Wandering Fix

I’ve been busy setting up and testing my M2 system. I noticed something very strange yesterday. Even though my Rover has a very solid fix that never leaves lock status, and even though both units are solidly mounted and do not move, the status location of the rover seems to wander around. The base coordinates NEVER change because they were measured and plugged in manually.

I set the number of points to 1000 and then let the system run all afternoon just sitting there. By the time the dust settled, the error spread had grown to about 15 cm or 6 inches. That’s a VERY LONG WAYS FROM the 1cm I was expecting.

Frankly, I thought that was quite unreasonable given how close together the units are and how solidly placed they are. It’s certainly nowhere near the expected 1cm accuracy I expected from an RTK system. Then I wondered… Is it possible that this is a map of the uncorrected readings for the afternoon? If so, that performance would be AWESOME!

That thought got me to wondering some more. I’ve noticed that the web portal screen readings for the rover wander around quite a bit even when a solid RTK fix is achieved. Is it possible that these status readings are also uncorrected?

If the status screen location on the rover is uncorrected, then my base calibration is wrong too. I used the readings on the status screen to find the error for the monument location and then used that error to correct my base location. But now I’m thinking that the status screen might be displaying the uncorrected location, not the corrected location. So my base location could be wrong too! Not a big deal for precision, but not Great for accuracy.

So, which is it? Is my 1cm precision all messed up, or is the status screen displaying the uncorrected coordinates?

If the accuracy is messed up, how do I fix it?

If the status screen location is uncorrected, how do I get corrected coordinates?

Assuming my suspicions are valid and the coordinates are uncorrecred, how the heck do I get the corrected location to 9 digits? Do I really have to create a log to download and look at later? That seems rather silly. One should be able to get a corrected location on the spot.

Note to Emlid - regardless of what it really is now, consider displaying both uncorrected and corrected coordinates on the status screen. This would eliminate the ambiguity and confusion as well as be very useful during setup. Heck, I would also consider displaying the difference.

1 Like

I would check your mount points again, these units will pick up absolutely any movement what so ever. It looks to me like one of your units is shaking back and forth.

Mounted on the end of a pole that is waving.
Or mounted to a wood structure that is picking up foot traffic.

The surveyors on here have put the RS2 up against a total station and had comparable accuracy.

Base mounting for survey is usually a tripod or concrete monument to prevent this.

Another user on here uses reach to monitor the flex in a dam wall, very sensitive.

I can also measure vibration on my tractor, on vs off using reach.

5 Likes

What does your rover environment look like?

Centimeter precise means centimeter level, meaning lower 1 digit cm precision, not sub-centimeter.

You cannot expect this of any RTK system. Especially in so short observation windows. You will always have a standard deviation attached to such measurements.
You will also see a big difference from the Kinematic RTK mode to the Static RTK mode. You will see a significantly smaller stdev from the Static mode.

There are many things you can do. The easiest is provide a cleaner receiver environments for both base and rover. You can do this by:

  • elevating your units further (at least 2 meters over ground)
  • move them away from trees and buildings. Especially true for the base!
  • stay away from heavy machinery and power lines.

Nop, the dots, are green, so this is the fixed solution.

4 Likes

Nope. Both rock solid. The base is mounted on top of a concrete light standard at the back west corner of my fenced in backyard. The Rover is on a survey pole tie-wrapped to the back East fence corner post. Neither one moves more than 1/2 a millimeter even in strong gusts of wind.

The rover location as displayed on the status screen tends to stay in one general place for quite a few minutes. Basically it slowly drifts over time to other places in the same general location up to 6 inches away.

There is no foot traffic or anything else in the area.

I have very high confidence in the environment.

All the signal bars are at the top of their range.

What you describe for yourself and other users is what I expected.

My Rover is on a fiberglass survey pole tie wrapped to a cemented in steel fence post about 2.5 meters above the ground with a totally clear view of the sky in all directions above 15 degrees or so.

My angular filter is set to 20 degrees. My enclosed back yard is inside a less-meticulously kept 4 acre farm yard. Basically a wide open sky view with no interference or signal noise.

Holy crap! That IS NOT WHAT I EXPECTED! Single digit centimeters means up to 9 cm. 9 centimeters is 4 inches. A four inch error might work for spreading fertilizer or spraying, but in planting it would be intolerable. Are you SURE of what you are saying? NOTHING I have ever read about cm precision suggested it was anything other than 1cm or less. I could tolerate maybe 2.5 cm (one inch), but 4 inches is unacceptable. I’ll be driving on top of previously planted seeds! This makes me wonder if I have just wasted a lot of time and energy! Then again, how can that be right? All kinds of farmers around here use professional systems that are straight as a laser beam.

All of these conditions are met in my current test setup. I am a farmer with a big farm yard on top of a hill. The barn and all my equipment is 250 yards away. There are no power lines. Both units have a clear view of the sky in 360 degrees and both show very strong high signal to noise ratio reception.

I must say that my head is spinning right now. I sure hope you are being pessimistic (managing expectations). Otherwise, I have wasted my time and money if that is what I can expect for my application. That said, I understand that it’s not a good idea to have unreasonable expectations nor to suggest so to other users. So I get where I hope you are coming from. I just pray for much better than that for my situation.

Centimeter precise means centimeter level, meaning lower 1 digit cm precision, not sub-centimeter.
[/quote]

M2
Positioning

  • Static horizontal 4 mm + 0.5 ppm
  • Static vertical 8 mm + 1 ppm
  • Kinematic horizontal 7 mm + 1 ppm
  • Kinematic vertical 14 mm + 1 ppm
1 Like

I wouldn’t panic. You’ll find things will be fine. M2’s (much less all Emlid products) are extremely reasonable in cost even if you feel you wasted cash… which you have not. Anything other than Emlid is definitly MUCH HIGHER cost. Be patient, you’ll get it worked out. The people here are absolutely great!

You might want to also provide some photos and details of your setup… i.e. antennae, ground plane sizing, material etc.

After all, it’s the M2 route which you assemble yourself versus the RS2 route, which things are perfected for you out of the box… not that you cannot achieve these same results as a RS2, but may be something not quite right where you may need to rectify something.

Edit: I see you have lots of prior posts about these things leading up to this. Good luck with everything. The RS/RS+/RS2 units pretty much ready to go.

2 Likes

That’s perfect. And much more like my expectations.

Thank you.

Is that a quote from Emlid’s documents?

1 Like

Can you post a few pictures?

Any reason for the 20 degrees? If you haven’t got obstructions, try 10 or 15 degrees. You can also try setting your SNR mask to 40, that would give you a stronger fix, but requires little to no sky obstructions.

Yes. But I did say lower 1-digit, so not 9 cm. Expect 1-4 cm. Sub-centimeter will never be RTK-territory. Even postprocessed GNSS would struggle to achieve this. This is total station territory.

And on another note, can your machines really use a sub-centimeter precise signal ? Remember the mechanical slob, electronic delays in the systems, imprecision in control-loops, along with tilt and pitching of the gear. If you look at the sum of errors here, the are probably 5-10 cm, if not more.[quote=“Susquatch, post:5, topic:24711”]
All kinds of farmers around here use professional systems that are straight as a laser beam.
[/quote]

Many farmers use WAAS based systems, where we talk about sub-meter, not centimeter.

Would it be possible for you to upload the ubx or fines log from your example?

It is, yes. But as with every specification on any product, these are best case, and your environmental factors will vary your mileage!

1 Like

I am not expecting perfection. But I do want something usable. 2.5cm would be more than enough for me. I fully expect that this project will be a work in progress for years to come with continuous improvement always in mind.

I have posted parts of my project around this forum as it comes together to get input and advice as it evolves. I also plan to post my entire project in the projects section when it is complete.

In the meantime, here is a shot of my rover in its current phase of completion.

The base is in process right now but it will be similar. For now it is attached to a steel bar on my lamp post. Here is its central backbone mounted in my lathe.

I plan to mount the base to a 5/8-11 post permanently attached to a tall concrete and steel lamp post in my back yard instead of on a survey stick but it is currently planned to be similar to the rover using an aluminium pot to provide rain protection and a ground plane for the antenna. The current plan for the rover will be to make it swappable between the survey stick shown below or to a 5/8-11 stud at the top of a bar located on the brush gaurd of my tractors.

The basic system has an internal aluminium backbone with a 5/8-11 bottom threaded hole and a top flange to attach the pot and the antenna. Inside you can see the Lora radio, the M2, and a high capacity USB Battery Pack. They are all temporarily attached using tie-wraps but at some time in the future, I’m planning to make some cradles to hold them so they can’t vibrate loose. I also plan to do some vibration testing for the rovers and add lead weight or rubber dampers as required to dampen any tractor induced vibrations (basically move the natural frequency away from the exciting frequency). Tractors are big heavy things that don’t really vibrate as much as you might expect.

1 Like

I still do not trust the tie wraps or fence. Push your fence in both directions and measure the deflection.

Is you rover set to static or kinematic? if your rover is in static it knows it should be motionless and helps filter out things that may be perceived as movement. If its set to kinematic it knows it should be in motion and has special filters for that as well.

Here is a grouping of 100 counts using an awful fix from a non emlid base 30km away, the rover is an m2 and is completely motionless. The scale of the map is 50mm i just could not screenshot all the info.

2 Likes

Remember that this antenna-design isn’t really meant for Ground-use in my mind.
It works very well on UAV’s being high above the ground and thus naturally don’t have to deal with multipathing.
However, on the ground, not so much.

5 Likes

The 2 horizontal bars might even thermally expand and contract during a day, moving the pole around, adding to the sum of errors.

1 Like

The M2 showed me that my 72" lawn mower was actually more like 69.5" blade tip to blade tip, its very accurate.

2 Likes

See previous reply.

Yes there is. The pictures show my development setup. However the field setup will not be so favourable. My fields are long and narrow with tall hedgerows and forests around them. Also, my tractors have parts (like an operator cab) that stick up. I chose 20 to minimize most of that. If anything, I may need to increase it.

I can probably live with 4cm, not 9. I had hoped for 2.5cm (1 inch).

All not a concern. I live in flatland. No farm equipment involved (yet). No autosteer (yet) - manual guidance only. Etc etc. Accuracy isn’t an issue until you go to follow the same path again and don’t know where the first one is. Tractor tires are big, plants and seeds are fragile, rows are narrow. A few inches one way or the other equals dead plants. I’m sure you get the idea. I do a pretty good job by eye. I want to be better than that. 2.5cm will allow me to do that. Farmers don’t use waas for anything but wide pattern spraying and before-plant broadcast fertilizing. It’s a big help for that, but you are not gunna plant, row weed, row fertilize, or harvest with WAAS. Ask me how I know…

I’d love to. But I don’t know how to do that yet. Learning how is on my to-do list but hasn’t been a priority yet. The weather, my lathe, and the bride have been calling me!

1 Like

Yeah, not a fan of pots and pans. That’s probably part of the biggest problem. Thick and heavy to? : / try something else more typical of what other users have done.

20 is already a bit high, increasing it further will lower your PDOP even further. As a rule of thumb, for a single constellation (GPS only i.e.), your PDOP should not be over 2-3 for a good RTK-result.

But can one even manual steer to within 1 cm in a big tractor on semi-even ground?

I hear you, but… I have my reasons.

Rain protection
Temp control
Packaging
Spider & wasp fence
Low cost

I’m not married to the pot. I’ll jump ship the minute I hear a better idea or a good reason why not.

Reply @Susquatch

I really like 12degrees, and 30db for equipment. Both picked from reading studies on GNSS performance testing of these parameters. I still have copies somewhere on my hard drive.

I think your pot really wont affect the accuracy of measurement much. But a base really needs to be solid, you can steer the tractor by running around with it.

If your base is 100%, then your rover will get you some excellent data, that sometimes finds things that you think are stationary move quite a bit.

2 Likes

What about your own time spent re-inventing the wheel ?
I know that M2+Antenna combo is a nice, cost-effective solution for a ground-rover at first, but, if you add your engineering time on this project already, maybe the RS2 is suddenly cheap?

It already offers you solutions on
Rain protection
Temp Control
Packaging
Antenna design
Spider & Wasp fence
Easy to mount/remove
etc…

2 Likes