Wandering Fix

Thanks Bryan. That’s more or less what I have been using for a battery for my Rover too. I’ve been using a charger adapter for the base and I thought perhaps it might be too noisy. So I tried a battery there too and it made no difference.

No testing today either. Maybe tomorrow.

1 Like

Others have raised concerns about the pot too, so testing without the pot is a priority for me.

However, it’s important to know that the pot is Coated high grade aluminium about 4mm thick. The base is perfectly flat and polished with no coating on it and it is only slightly larger than the size of the ground plane recommended by Emlid. In fact, it was Emlid’s antenna recommendations that caused me to buy it. The pot was the cheapest way I could find to buy a nice thick flat piece of solid aluminium that met your size recommendations. Originally, I was going to cut the sides of the pot off, and just keep the flat part, but then I realized the sides of the pot made a perfect weather enclosure and the open bottom would ensure that it would never get too hot. Therefore, I have kept the sides for now. If testing shows that the sides are bad, they will be replaced with an upside down plastic margarine container because I really do like the weather and temperature protection that I believe an upside down container provides. If subsequent testing shows that the ground plane is bad, then that part of the pot will go too. I’m not married to anything except getting the best performance that I can. But I am a bit stubborn about not letting unconventional ideas get tossed without giving them a good try.

I confess that I do regret my antenna purchase though. The Helical unit was the one packaged with the Emlid. I just didn’t know enough at the time to realize that it was specifically intended for drone applications. I assumed that it had to provide decent performance or you wouldn’t sell it.

I can’t test everything just yet. We got a dumping of fresh snow yesterday. I have to wait a bit before I can get outside again. But I’ll keep you posted about my findings and progress.

1 Like

Tallysman Antenna

@Susquatch You seem to be the first guy who has had issues with this milspec antenna.

If you look at my screenshot with the poor fix off a distant off brand base, it works quite well under very poor conditions.

Also i think you will like that its the only one that sheds dust in AG applications, you never need to stop for a wipe.

The factory that makes it is in Ottawa Ontario, right close to where you are at.

1 Like

Hi @Susquatch,

I understand your point regarding weather and temperature protection. However, it’s indeed hard to predict how it affects accuracy without additional tests. So, yes, please keep me posted. I’m interested in figuring out your case.

Reach M2 antenna should work well independent of its application. It was designed for surveying in general, not for drone mapping in particular. I’d say that there is no big difference between these spheres in terms of required antenna specifications.

Also, we’ve tested our antenna in different conditions, and it works the same way. So, the issues you experienced aren’t related to the antenna. We’ll try to sort them out.

Great day today. I cheated a bit on my plans. My bride had to help out with our grandkids so I have the day to play with my gps system.

I had originally planned to test without pots out in the open today as my first priority. But convenience got in my way. As I mentioned previously, I finally finished the pot enclosure for my base, and I had to establish a new base location anyway, so I set my base up in the regular location WITH THE POT to do that. Then I set the rover up in the planned new location again WITH THE POT to get a location that could be plugged into the base.

The first thing I noticed is an immediate solid fix. So I set the status map to collect 10,000 points and let it go.

Amazing… Everything seems to be working just fine! The fix stays in the middle of the cluster the vast majority of the time and only wanders a little from time to time. So I feel pretty good about the accuracy for my planned field work.

Both Static and kinetic work almost the same. Maybe as Potatoe Farmer said a wee bit better in static mode.

Maybe the pot works after all by reducing multipath at the base! (reflections from the house, the fence, the pergola, the bushes), Or maybe it’s just the rover that is working better in the new ideal position with less multipath. Or maybe it’s a combination of both.

After lunch, I plan to move the rover back up to the house to see if it was the rover, or the pot, or perhaps both. Nothing will be proved absolutely, but it doesn’t matter. It’s working in the configuration I originally hoped to achieve, so I am MUCH happier.

After the simple tests this afternoon, I have to work on optimizing my standard setup. What constellations, what update frequency, etc etc

Here are my current settings, provided both to solicit comments and suggestions, but also just to document what worked so I don’t forget.

Edit : no idea why the submitted format is so messed up on the forum. I’ll take screen shots and edit again shortly.

Comments and recommendations appreciated


I moved the Rover back up near the house. No difference, works fine.

So based on this one pair of tests and very limited experience, I would say everybody was right. The problem was too much multipath around my base (house, fence, bushes, pergola, etc). And, it seems the issue was significantly reduced by adding a ground plane to the base. (I’ll refrain from calling it a pot…).

I plan to try a few field runs shortly. If they work out, the next step is brackets and power supplies for my two tractors.

I did a quick field check. Seems to be working very well now. Using “Field Navigator” on my Tab4 Tablet, it was pretty easy to setup a field and do a little pretend narrow spraying on my atv.

They say we should be careful what we wish for. Boy does that ever apply. It may take me a lifetime to drive to those new guideance lines. SENSATIVE!

A marker disk line might be easier.

Spraying and spreading should be a piece of cake though.

Question: My main tractor for spreading and spraying has no cab. My Tab A is really hard to see in the sun. My S21 phone is easier, but isn’t very big. What do you use for a display? Does a hood help?

1 Like

Ok some setting suggestions

Base set to static, Tractor rover should be to kinematic. Kinematic will help on the tractor because the filter will assume you are moving on a path increasing the accuracy of the path.

Mask; I would set the mask to 12 degrees, if your under trees it will still be able to squeak out a position, this will be the same on base and rover. The emlid will only use the best satellites for the solution anyway, the rest it just tracks until they become the best to use. Its only using 3 to 6 to compute a position.

If you are taking the base with you and distance is not an issue "Within the section, 640acres " to the tractor try a higher data rate. Somewhere around 19200baud, the emlid lora values are a little lower than the normal 232 baud values. I always leave radio power cranked no RF emission rules apply at 0.1w in my area , frequency is whatever works best for you.

The faster you can pump the correction in the faster it can be used to compute a solution.

You only need to send the corrections that the rover is using for a solution. In your current state Beidou would not be needed.

But I suggest using all constellations, except QZSS in north America. Reason being if you loose fix at the tractor the computed position off the Emlid is better than WAAS accuracy.

Here is a field harrowed using an m2 all constellations active and no rtk corrections, still very straight.

I am not sure what Hz your light bar program requires, but you wont be able to react faster than 1hz steering by hand anyway. With automated guidance the fastest it can handle 5 or 10hz is beneficial, it allows the errors between readings to be smaller and requiring smaller steering movements.

All subsequent settings off, perfect no use chewing up processor power on stuff that you are not using at all. This will also lower power requirements, and module heat creation.

For your base coordinates, if you do have a good spot to put the base year after year, survey in for the maximum time. Once the reach has coordinates, find the drop down that locks those to manual. Write down the coordinates for future use. As long as the same coordinates are used for that exact spot are used, your passes will be the repeatable year after year.

Your passes will be completely accurate relative to themselves and repeatable, though the position may no be absolutely precise to the survey. Absolute Positional accuracy is not required for farming, and requires PPP, or other survey methods to accomplish.


Wow! What a fantastic comprehensive response. When I saw your harrowed field I got homesick. I miss the prairies…

Got it and understand why.

Hmmmm, I had trouble with a low mask at the base. There is a forest to the west and a house to the south. I set the mask based on an angle measurement to clear them. I’ll try a few other settings over the weekend. Does the rover really have to be the same as the base? Can it be a lower angle? And if so, what happens if the rover finds a good satellite at a low angle but doesn’t get a correction for that Satelite from the base?

I don’t plan to move the base at all other than back and forth from the house to its location at the top of my lamp post. But the wife wants a clothes line, so I may move it to the top of a clothes line post if that works out. However, I tried higher data rates and the fix loses reliability for some reason. But I’ll try it again now that the ground plane is installed.

OK, I’ll kill corrections for Beidou. Frankly, I’m not sure why I had it set to send corrections for anything but GPS & GLONASS.

Ya, I had the same issue with this as for the baud rate. For some reason the volume of data seems to choke my system. That’s why I disabled everything but GPS & GLONASS. I’ll try it again with the ground plane.

So beautiful…

I hear you… I did have problems with the program at faster update speeds anyway. It started throwing triangles instead of a path. I might even slow that down some more. I am old and slow… I won’t really know what I can follow until I actually try. It was all way faster than me today.

Actually, I have a nearby monument and I used that to establish a good fix for the base. See my other thread on establishing base coordinates. Establishing a base coordinate without a reference

Bottom line is that I created a corrected manual input by comparing my rover to the monument actual. So my fields should all be good both year-to-year and true to actual accurate Lat/Lon and I should be able to survey with it too.

In summary, I’ll play with your suggestions to get as close to them as I can and let you know how it goes. With the ground plane in place, it may turn out to be easier than I fear.

Thanks again!

I sure do love your farm…


The mask is anything below that angle gets ignored. But now the fancy math comes in since one is mounted to a tractor that does not stay perfectly level. If your mask is set 25 degrees, and now your tractor nose is pointed 25 degrees in the air driving uphill , you have effectively made a 50 degree mask in front of yourself that drops a lot of satellites in view. The antenna sees all ways, but not out the bottom.

If the mask on the base is high and ignores satellites that your tractor needs corrections for you end up loosing fix. Best to have them set lower and the same.

Well why not set them at 0? you might ask, the lower toward the horizon the more messed up and dirty the satellite signals ( em radiation) are due to having to travel through more of our atmosphere. The same way light (em radiation) gets messed up making the sun look huge at sunset.

Lora and baud rate, the lower the baud rate the further it will go unobstructed.
The higher the baud rate, the more data can flow faster.
Using slow baud rates means you will have to use less constellations that reduces data.
If we are only using one constellation, the lowest amount of data, it is still advantageous to use the highest baud rate. If a re transmit is required it takes up a fraction of the cycle, also less chance of running up against the output buffer.

But Lora is line of sight, it pretty much only works well if you can visibly see the other antenna.


OK, got it and now understand better why. Thx.

OK, I have seen that elsewhere too. But I ran into trouble at high rates for reasons I don’t understand. This might turn into a matrix of testing to wring out what is best for me.

Although I was expecting that, for some reason, I have not found it to be true. When I was establishing my exact base coordinates, I had to use a monument that is 2-3km away. The radio had to go through my house, through a hardwood forest (oaks, maples, walnuts, and wild cherries), across a large farm field, through a small subdivision, and across another field. I fully expected it to fail but had to try. My fallback plan was to shoot along the highway to establish a temporary base and then up to the house. But it worked first crack - albeit probably with the 6" wander error we have been discussing here. I cannot explain this but am thrilled with it.

I’ll try using higher baud rates with the fewer constellations I am using now. It would be nice to see Emlid publish some limits on the inter-relationship between baud rate, updates, correction rate, etc. Frankly, it all seems weird to me. With processor speeds like those commonly available today, I would have thought that 112,000 baud would be like a week long nap between bits for the system.


Off topic, but just curious… can Field Navigator or the like do corn mazes? Wonder how detailed if so? Would be nice if could import vector paths to follow for elaborate work. I.e. .eps, .svg, .dxf, .shp, .json, etc. I know their are converters and ways to deal with the vector curve line work, but the app or software to pull these in for a gnss guided mobile wheeled rover/drone etc is what I am not finding. Just seems very simple limited flight path capable? I.e. rows on a field. I need something that can input hundreds or even a thousand or so wayfinder points versus just a couple digits of them. Or can even do arcs and curves in addition to line segments.

PS, glad you got your wandering fix problem solved… Emlid products are great! ; )


So, remember that I am a greenhorn…

But I think it should be easy to do. Instead of thinking of a path within a field, think of the path as though it IS THE FIELD. You can create your field using manual methods or by importing coordinates. Just make your field the usual way and then draw your maze within that field. The first field isn’t really that important. It only serves as an outline for you. The inside field is the one you need. Just spray or mow it in the usual way. Sounds like a fun project!

Edit - It is prolly important to set the impliment width to the maze path width and set the tractor to impliment length to zero so your corners work out properly. Also, most of the Ag programs that I have looked at will allow the user to import fields. The authors probably never imagined a use like that, but I can’t see why it wouldn’t work.

Here is a quick concept test during and after field creation.

My program doesn’t handle rounded edges, but many do. When I want rounded edges, I just add a million points to define the curve.


Hi Potatoe Farmer,

I’ve been looking for low ripple usb power supplies to provide the power to my rover M2 from the tractor. There are lots of choices but I have not been able to find anything claiming low ripple. I believe this is important. Where did you get the low ripple usb power supply you mentioned and what model is it?

for 12V to 5V buck converters and usb power supplies off Ali are usually pretty good.

Most ripple is stepping down from 120 to 24, 12, or 5v. Then I use instrument power supplies that I get out of old industrial PLC equipment before landfill. The Weidmuller and Phoenix power supplies are pretty pricey but the output is like a battery on the oscilloscope.

A cheaper alternative is power supplies for external hard drives, you will see on the adapter “for computer use” or something to that effect. they are dead flat too. I have some from old sea gate 3.5" externals.

I see.

In my case, I am looking for a 12v to 5v converter for the tractors. I’m not sure how they are made these days. They used to be shunt supplies in my day. That would have no Ripple at all but be very inefficient and get quite hot. I’ve been contemplating just making one cuz I don’t care about the efficiency or the heat. But if the regular 5V adapter blocks work, no need to re-invent the wheel.

I like your analogy of flat as a battery. Excellent visualization. LOL!

I also like your idea to use an external drive supply for 120 to 5. I’ve got a bunch of those laying around too.

US $1.78 30% Off | In stock!5 Types Car Power Technology Charger DC Converter Module Single Port 12V To 5V 3A 15W with Micro USB Cable Durable

I usually get these off Ali, they work good. I find if your power supply is dirty or sized to the bare minimum everything works harder.

Ive measured surface temps of the m2 around 55C under normal operation.

Dirty power makes them heat more

1 Like


It’s potted, but it looks like a shunt power supply to me.

Did you measure the Ripple at various draws? Does a cap at the M2 help?

Did not add any caps

But I do over size them if I need 2A i get a 3A supply.

1 Like

Sounds good. Prolly perfect for the tractors. Might put a scope on it to see what it looks like.

Thanks once again.

Edit - It is ordered.