Sticking a GPS module in a project has been a common occurrence for a while now, whether it be for a reverse geocache or for a drone telemetry system. These GPS modules are expensive, though, and they only listen in on GPS satellites – not the Russian GLONASS satellites or the Chinese Beidou satellites. NavSpark has the capability to listen to all these positioning systems, all while being an Arduino-compatible board that costs about $20.
Inside the NavSpark is a 32-bit microcontroller core (no, not ARM. LEON) with 1 MB of Flash 212kB of RAM, and a whole lot of horsepower. Tacked onto this core is a GPS unit that’s capable of listening in on GPS, GPS and GLONASS, or GPS and Beidou signals.
On paper, it’s an extremely impressive board for any application that needs any sort of global positioning and a powerful microcontroller. There’s also the option of using two of these boards and active antennas to capture carrier phase information, bringing the accuracy of this setup down to a few centimeters. Very cool, indeed.
Thanks [Steve] for sending this in.
This is great! I need to start checking Indiegogo once in a while.
Anyone know what the software ecosystem is like for this LEON architecture? GCC or otherwise free tools would be preferred, of course.
Looks like it’s a SPARC core so GCC should work without too much trouble.
What about the Solaris compiler? It’s an olde beaste from bygone days of yore but it does generate good SPARC code.
It’s a LEON3 synthesizable Sparc-v8 processor designed by Gaisler Research used inside the chip. It has its own supporting gcc tool-chain. Using a different Solaris compiler likely will take much efforts bring it up. If using NavSpark, it still has 490KByte Flash + 45KByte RAM while running 10Hz GPS, still plenty room for small applications.
Interesting that it’s LEON and not ARM. I have seen digital cameras with SPARC chips in them before though.
http://akizukidenshi.com/catalog/g/gM-06896/ — This little ZigBee module (you can run your application code on the module too..) is OpenRISC based. Maybe ARM licenses are a bit expensive for lower volume vendors.
The VHDL for Leon was always free open source so if you are developing with FPGAs it makes sense to stick with it I guess.
5-stage pipelined LEON2 is open-source, free for all use. 7-stage pipelined LEON3 needs to be licensed for commercial use. It’s much cheaper than using an ARM core.
How much can an ARM core cost? I’ve ordered a bunch of arm CHIPs (i.e. the hardware) at EUR 0.32 a piece. (They are hinting at the 32 bits of the architecture in the price, just like TI is hinting at 430 with their launchpads).
The thing is, when you’re selling at such a huge volume, the cost of the license (which I would guess is anywhere between $100k-$1million) is easily amortized. For someone who isn’t in the same class of huge companies like TI, this isn’t really a realistic solution.
You’re off by one, the LEON1, LEON2, and LEON3 are all FOSS, whereas LEON4 is proprietary.
I like!
How nice to receive all three systems with one cheap board!
I really like :-)
(do wish it was a uC I was used to..like ARM)
It’s truly a good looking design. I think you’ll be pleasantly surprised by this architecture – something worthy to add to the resume.
The title is missleading. Reading the perks descriptions it’s not three systems with one board but GPS only, GPS + GLONASS or GPS + Beidou!
The “Piksi” project (http://swift-nav.com/piksi.html) is charging $700 for a pair of receivers and until now was the only consumer-level GPS system capable of centimeter-level precision. In an industrial setting you’d be paying many thousands for such a system.
I’m a little wary when Navspark says they can do it for $30. It would immediately make many forms of surveying equipment obsolete.
People have achieved decimeter-level resolution using custom firmwares on OpenMoko. I guess as long as you have low-level access to the GPS signals and the hardware for accurate carrier phase measurement, the rest is just software (e.g. RTKLIB).
But it still does require a base station within some 10 km range. That is why their NS-RAW perk includes 2 receivers (which seem to be USB only..)
There is a serial output line available, I was wondering the same thing if you would need a USB host for the NS-RAW, but that is not the case after asking Oliver about it. The TX pin of the Venus822 which goes to the UART to USB bridge chip is pulled out to a pin and available. No RX line though since it would conflict with the RX line coming in from the USB bridge. I haven’t done much reading on it yet to confirm, but my theory is you could just tap this TX pin out to a radio module and relay the data back to whatever you have running RTKLIB. Operating under the assumption the module just boots up and starts spewing info though, since a RX line isn’t available to talk to it. I’m going to be in for a pair.
You could always just use a RPi, Cubieboard, or some other cheap/little dev board as a USB host for one or both RTK receivers.
That’s a lot of overhead/added expense if all you want to do is relay it to an XBee or something…. Sure, one side of the NS-RAW pair might be on a higher powered unit like a raspi, but either the rover or base station just needs to be radio’d back to the other of the pair, and having a serial out to a radio would be much simpler and cost effective…
You need a good antenna if you want centimeter level precision. This will cost a few hundred euro’s if bought new (but look on eBay).
Professional GPS antennas are indeed very expensive. But Michele Bavaro says that a cheap patch antenna with a 20cm ground plane (piece of metal) is often good enough:
http://www.onetalent-gnss.com/faq
Also if you read the comments, that processor isn’t capable of running rtk lib, so you need to ship the data elsewhere for processing
NS-RAW functions as a GPS measurement engine, sending out measurement data to a host computer running RTKLIB software. RTKLIB takes correction data from some reference station and measurement data from NS-RAW, compute NS-RAW’s antenna location, resolving carrier cycle ambiguity to determine position to fraction of a wavelength; i.e. fraction of 19cm = 3e10/1575.42MHz. Conventional GPS receiver determines position to fraction of a chip (1023 chip in 1millisecond); fraction of 300 meters, so it’s about 3m accuracy.
Figuratively speaking, Piksi integrates GPS measurement sensor + RTK software + RTK computing hardware + wireless transmission into a box, justifying the price. One can also purchase a $179 LEA-6T GPS module to use with RTKLIB though.
Looks cool, although not totally open-source. “the GPS / Beidou / GLONASS navigation kernel library integrated will remain in library file format.”
sounds too good to be true
I want to see a dash mount GPS unit for vehicles which receives signals from all three systems.
If this system is capable of sub-decimetre precision, it could be a game changer.
Article is misleading. The Navspark has 3 variants available, GPS only, GPS+GLONASS *OR* GPS+Beidou. It does not do all 3.
Literally just the other day I was searching out receivers that were capable of both GPS and GLONASS…
AWESOME! Thank you for reading my mind!
I already backed this and I’m really looking forward to this board. Hopefully they will get the requied funds. I chose GPS+GLONASS.
The hardware is a Venus 822 dev board.
So which other positioning system is more useful, Glonass or Beidou? From the early bird sell outs it looks like it is GLONASS. Could that be due to more satellites being available?
Since all these systems seem to work in the same frequency band (1.5GHz?), will bad reception in one system coincide with bad reception on the alternative system?
Looking at wikipedia, the Beidou system only has coverage for China and ‘near’ neighbours http://en.wikipedia.org/wiki/File:Beidou-coverage.png
GPS and GLONASS offer global coverage, again looking around, GPS is better around the equator and GLONASS is better at the extremities. Combining GPS + GLONASS seems the best option to me.
Great, I bought a pair of GPS/GLONASS boards before I noticed the NS-RAW option…now I might have to order another pair! Oh well, I guess it will help push them closer to funding. They only need $637 as of now…
Assume someone were to purchase two of these NS-RAW boards with the included active antenna, and possibly adds the previously mentioned 20cm ground plane. Also assume no specific RF/GPS/RTKLIB expertise, only general computer/electronic/MCU skills.
How likely is it that they’ll be able to easily achieve near decimeter precision, outdoors or in the average residence, in the southern US?
The reason I’m asking is because I spent a few minutes searching for reports of using RTKLIB for DGPS. Though it seems to be theoretically possible with several existing GPS modules, I found no one who actually obtained a significant improvement over regular GPS.
Perhapst this would suggest success: http://michelebavaro.blogspot.com/2012/02/low-cost-rtk-performance-round-up.html
Thanks. But while low-cost GPS modules were used, they appear to have been used with a high end antenna – priced between $1,400 and $2,000! So I can’t consider this a reasonable predictor of success with low cost hardware.
Right now I’m using RTKLIB with a single ublox LEA-6T module and a $6 ebay puck antenna on a groundplane. My PPP-Static solution is consistent from day to day to better than 10 cm horizontally and just over 10 cm vertically, but that’s using combined fwd/back filtering and a 24-hour observation period. I have not tried any RTK / DGPS.
Does this device have the speed/altitude restrictions that other gps receivers have? If not, maybe it would be a good thing for model rocketry.
The “Ultimate GPS” (MTK3339 chipset) that Adafruit sells has been tested to work above 27km altitude. The only limit I know of is the US ITAR GPS limit of 18 km and 515 m/s, so presumably that GPS has no altitude limit at all.
Standard COCOM limit still applies to NavSpark. That is it’ll work correctly if threshold of (speed < 515m/sec) and (altitude < 18km) are not both exceeded. It'll still work if either one is exceeded. Such limit imposed on commercial GPS receivers is to avoid being used in missiles, a limitation set by software, not inherent limitation of the hardware.
i’d very much see a microSD slot on that device to have (nearly) unlimited storage space and easy data transfer without any drivers involved.
What is the power supply on this device?
Oliver Huang said today a uSD slot will not fit (due to existing components top & bottom), but it could of course be added with a 2nd board stacked on top. The NavSpark board runs from USB +5V although the GPS receiver (Skyworks SE4150L) and the baseband processor (LEON SPARC v8) run on +3.3 V so I assume it could also work from +3.3V external.
In addition, I’m told the NS-RAW version is *not* user programmable. The programmable ones do not have raw output.
For easier reference I’m assembled an (unofficial) “NavSpark FAQ” which is a collection of the info now spread out through the comments section of the Indiegogo page. http://pastebin.com/1f5iXwiT
Mine has a fix!
can it be used on a DJI Phantom Vision +?