SparkFun Forums 

Where electronics enthusiasts find answers.

Tips and questions relating to the GPS modules from SFE
A friend of mine and I are trying to build and then document in laymen's terms, a total RTK-GPS system that will be able to do very accurate (hopefully <10 cm similar to high-end commercial systems) GPS positioning using inexpensive hardware and free software.

(see )

Basically, we feel as if super-accurate GPS positioning is the sort of tool that makes people have creative ideas of what they can do with that data.. For example, given some time, it may be possible to -say if there is an earthquake, the earth moves.. measure that change, measure continental drift.. measure the rebound in formerly-glaciated areas we are seeing now after the ice age.. build accurate 3D models for archaeology, etc..

It appears that the weak link in the chain for us at this point may be the antennas.

We want an accurate, low-cost antenna that is able to receive righ-hand circularly polarized signals from the entire sky at as close to a single point, electrically as possible, while rejecting reflections well. Its turning out that issues that are not important when the criteria for success is merely getting lots of satellites still make it impossible for a lot of antennas to work well in a high precision application, because they delay the signals coming from some directions enough more than others to make the fixes literally jump around a substantial amount from satellite to satellite. (if you are looking at the raw data coming from the processor)

Ground reflections, probably, are what causes the wildly jumping errors we are seeing. Some antennas (choke ring antennas) cancel them out..

We understand that both in theory and in the real world errors can be reduced to very small distances, in the real world they often are- with expensive antennas.. Thats why they are expensive. The better antennas are also often relatively bulky. They have to be positioned properly, too, to actually be accurate.

What are your experiences with GPS antennas?
Last edited by trillium on Fri Aug 26, 2011 11:05 am, edited 1 time in total.
Hi Trillium,
Before pointing to the antenna as the culprit, it would help to understand what you mean by high precision (millimeters, inches, feet, meters?), and by wild jumps. Also, are you looking at the NMEA messages, and is your receiver using code phase or carrier phase measurements?
Generally, wild jumps can be caused by
a) multipath, i.e. reflections by eg buildings which temporarily become stronger than the direct signal.
b) a change in the set of satellites used to compute the location fix. Normally, only a subset of all visible satellites is used for the computation, the others are just tracked, or kept "on standby". Depending on evolving geometry, visibility etc., single satellites are swapped in and out.
This is what I am using.. with a Skytraq-module-based receiver:

T.Takasu and A.Yasuda, Development of the low-cost RTK-GPS receiver with an open source program package RTKLIB, International Symposium on GPS/GNSS, International Convention Center Jeju, Korea, November 4-6, 2009 ... b_revA.pdf

also see

I also should edit my previous post to include the fact that I also now realize that it makes a LOT of difference that my homemade antenna is in a skylight at a 45 deg. angle - (also only a 2/3 sky view)- and NO ground plane per se yet...
So I need a better antenna, and I also need to get it as much in the clear as possible.. either up on a pole, or possibly, down, on a flat metal ground plane in the middle of the lawn.. (we'll have to see which one works better)
Some of the jumps are very large.. I don't know how large they translate into on the fix because the fix isn't calculated from just one.. As you, elevator4, bring up, though, its also quite possible its not the antenna, it may be something else.. too.

But I suspect it is antenna issues, as they would explain this. I really don't have any other explanation for what I am seeing..

My experience so far seems to be that I get more accurate fixes when I am low down nearer to the ground, even with a limited sky view than when i am up near the skylight with a larger potential for ground reflections of substantial distance. There are some large objects nearby that might be quite reflective to RF coming from satellites near the horizon, if its bouncing up.
Not likely afordable, but the Antcom active L1/L2 Choke Ring pipe mount antenna is a good candidate. Just need to wash the bird droppings off it once in a while.

All GPS units will have out of family pseudo ranges even on perfect days. I get "out of family errors" from my GPS on occation and the unit is nailed in position. I suspect birds and rain aggrevate the problem. God forbid you are moving or have multipath issue. Remember that at L1 frequencies lots of items will reflect and or absorb. Also remember that the GPS signal is hardly above the noise floor. It is amazing the system works as well as it does.
thanks for the extensive edit to the OP after I posted my reply - it sure makes the latter look like that of an idiot.
Anyway, back in 1999 we were eyeing structural movement and landslide-prone regions. Using a novatel frisbee-shaped antenna and CMC Allstar receivers, we were able to achieve sub-centimeter precision. See ... df/c99.pdf for an example application involving 10 permanently installed "rovers".
The key was in the firmware and post-processing, actually. Even using regular patch antennas (CMC again) the results were consistently better than 2-3cm.
If I were you, I'd start with choosing the right receiver. Skytrax' founder, a longtime friend of mine, may be able to provide some guidance as to what you may expect from their chip in terms of CP performance.
Yes, that makes complete sense. I would love to be able to make something like the cool novatel antenna, but its beyond my technical skills - I have a friend who I am brainstorming this with and Ive been doing a lot of reading. The best we have been able to come up with to date is a shorted annular patch antenna, a ring shape, with a hole in the middle thats shorted around its edge, sort of a very large circular via.. which can be made on a PC board in a square shape.. not too large. It needs two feeds at 90 degres apart, and a phasing network of some kind.. still trying to figure that feed part out.

So.. there are a lot of isolated geeks with the similar idea, and I think we need to get together so we can work on this as a group.. I am not an RF person , I am a web person, and I am learning as I go. Luckily I think this is really interesting so thats a good motivator.

You must mean Oliver Huang of Skytraq? From the single email exchange I had with him he sounded like a smart guy and I think that his products are very good. I would be surprised if I can't do as well as receivers costing much much more, perhaps even the limit for the technology with it.

It would be great to have Skytraq give a modest level of technical support (information, not money) to the open source community.. help to know what was possible.. I think RTKlib is an independent project, and I get the impression they could use some more info on the Skytraq's internal command set..

They could really turn the game around with a committment to democratize the technology.. As it stands now, there are only projects like Michele Bavaro's Yuan 10 receiver - I think he is the first to sell a product - at least in the English-speaking web, based on that chip.

But Skytraq can't be making it for nobody, where do they go?

Its a game changer..

Two of te Ublox chips also work with RTKlib.. three if you use a specific firmware version.. Also there are other chips out there that let you access raw data.. most more expensive or older/less sensitive I would imagine..

This post-processing/low cost DGPS meme has been building since at least 2002. I'm sorry I went back and edited the original post, I did that because in retrospect, I thought I could have explained what I was trying to say better.

I really am just learning this.. I make huge mistakes in my assumptions all the time. I am lucky that I have gotten help from people, thats been very helpful.

Its time to do this thing and get people excited about a technology that most of us only know a little, but I am starting to think, could make *mathematics* the new, cool gateway drug.
fil-freak, what is the difference between a "timng" antenna and a geodetic grade GPS antenna? Do the timing antennae take pains to only reciive satellites higher than a certain angle?

Also, "out of family" error just means a statistical outlier, (i.e any odd multipath error that isn't repeated..) right?
Not sure about the difference in antenna types. This was picked by our antenna specialist for roof top environmentally sealed stationary active antenna. I do notice that most SVs bellow about 25 degrees seem to be masked. I think that has to do with the choke ring.

"Out of family" is a pseudo range that solves to a solution that is well outside the statistical average of the other pseudo ranges. Do have this type of error, you must have at least 5 Svs locked as you need to have an overspecified solution to determine "out of family" solutions.
Elevator4, thank you for posting that paper on the Genoa breakwater experience.

Are you, yourself in Italy?

I am thinking you might know my friend Michele Bavaro, who built the little receiver that I have been playing with.

This is his web site..

His Skytraq S1315F-based receiver, the Yuan-10 is at: ... ers/yuan10

Its quite amazing, this tiny, self contained carrier-mode GPS receiver..
trillium wrote:I am not an RF person
Myself, just a tiny bit, except antennas, which I consider magick. But I halfway remember a DIY antenna project way back then... maybe Radio Amateurs, let me do some research.
Also, here's a few suggestions for bug-hunting: 1) do a zero-baseline test with 2 identical receivers. 2) do static short baseline tests (say 1 meter with defined orientation) between two identical receivers but different antennas. Say one patch, and one piece of wire (not RHCP, picks up everything :lol: ). You may want to have the tests run for hours to cover a range of sat geometries. 3) Play with masking angles (you should be able to configure the receiver for the purpose).
trillium wrote:You must mean Oliver Huang of Skytraq? ... But Skytraq can't be making it for nobody, where do they go?
That's the man. AFAIK Skytraq are doing well, they are still at an early stage (and pretty busy...). I am currently designing their '638 into three projects: a logger, a tracker w/GSM, and an aircraft/RC AHPRS. Any of the three could be used as hardware platform for a CP rover application BTW.
trillium wrote:...Genoa breakwater experience. Are you, yourself in Italy? I am thinking you might know my friend Michele Bavaro, who built the little receiver that I have been playing with.
Yes, temporarily. I would love to meet Michele for a chat sometime. Nice work he's doing.