Do you think if I mount the GNSS Multi-Band L1/L2 Surveying Antenna - TNC (TOP106) to a mast attached to the bulldozer blade, that the vibrations could destroy the antenna?
If so, can you guys recommend a Surveying Antenna suitable for bulldozer blade mounting I can use with the SparkFun RTK kits?
I suppose I won’t know until I try it, and I could just buy some extra antennas since they are so cheap anyway. I would imagine the “ruggedized” ones are much more expensive.
As a matter of curiosity why not mount it to the top of the cabin?
Because @jeffersonkim is referring to what the construction industry calls “Machine Grade Control”.
The purpose is to be able to know where the dozer blade is in reference to the Proposed Grading Plan.
The Operator will know how much cut/fill any area requires based on the location of the blade’s cutting surface (vertical is the most important).
Machine (Grade) Control has experienced a huge amount of growth in the past decade.
I believe it followed the Agricultural Industry that got it’s start in AutoSteer.
The Open Source community is supported by AgOpenGPS and projects like OpenGrade3D.
@jeffersonkim , I wouldn’t be afraid of using TOP106 antennas for a rough grading application. It’s a good design and a great value.
For Finished Grading, the construction industry typically adds Total Station/Laser tech to reach a final bluetop accuracy of ~1mm, when required by the project.
Vertical lines are spaced 7.5’ apart along a 52’ span between the red lines to maximize the 9’ blade. I edited the KML files manually referencing an online calculator and excel spreadsheet for correct spacing.
Each of the horizontal lines are spaced 50’ apart to match engineered drawings that specify altitudes along the center of the finished road.
The theory is that combined with Caterpillar’s AutoAssist features built into their D3 dozer (can set incline/decline & cross slope of blade relative to the vehicle), along with 1/2" altitude accuracy of the GPS blade, this should be sufficient for this project.
It’s just a 1/2 mile straight road.
This is a pretty fascinating topic. Thanks for the in depth explanation.
In general GPS antennas are not very fragile. But for very precise vertical positioning a second for.m of sensor is used in general.
I’m pretty sure the OP @jeffersonkim already knows this, but it’s an interesting topic and I’d like to mention the next step for others in the future.
The user will usually perform a Localization or Site Calibration. The Engineered drawings are typically in a local system, NAD83 State Plane Coordinates for example. On such a small site, the Grid Coordinates wont pose much of an issue, but the elevations will. You basically need to identify the Elevation difference from your GNSS observed points verses the existing elevations on the plans. Depending on the Vertical Datums used, this could be a difference of several feet. Once you localize to the site, it wont matter, as you will apply that difference to all your proposed points (you shift your proposed surface to match the local system, even if it’s an arbitrary system).
I’ll admit that I’ve oversimplified this, but localization allows you to Stake-Out points straight from your GNSS.
Thank you. I’m new to this, so I did not know. Will using an NTRIP provider or Spark Fun RTK Reference Station help improve the vertical accuracy?
It appears on SW Maps, that I’m getting about 8" vertical accuracy with the SparkFun RTK Express out of the box simply in Rover mode without any additional correction data.
(I went ahead and ordered a RTK Reference Station & will set it up at the construction site. The antenna will be within 50 feet height of the road altitude, and within 1/2 a mile.)
Be sure to check out the RELPOSNED message on the ZED-F9P. RELative POSition North East Down, gives you the relative position of the Rover compared to its Base. In a “Moving Base” scenario, this can provide excellent relative positioning, but not absolute. For absolute accuracy, the moving base needs its own corrections, from a local fixed base.
Another user was trying to do (almost) the same thing here. Their application was drone orientation, compared to a moving vehicle. But the principles are the same.
Hope this helps,
Paul
Operating your own Base at the project site will provide the best results, by far.
The Rule of Thumb that’s always quoted for RTK is 1cm + 1ppm.
That’s saying, that under the absolute best test conditions (and very expensive antennas), the RTK relative accuracy is 1cm + 1 cm for every 10km (6 miles) of baseline (distance between base and rover).
Generally speaking, I wouldn’t put any confidence in the GPS Uncertainty (Perceived Accuracy) values reported by any GNSS device. The device would need to know it’s actual location to be able to provide any meaningful data.
There are several ways to accomplish what you want to do.
I think the easiest (if you don’t have Surveying Software and Data Collector for coordinate transformation on-the-fly ) would be to setup your semi-permanent base for the project and get it’s position sorted out, just like normal. Then locate/store the positions of various site features on the construction plans with your Rover, especially any control points. In your MS Excel Spreadsheet, you will define the N,E,Z offsets for each pair of positions for your Site Calibration (Published - Located) and establish a best-fit offset. Then have MS Excel create a new list of proposed stakeout points that have been adjusted using the “best fit” offset value for N,E,Z across the entire project. Now your points have been Localized to the site so your SparkFun RTK Rover positions can be used easily and directly.
It’s basically similar to opening the design in AutoCAD and moving the project site to ITRF, which is the reference frame of your RTK combo.
Great ideas! Thanks.
I just wanted to provide an update.
I just set up rtk2go as the caster and now sending out coordinates.
Since the final road height will need to match the public road, I can actually use the public road as my “monument” and then reference the rest of the road to that since I already have markings for finish grade.
We also mounted, using a piece of wood as an “adapter” of the antennas to the dozer blade. Excited to see this work!
I love it !
You might be the first person to bring SparkFun RTK into the Machine Grade Control Industry?
As you already know, Machine Control is generally terribly expensive.
I would love to see a picture of the road profile when you get it graded 
Did you go with a single mast at the center of the blade?
Yes. Single mast in the middle.
Well, it’s my own road, so I’m not trying to impress anyone. But if I were a contractor, it may look a little sketch to clients, especially the larger entities that have deadlines.
Likely you’d see more contractors in my area using it if they had reliable tech support on-site. If you added in the cost of hiring a tech-hand to set everything up, I’d imagine paying $100k for a Trimble doesn’t seem so unreasonable, as I believe the Trimble/Cat partnership would provide ongoing support.
The wider appeal use case likely is with the farmers in my area. But I won’t be able to do that until I actually work in the field and work out the kinks.
Also to clarify, it’s not machine controlled. I’m just going to have SW Maps open in the Dozer so we can see the altitude of the blade. Hopefully the built-in latest gen auto-assists will be enough.
I’ll keep this post updated as I work out the kinks, and maybe one day do a YouTube video.
During a shower, I was contemplating the annoyance of having to adjust my lines to accommodate the fact that the antenna is only in the center. Then I realized that there are locations for 3 masts on the blade.
This means that if I attach a $250 or so mast onto each of the three locations, (left, right, center), then I can simply reattach the antenna to each of the masts depending on which side of the road I need tracking.
I’m using this clamp along with the standard 5/8" bolt from Home Depot for a really cheap “mounting” option that makes it easy to switch between the masts as needed.
It’s still to be seen how these handles hold up with the movements of the blade, and I may need ot look at some stronger clamping actions.
The other benefit of having an adjustable clamp is that I can have the antenna really sticking out toward the edge of the blade and at various angles depending on the task.
Also, I discovered that there are $100 5G Tablets that I just ordered with T-Mobile $10/month data-only cards. To be mounted in the dozer with this tablet holder. https://www.amazon.com/dp/B0936P1MNB
The one annoying thing we discovered is that when you lose data on the phone from switching to Wi-Fi to 5G, it also loses NTRIP connection, and you have to reconnect.
Another idea is that if I really want to permanently attach the antennas to the masts, I can have 3 different antennas on each of the masts, and then use something like this antenna switcher for the operator to switch between different antennas.
ANT 1 = Left Mast
ANT 2 = Center Mast
ANT 3 = Right Mast
I will post separately for this idea.
Some issues faced:
Just how hostile some Dozer operators toward being micromanaged to such precision
The NTRIP connection drops every once in a while requiring the operator to manually reconnect
Many of the operators hardly know how to use a cellphone, let alone connect a bluetooth device on an Android.
Most contractors are technologically challenged and you need constant monitoring for the system to work. Things will just randomly break and you have to painstakingly go through the steps to troubleshoot. Meanwhile, your Operator is getting angry because they just want to get the work done. They will need someone on hand at all times to troubleshoot.
If you’ve seen an Operator throw a phone because of how frustrated they get, you can get a sense of just how frustrated they will be with the survey equipment attached to their machine.
Even bluetooth headsets can be finicky. Bluetooth is finicky.
You will likely need to monitor your operator to ensure they’re driving on the designated lines. This may mean installing remote software on the tablet in the cab.
Operators are not used to such level of scrutiny.
I’m still pushing through the problems hiring forward thinking Operators. And it’s still cheaper than hiring a separate surveyor just to monitor the blade height, or setting up an automated laser system which you’ll need to measure anyway with a survey stick.
The biggest challenge is likely finding an Operator willing to work through the prototype phase. It’s just not as integrated and reliable as the $100k trimble systems (I’m assuming). And the operators may not be as hostile if they think it’s integrated into the system, rather than a separate android tablet hanging in the cab.
I just wanted to say thank you for writing this up! I’m enthralled with my own naivety and the obviousness of what the operators are up against - and I’m taking notes! This is all really good info for all our PNT products.
Ok, perhaps a breakthrough in Equipment centered around a laser dual-slope system. I’ll go through the items and then provide a description of use.
Milwaukee just came out with their own dual-slope laser system that at 100 feet is 1/16" accurate. This seems on par with the competitors that have been in the business a while like the GL622N. M18 probably can’t do 25% slopes though, and I’m not sure if it can do decimals. Seems much more robust in terms of drop proof. I’ll let someone else do a more detailed side-by-side.
https://www.acmetools.com/johnson-level-remote-display-40-6792/049448067923.html
If you understand the items, you can probably understand where this is going. The engineered road plans have centerline elevations every 50 feet. That’s where the SparkFun RTK shines. I can set the blade every 50 feet, and then from there calibrate with the dual-slope laser for the space in between.
Here are the general steps:
Confirm elevation of end of pavement of public road connecting to along centerline with the SparkFun Rover.
Adjust all road elevations in your plan to that reference point.
Make sure you are using a secondary GPS device to double check the measurements of the Dozer antennas. They jiggle and move all the time. You could also create a monument (concrete form tube in ground) on-site near the end of pavement. I’m pretty paranoid, so I’m probably just going to check the end-of-pavement or monument every morning to make sure it the elevations don’t start moving around in the middle of the project.
Place center mast of the dozer antenna to be on one of the known elevation points of the road on the centerline.
Setup laser nearby with appropriate dual-slopes needed for the 50 foot section
Set AutoAssist in Cat dozer to correct settings
Setup red laser receivers to pickup the laser
Proceed forward along pre-planned travel lines where Operator will focus on keeping the blade along the laser until they reach the next known elevation point in 50 feet.
Verify dozer blade is at correct elevation according to plans and Survey.
Adjust as needed
Basically, you have a triple system check to verify you have the correct slopes:
As a 4th control, you can have someone with a separate RTK Express on monopod spot checking to ensure the blade antenna didn’t get all weird.
I got the parts in order, so hopefully by next week, I’ll be able to provide an update on how it works.
The laser system is absolutely necessary between the 50 foot benchmarks.
The auto-assist in the dozers help with cross slope, but can’t be relied on between the 50 foot benchmarks.
Today, since we had about 4 feet of soil to remove, we just calibrated the bottom of the dozer blade with our other GPS stick.
Essentially, the dozer can just set his blade on the ground, and knows the exact elevation of the blade and references that with the plans on the tablet for minimal height.
I am adding a track loader with an auto laser tracking blade. That’ll likely arrive next week.
