what you demonstrated.
It seems pretty clear that you have demonstrated hysteresis, which you call wind-up (I think in control systems work wind-up is a different effect). I'm not sure how you checked for slack.
You could try this: point to one target, gently twist the tripod clockwise by hand, release, and take a reading. Then repeat counterclockwise. This should give a ballpark estimate of tripod hysteresis.
Similarly, repeat by grasping the instrument body and twisting gently each way so it twists the tribrach. I found that my (much sloppier) measurements improved when I tightened the friction adjustments on my tribrach screws.
Finally, put a finger on each end of the telescope and very gently push each way. There is a possibility that there is slack in the instrument itself. About 20 micro-inches (a half micro-meter) of play in the bearings, could cause the encoding disk to move this much relative to the sensors, or the telescope to move sideways with respect to the instrument body.
>
> What you are looking at is a the average of 10 or so sets of observations taken to 6 targets in a forward and back direction. The order is t1, t1, t2, t2, t3, t3, t4, t4, t5, t5, t6, t6, t6, t6, t5, t5, t4, t4, t3, t3, t2, t2, t1, t1. The Y-axis is in decimal degrees.
For some reason, I'm not making sense of the graph. What do the numbers 1-12 represent?
I took it as numbering his measurements
t1, t1, t2, t2, etc with plot numbers
1, 2, 3, 4, ... 11, 12, 12, 11, .... 1
> I took it as numbering his measurements
> t1, t1, t2, t2, etc with plot numbers
> 1, 2, 3, 4, ... 11, 12, 12, 11, .... 1
from Conrad's original post above:
> What you are looking at is a the average of 10 or so sets of observations taken to 6 targets in a forward and back direction. The order is t1, t1, t2, t2, t3, t3, t4, t4, t5, t5, t6, t6, t6, t6, t5, t5, t4, t4, t3, t3, t2, t2, t1, t1. The Y-axis is in decimal degrees.
>
> The first sight to each target is assigned zero status and the differences are with respect to the first sighting. I decided to check the numbers because I seemed to get good angular comparisons between the first and last observations (on target 1), but the spreads on targets 3 and 4 seemed to always be larger than the spreads on the other targets. The graph bears this out. Interesting is the dip in the first re-pointing of target 1. I put this down to me always turning the instrument anti-clockwise to line up target 1 to begin the set quickly by hand. I suspect things bounce back somewhat quickly between the initial target 1 sighting and the next re-pointing after this large torque applied by gripping the instrument by the standards and twisting.
So, as you understand the program, is it that:
- two pointings are made on each of six targets and the instrument is turned CW from target to target, with the angular magnitude of the second pointing on the lower limb of the graph representing the difference between it and the first
- then two more pointings are made on the six targets in reverse order with the instrument turning CCW from target to target, but the angular magitude of the upper limb representing the differences between them and the very first direction to the target (on the lower limb)?
If the problem is the tripod rotating, you'd think that just always turning the TS CW between pointings would answer the question. I'd be interested to know whether the instrument was in direct sunlight.
Hello Bill,
> It seems pretty clear that you have demonstrated hysteresis, which you call wind-up (I think in control systems work wind-up is a different effect). I'm not sure how you checked for slack.
When using this gear before I have physically twisted the tripod by hand with quite some force in order to make it find another resting position if slack was present. I've done the same with the tribrach by physically twisting the instrument and tribrach by hand to try and find any slack. Also made sure all screws were tight on the tripod. That's all I know how to do.
> You could try this: point to one target, gently twist the tripod clockwise by hand, release, and take a reading. Then repeat counterclockwise. This should give a ballpark estimate of tripod hysteresis.
Done this already but didn't take readings. The torque involved by motoring the instrument on a low speed setting should be a long way short of anything I can do by hand.
> Finally, put a finger on each end of the telescope and very gently push each way. There is a possibility that there is slack in the instrument itself. About 20 micro-inches (a half micro-meter) of play in the bearings, could cause the encoding disk to move this much relative to the sensors, or the telescope to move sideways with respect to the instrument body.
Will try this. Thanks Bill.
Hello Kent,
> So, as you understand the program, is it that:
>
> - two pointings are made on each of six targets and the instrument is turned CW from target to target, with the angular magnitude of the second pointing on the lower limb of the graph representing the difference between it and the first
>
> - then two more pointings are made on the six targets in reverse order with the instrument turning CCW from target to target, but the angular magitude of the upper limb representing the differences between them and the very first direction to the target (on the lower limb)?
Bill has it right. The number represent the order of observation, and graphing them this way also gives a visual idea of the CW then CCW order of observations forward and back against the difference from the first pointing.
So obs 1 is the first sight to target 1 and obs 2 is the re-pointing. Obs 11 is the first pointing to target 6 (so it it's sitting at zero y value) and obs 12, 13 and 14 are the next 3 pointings to target 6. No motoring is done at this end, just 3 re-pointings so I can't explain the slightly increasing reading over the 4 obs. Perhaps there are also bearing effects in there too.
> If the problem is the tripod rotating, you'd think that just always turning the TS CW between pointings would answer the question. I'd be interested to know whether the instrument was in direct sunlight.
No sunlight. It's in my garage and typically the instrument is left on the tripod for several hours before the first obs is made.
> No sunlight. It's in my garage and typically the instrument is left on the tripod for several hours before the first obs is made.
I think I'd be interested to know whether the instrument just repointing on the same target (as was done for pointings 11, 12, 13, and 14, I take it) over a period of time exhibits any similar pattern.
