Most accurate gps chipset

[jiraya]

Working on a laptimer project. Speed and Positional data accuracy is of primary importance. Browsed over the following options.
Comments / Suggestions Please. If there is a chipset which is better but missed below please educate me

List compiled from https://www.intorobotics.com/14-gps-mod ... i-project/

1. NEO-6 GPS module
2. RasPiGNSS
3. Ublox MAX-M8Q
4. GPS shield
5. SiRF StarIII chipset
6. Adafruit GPS Breakout
7. 3DR uBlox
8. 3G/GPRS shield

*Info collected

[topguy]

Best antenna wins.

A combined GPS/GLONAS + compass + accelerometer, and your accuracy should be as good as they come.

[Heater]

What are you trying to measure that lap times of? And how accurately?

That rather determines the positional accuracy and update rate you need.

[bensimmo]

There has been a new leap in tech on this, I think Broadcom (47755 iirc) are making some with accuracy down to less than a meter (30cm/1' )now and probably others may be using it as well.
Look for L5 use.
Normally you are looking at about 5m accuracy, less if you sit there for a while and could get to 3m


Price, when and where...

[scruss]

As topguy said, best antenna wins.

If you're looking for something really accurate, any RTKLIB-compatible chipset will do. It's not fast, and needs a nearby telemetry link and a lot of number crunching, but it can give you centimetre precision.

[Heater]

There is the thing " It's not fast". With that and the position accuracy one can expect I can't imagine how GPS alone makes a good lap timer. Especially in close run events.

Unless you are racing snails around Silverstone Circuit

Or a yacht race perhaps?

[scruss]

Most of the more modern chipsets can kick out updates at 10 Hz or better, so while not ideal for F1, might be useful for some track applications.

[mfa298]

Best antenna wins.

Not always, As long as you have enough satelites you should get an accurate fix, but if you're picking up weak bounces off a building that could affect the accuracy.

Most of the more modern chipsets can kick out updates at 10 Hz or better, so while not ideal for F1, might be useful for some track applications.

Just because you're getting 10 positions a second doesn't mean they're accurate positions.

---

Some GPS modules can be configured to work in different modes, e.g. the UBlox has various Dynamic Models it can use, these vary the assumptions it can make which might lead to greater accuracy at the expense of other factors (altitudes it will cover, speed it can travel, needing to be stationary).

[bensimmo]

It's not just that, it's just a restriction in the system used.
Only recently has a higher accuracy channel been available from the satellite

Even a GPS/GLONASS combination is only down to 2 meters iirc. Though it does give more Satellites to get a lock on (hence less drop out)
You'll need an additional systems for what you need. (If you actually want something more than a random looking chunky loop).

Anyone that logs rides and things can see full well how you can be riding through fields and houses, cut corners etc.

Hopefully the new generation will actually improve this.

These are the systems they're hoping to use to know which lane you are actually in.

[jiraya]

Thank you all for chipping in. I have narrowed down to evaluate the adafruit gps breakout and ublox neo M8N chip.

Testing the adafruit gps chip i see i get back data only at 1 second intervals. The data sheet says 10hz. Is there some config necessary to let gpsd know i want max updates ?

[topguy]

You have to tell the GPS to send positions faster than 1Hz. This usually means you also have to increase the baudrate or this will become the bottleneck preventing the GPS of doing 10Hz for example.

Different GPS chipsets might have different commands to do these things, and GPSD is not programmed to know all the proprietary command sets for all GPS-chipsets.

Googling the "name of chipset" + "programmers manual" often gives you the PDF you need. I dont know is GPSD support sending special strings to the GPS on startup, never tried it.

[Ronaldlees]

I wonder if Galileo is possible yet with the Pi:

viewtopic.php?f=46&t=168473

viewtopic.php?f=46&t=177228

It was said to be pretty accurate.

[Ardusimple]

In case anybody is interested we just launched the simpleRTK2B, the first low cost evaluation board for the ZED-F9P GNSS module, compatible with Arduino and Pixhawk autopilots.
If you want to see all the details you can go here: https://kck.st/2NhLhBW

[skypuppy7]

AIUI, the clocks are some of the most accurate on Earth and in near-Earth orbits. You don't need location to get accurate clocks. If you only want lap times, click to start, click to end, subtract start from end and there ya go. Any GPS will do. You'll need a small program to do that on the RPi.

That said, I've been playing with the Ardusimple F9P boards and I am -extremely- impressed for their positional accuracies relative to costs.

[raspitrick]

There's a lot to consider regarding GNSS accuracy.

Good antennas is one point (make sure it fits your GNSS chipset capabilities!), but accuracy highly depends on what and how the GNSS fix is calculated, and how many GNSS constellations (GPS, GLONASS, GALILEO, BEIDOU, ...) and which GNSS bands (L1, L2, L5, ...) the receiver supports (which determines the number of satellites and frequencies available for calculating a decent solution).

Basically, there are three variants:
- Code solution. This is the standard solution you'll get from your OTS GPS, mostly delivered in NMEA-0183 format, that gives you an accuracy (RMS of position error) of around 2,5m 1-sigma in 2D (roughly 2m CEP).
- Precise Point Positioning (PPP) solution: In that case the GNSS receiver needs to deliver raw satellite measurements, which are fed into a software (like RTKLIB) that computes a PPP solution with the help of (public or commercial) correction data (e.g. from IGS). Correction data can be made available pre-flight, online, or during post-processing. Accuracies of approx. 30cm 1-sigma 2D are feasible.
- Real-Time Kinematics (RTK) solution: In this scenario a (fixed) base station is provided in a "nearby" location, that delivers correction data in real time to the rover. Both the base station and rover GNSS chipsets need to be able to provide raw satellite measurements, and again with the help of suitable software (RTKLIB) a so-called carrier phase solution fix can be calculated, with accuracies down to 3cm 1-sgima 2D. The drawback is you need a permanent connection to the base station.

You'll find more information about GNSS constellations, bands, and solutions, at Navipedia. You can download a port of RTKLIB for the Raspberry PI at https://drfasching.com/downloads#rtklib.

The RasPiGNSS "Aldebaran" based on the NV08C-CSM which has been around for many years now can do all that, albeit only on the L1 frequency. However we'll shortly release two improved successors, the RasPiGNSS "Betelgeuse" (based on the uBlox ZED-F9P, supporting the L1/L2/E5b GNSS bands), and the RasPiGNSS "Copernicus" (based on the uBlox NEO-M8T), that support more GNSS constellations and bands (watch http://raspignss.tech for announcements).

[CAredJAG]

The UBLOX C099-F9P application board is a very well sorted-out solution for RTK, it has onboard wi-fi and BT, matched with a decent antenna it will perform at modest distance base-rover configurations as well as vastly expensive alternatives. If cost is no concern I might look at Septentrio products.

https://www.u-blox.com/en/product/c099- ... tion-board

[philrandal]

I wonder if Galileo is possible yet with the Pi:

viewtopic.php?f=46&t=168473

viewtopic.php?f=46&t=177228

It was said to be pretty accurate.

Supported on the ublox m8 for some time.

I use an Uputronics ublox GPS hat as a GPS source for ntp, using the Galileo constellation.

http://www.philrandal.co.uk/blog/archiv ... y_213.html

https://github.com/philrandal/gpsctl

Cheers,

Phil