Surely at this level of precision using fixed height Invar staves (no joint to wear etc.) and field methodology are going to make as much difference as which brand.
Have to disagree about optics, the Wild/Leica glass I've used has always been better than Sokkia. Not that Sokkia is poor quality at all.
Unless you are trying to shoot through small gaps in machinery like Chris notes about above not sure why you would be choosing a optical level over a digital one?? Digital is faster and you can't transpose readings. Leica LS digital levels with Invar staves have same precision spec as optical ones you noted.
What staves do you intend to pair with the level? What turning points, instrument shading etc. Would need to know the whole package you intend to use to really comment.
@lukenz We are going to use Leica fibre glass staves the project includes a laser scanning control network in a factory and then use the existing network installing machines.
We have an upcpming project to establish a network in a metallic bridge.
We want the optical because in some places may we have vibrations.
Also I am opened to alternatives either in level or in instruments, cost is not the problem.
We want the optical because in some places may we have vibrations.
If vibrations are the concern, then it seems to me that the instrument question isn't optical versus digital, it's automatic leveling versus manual leveling. The NAK2 is an autolevel, so severe vibrations will make its compensator bounce around.
@jim-frame thanks for the answer. Yes I was meaning manual levelling you are correct. Maybe the na2? Other alternatives we have in manual levelling instruments?
For your application you will need to check the specs carefully as current Leica NA2 and LS digital ones have compensators I understand so unsuitable for your purpose as compensator will vibrate on bridge/in industrial site as @jim-frame has already pointed out to you.
The "A" in NA2/NAK2 I assume standards for automatic (i.e. has automatic compensation)
You need to talk to a good dealership who can supply/support for you!
Not sure the Leica/Wild N3 is still in production (looks like a few around online, not sure about parts though). Looks to be a tilting level with split bubble not a compensator.
Sounds like a job we did last year for a client, it was a big job for scanning, about 1.2 km long (0.72 miles), they had a scanning standard that required 2mm accuracy and nobody was willing to have the standard revised. Since no confidence level was quoted we nominated, 1 standard deviation (68% confidence) and we quoted "as accurate as practically achievable", with a caveat that all long shots needed to be performed during overcast weather. We used an SX12 doing heaps of rounds for network strength and we used long occupation GPS running off solar, we ran the observations through AUSPOS (Australia's equivalent of OPUS), we used a choke ring antenna and two antenna's with ground planes. We used the GPS vectors and covariance matrices from AUSPOS and ran them through least squares, along with Total Station, Scanner and Level observations. The other scanners we used were a Topcon GLS (prism target observations in least squares), a Leica RTC360 (trimmed at 35m radius) and a NavVis VLX (operational plant, terrestrials couldn't handle the vibration).
After about 18 weeks of GPS observations processed through AUSPOS, least squares was reporting 1mm @ 95% confidence in 3D, most of our network was within 3mm and some 4mm with a few outliers above that (retroreflective sticker targets). Although there was no absolute positioning requirement in the spec, we figured it would assist tightening our control network, which it did.
Scanners have the same problems that total station direct readings have, low angles of incidence of the laser beam isn't going to meet spec. The SX12 was the closest instrument to spec but would have been impractically slow (5 arcsec angular accuracy), the GLS range accuracy isn't quite as good as the SX12 and it has 6 arcseconds angular scan accuracy. The RTC360 has 18 arcsecond angular accuracy. Where the SX12 and GLS have a big advantage is with positioning accuracy. B+W checkerboard target centring accuracy is 2.5mm standard deviation, so it was going to be out of spec. Interestingly the NavVIS scans when compared to Topcon scans were well aligned for large surfaces, the NavVIS had some issues capturing details, but it was the best we could do in a running plant.
After the completion of the project, someone tracked down the original author of the scanning standard, it turned out that 2mm was based on the range accuracy of a single scan off a scanners spec sheet.
@pfirmst Thanks for the answers I have never thought about using choke antennas in the project! It’s a bit time consuming but I read it well you wait 18 weeks almost five month for observations?
We'd visit the GPS stations each morning to check level and over point. AUSPOS can process a week of observations, so we'd grab that data once a week, process it, then add it to our ongoing least squares analysis. OPUS processes up to two days of observations, so it would require more regular processing, but then I think it has better least squares software support, so maybe less data entry. We picked the reference stations with atomic clocks for our GPS vectors.
Obviously, you need vectors which are in a fixed time reference frame, to compensate for continental drift. The network was very tight, any control point you pointed the total station at would typically read within 1mm.
These SMR prisms have excellent centring accuracy, we also found Topcon's GLS would get vertical angle errors on 62mm diameter prisms within 135m. Nothing beats prisms for accurate observations and centring accuracy. The problem with using point clouds for registration, when high accuracy is required, is surface properties, like colour and roughness will introduce error and these will propagate. Using control scans, captured from least squares analysed positions prevents errors accumulating in point cloud registration.
We found when we imported the SX12 scans (e57) into Leica Cyclone (control scans for the RTC360), that Cyclone wouldn't preserve the original position of the scan correctly. Cyclone had no problems importing the PTX scans from Topcon, so we used them instead.
Turns out the SX12 and GLS work really well together, the SX12 for window scans and lots of rounds, while the GLS gets heaps of coverage using road mode full dome scanning.
I'm currently working on mixed pixel filters for the GLS, it captures points behind guarding, which is very useful, but standard mixed pixel filters remove them. The instrument will produce mixed pixels between surfaces closer than 350 to 400mm.
The SMR prisms have 0.1mm centring accuracy, the precision tribrach carrier adds another 0.1mm of radial runout (I checked with a dial gauge). The SMR prism extension has a datum flange, for consistent height above the carrier, height is set at the standard 196mm above the tribrach.
We'd visit the GPS stations each morning to check level and over point. AUSPOS can process a week of observations, so we'd grab that data once a week, process it, then add it to our ongoing least squares analysis. OPUS processes up to two days of observations, so it would require more regular processing, but then I think it has better least squares software support, so maybe less data entry. We picked the reference stations with atomic clocks for our GPS vectors.
Obviously, you need vectors which are in a fixed time reference frame, to compensate for continental drift. The network was very tight, any control point you pointed the total station at would typically read within 1mm.
These SMR prisms have excellent centring accuracy, we also found Topcon's GLS would get vertical angle errors on 62mm diameter prisms within 135m. Nothing beats prisms for accurate observations and centring accuracy. The problem with using point clouds for registration, when high accuracy is required, is surface properties, like colour and roughness will introduce error and these will propagate. Using control scans, captured from least squares analysed positions prevents errors accumulating in point cloud registration.
We found when we imported the SX12 scans (e57) into Leica Cyclone (control scans for the RTC360), that Cyclone wouldn't preserve the original position of the scan correctly. Cyclone had no problems importing the PTX scans from Topcon, so we used them instead.
Turns out the SX12 and GLS work really well together, the SX12 for window scans and lots of rounds, while the GLS gets heaps of coverage using road mode full dome scanning.
I'm currently working on mixed pixel filters for the GLS, it captures points behind guarding, which is very useful, but standard mixed pixel filters remove them. The instrument will produce mixed pixels between surfaces closer than 350 to 400mm.
@johnymal That's right, we were on site for 5 months, it was a big job. Two to three weeks will still get you good accuracy, but we're talking 4 to 5mm @ 95% confidence in 3D. GPS is helpful as it minimises distortions in the control network.
@pfirmst I would like too see the atomic clock as from my search I only find expensive ones for scientific use.
If you like post some more fotos
smr prisms is a good solution already have one but I want to order two more.
Which program did you use for least square adjustment? Starnet?
I have surveyor background as I do this for the 20 years now.
I will do the laser scanning with a Leica P40.
Its the first time that I am trying to establish a so precise 3D network.
we have an upcoming project for scanning bridges combine it with photogrammetry. Also we have to scan a big industrial space all the above has horizontal and vertical accuracy 2mm. That’s drive me crazy but is challenge for me also.
@pfirmst I would like too see the atomic clock as from my search I only find expensive ones for scientific use.
We choose CORS reference stations with atomic clocks, these are also fiducial coordinate sites, fixed in our LSA. Our receivers don't have atomic clocks.
