Author Topic: Affordable Centimeter-Level GPS Accuracy  (Read 20761 times)

0 Members and 1 Guest are viewing this topic.

Offline RickHTopic starter

  • Newbie
  • Posts: 1
Affordable Centimeter-Level GPS Accuracy
« on: January 14, 2014, 08:31:56 am »
   A plug-and-play carrier phase raw measurement output GPS receiver that works with open-source RTKLIB, capable of generating centimeter-level position accuracy at $50 a pair should be interesting.

   Although not a complete stand-alone RTK receiver as the $900 a pair Piksi, but requires running RTKLIB software on another PC, it is still quite impressive that how quickly the centimeter accuracy RTK technology found in high-end high-price survey grade GPS receivers are becoming accessible nowadays.

   http://bit.ly/1dBpits
 

Offline JBeale

  • Frequent Contributor
  • **
  • Posts: 298
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #1 on: January 17, 2014, 05:14:11 am »
In case you don't like hidden URLs, the NavSpark project on indiegogo is here:
http://www.indiegogo.com/projects/navspark-arduino-compatible-with-gps-gnss-receiver

They are using the Skytraq Venus822 GNSS chipset and 100 MHz 32-bit LEON3 Sparc-V8 CPU+FPU.
I really want this project to work and the promised performance to be realized, but I'm curious if anyone here knows how likely this goal is to be achieved?  There is very little on the "affordable" end of the market with raw pseudorange outputs, as far as I know. The u-blox timing grade receivers do this but are rather difficult to source.  Anyway I plunked down my $50 so my fingers are crossed.
« Last Edit: January 17, 2014, 06:52:55 am by JBeale »
 

Offline Rasz

  • Super Contributor
  • ***
  • Posts: 2616
  • Country: 00
    • My random blog.
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #2 on: January 17, 2014, 07:30:34 am »
its chinese/tw = there will be no documentation or source code = good luck
Who logs in to gdm? Not I, said the duck.
My fireplace is on fire, but in all the wrong places.
 

Offline JBeale

  • Frequent Contributor
  • **
  • Posts: 298
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #3 on: January 17, 2014, 05:23:22 pm »
You could be right about the documentation. But some SkyTraq Venus GPS chips are being used in the hobbyist market already, for example: https://www.sparkfun.com/products/11058
« Last Edit: January 17, 2014, 06:01:58 pm by JBeale »
 

Offline JBeale

  • Frequent Contributor
  • **
  • Posts: 298
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #4 on: January 31, 2014, 05:35:26 pm »
If you're interested, I've re-organized some info from their Indiegogo webpage into my own FAQ on the NavSpark board: http://pastebin.com/raw.php?i=1f5iXwiT
 

Offline edavid

  • Super Contributor
  • ***
  • Posts: 3381
  • Country: us
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #5 on: February 04, 2014, 03:57:22 pm »
They don't seem to mention power consumption. I'd like to have centimetre level GPS accuracy but the microcontroller is superfluous. I don't want to learn another micro or use the Arduino IDE, I just want a nice low power module I can interface to other things.

The Venus822-BD datasheet says "Tracking 15mA @ 3.3V".  Is that low power for a GPS receiver?

Also, the "centimetre level GPS accuracy" version (NS-RAW) doesn't allow user programming, and it won't be low power because it requires a powerful external computer to process its output with RTKLIB.
 

Offline JBeale

  • Frequent Contributor
  • **
  • Posts: 298
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #6 on: February 04, 2014, 05:19:43 pm »
The Venus822-BD datasheet says "Tracking 15mA @ 3.3V".  Is that low power for a GPS receiver?
I think that is quite low power for GPS with continuous updates, eg. not sleeping 90% of the time- but that datasheet may be talking about the baseband Venus8 CPU only, and not the full solution including RF front end.
Quote
Oliver Huang at "Electrical Specs" in http://pastebin.com/raw.php?i=1f5iXwiT and also Commonly Asked Questions.pdf
Current consumption for initial full power signal acquisition / fully tracked navigation for NavSpark, NavSpark-GL, NavSpark-BD is roughly 38/30, 45/40, 45/40 mA at 3.3V using on-chip DC/DC switching regulator.
As far as I know, true cm-level GPS solutions are currently in the $1000 (and way up) class. It remains to be proven how well the $50 NS-RAW set does... I put in for one and I'm hoping to be impressed, but my suspicion is that if cm-level happens it will only happen under ideal conditions.  For example I have a ublox LEA-6T with raw output (bare module lists around $180) and after a week of tests, I find it gives me results consistent from day to day of about 20 cm in a static-PPP solution with 24 hours of averaging per data point.  I have an old SiRFStar-II chipset device with raw output (DeLorme Earthmate Blue Logger) and it is about 3x worse in resolution, using the very same antenna through antenna splitter.   On the other hand this is single-receiver PPP which is harder than DGPS with base and rover, but basically I'm just saying, don't expect too much.

That said; low power is meaningful even for NS-RAW, if you put it on a mobile robot, plane, quadcopter... you can use a small transmitter to send the data back to your laptop etc. and still get basically real-time position without having big batteries on your remote device.  Or something like a Raspberry Pi could work if you want the remote running RTK.

EDIT: By the way, I see there are a few websites like http://rtk-pi.com/ and http://tubbyaustin.blogspot.com/2013/06/rtklib-on-raspberry-pi.html which mention "cm-level precision" with cheap receivers but they do not show actual results, or really much detail that I could find.
« Last Edit: February 04, 2014, 05:56:20 pm by JBeale »
 

Offline Sigmoid

  • Frequent Contributor
  • **
  • Posts: 488
  • Country: us
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #7 on: February 04, 2014, 06:58:29 pm »
Quote
NS-RAW is to be used as it is without Arduino programability, due to some technical issue yet to overcome and we prefer not to over-promise.

vs.

Quote
Q: Is there any reason why you can’t configure the Navspark / Narspark-GL /
Navspark-BD to output the same data as the NS-RAW?

A: It’s mainly market segmentation issue, similar to software companies having
different price bundles for a product

I smell a rat. :)
« Last Edit: February 04, 2014, 07:02:10 pm by Sigmoid »
 

Offline ve7xen

  • Super Contributor
  • ***
  • Posts: 1192
  • Country: ca
    • VE7XEN Blog
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #8 on: February 04, 2014, 07:51:06 pm »
Hoping they'll be useful for timing, good timing modules are pretty expensive or old and power hungry / large. Backed. Might almost be enough peripherals onboard to implement a basic GPSDO without too much else.
73 de VE7XEN
He/Him
 

Offline JBeale

  • Frequent Contributor
  • **
  • Posts: 298
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #9 on: February 04, 2014, 07:59:20 pm »
If you read through all the comments on the Indiegogo site, you will note that RAW output was not originally part of the offering, but they added it by customer request. At first try they could not get RAW output working with RTKLIB from VENUS 822 so they were going to use an older VENUS 6 chipset as the backup plan, which they stated upfront.  Later you will note that they did report success with RTKLIB from the Venus 8 family, so after that there was no hardware limitation in place for Arduino programmability, so what happened then?

I don't know but it's easy to imagine there was some company meeting and the marketing-strategy people ruled out programmability for the RAW output version. Logically enough, because if it's just provided as library, you could buy the less expensive basic unit and simply reprogram it to be the (slightly) more expensive RAW one.  The fact that Oliver posted the point about market segmentation on page himself, suggests to me he's not trying to hide what's going on.  I can hardly criticize this strategy when it still ends up the cheapest raw-output GPS I'm aware of.

Compare that with the other vendors in the GPS space who do the same thing, but without actually saying it. In days gone by, the SiRF-Star II GPS chipset had an undocumented feature providing raw pseudorange and carrier phase. This was removed in SiRF-Star III, allegedly for this reason.

With the 10-nsec resolution I was also thinking about GPSDO.  I note they have a 24-bit PWM which when filtered might be useful as a DC control signal as well; so it's possible you might get it to work without needing a separate controller.
« Last Edit: February 04, 2014, 08:10:02 pm by JBeale »
 
The following users thanked this post: cdev

Offline edavid

  • Super Contributor
  • ***
  • Posts: 3381
  • Country: us
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #10 on: February 04, 2014, 08:44:36 pm »
Hoping they'll be useful for timing, good timing modules are pretty expensive or old and power hungry / large. Backed. Might almost be enough peripherals onboard to implement a basic GPSDO without too much else.

I don't think so... Oliver said:

Quote
Regarding GPS with position-hold for 1 satellite precision 1PPS generation, AFAIK only SkyTraq Venus638LPx-T, ublox LEA-4T/5T/6T NEO-6T, Trimble Resolution T, iLotus M12M Timing Oncore, NavSync CW-25 TIM are precision timing mode receivers with such feature. All $40 ~ $60 SirfstarIV, ublox-6Q/7Q GPS shields on hobbyist sites are normal GPS module for navigation/tracking application, not precision timing type and do not have position hold 1PPS generation feature, no different from NavSpark.

Maybe John can translate that, but it doesn't sound promising for GPSDO use.
 

Offline JBeale

  • Frequent Contributor
  • **
  • Posts: 298
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #11 on: February 04, 2014, 08:52:50 pm »
Maybe John can translate that, but it doesn't sound promising for GPSDO use.

I'd say Navspark could be used in a "basic GPSDO" (maybe not a "fancy GPSDO"). Oliver is saying that it's not as good or feature-ful for precision timing as a $200 solution (LEA-6T). However position-hold does not necessarily improve your timing over the long term, it just helps with short-term fluctuations. If your local oscillator is stable enough and you're only using the GPS for long-term drift correction, it may not affect accuracy much, particularly if your antenna has a good view of the sky. 

He also said Navspark does not have TRAIM, which is an integrity monitor that can warn you if the calculated timing error exceeds some limit, but that is just an information feature, it does not by itself change the accuracy.  You could get the same information by measuring instantaneous deviation between the 1PPS from the GPS and from your local disciplined oscillator.

It's just a question of what your goal is, and your budget. It's not a high-end GPS, rather the opposite, but for the price of one LEA-6T you can get quite a few Navspark units and compare their outputs.
« Last Edit: February 04, 2014, 09:16:38 pm by JBeale »
 

Offline ve7xen

  • Super Contributor
  • ***
  • Posts: 1192
  • Country: ca
    • VE7XEN Blog
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #12 on: February 04, 2014, 11:11:11 pm »
It's also programmable. Not sure exactly what that means, but it may be fairly easy to implement TRAIM and position hold in a custom firmware. Affecting how the time solution is calculated is probably not possible... but it'll be fun to play with anyway, I think. Seems a lot more practical than 20 year old Motorola Oncore units.
73 de VE7XEN
He/Him
 

Offline edavid

  • Super Contributor
  • ***
  • Posts: 3381
  • Country: us
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #13 on: February 04, 2014, 11:13:52 pm »
It's also programmable. Not sure exactly what that means, but it may be fairly easy to implement TRAIM and position hold in a custom firmware. Affecting how the time solution is calculated is probably not possible... but it'll be fun to play with anyway, I think. Seems a lot more practical than 20 year old Motorola Oncore units.

Programmable means you can run your own code, but the GPS code will just be a blob, so it can't be modified.
 

Offline amyk

  • Super Contributor
  • ***
  • Posts: 8240
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #14 on: February 05, 2014, 12:12:52 pm »
The Venus822-BD datasheet says "Tracking 15mA @ 3.3V".  Is that low power for a GPS receiver?
I think that is quite low power for GPS with continuous updates, eg. not sleeping 90% of the time- but that datasheet may be talking about the baseband Venus8 CPU only, and not the full solution including RF front end.
For comparison that is the same power consumption as this MT3339 module, but this
Quote
Current consumption for initial full power signal acquisition / fully tracked navigation for NavSpark, NavSpark-GL, NavSpark-BD is roughly 38/30, 45/40, 45/40 mA at 3.3V using on-chip DC/DC switching regulator.
suggests it consumes rather a lot more.
It's also programmable. Not sure exactly what that means, but it may be fairly easy to implement TRAIM and position hold in a custom firmware. Affecting how the time solution is calculated is probably not possible... but it'll be fun to play with anyway, I think. Seems a lot more practical than 20 year old Motorola Oncore units.

Programmable means you can run your own code, but the GPS code will just be a blob, so it can't be modified.
Modified easily, that is... "when there is a will, there is a way." ;)
 

Offline JBeale

  • Frequent Contributor
  • **
  • Posts: 298
Re: Affordable Centimeter-Level GPS Accuracy
« Reply #15 on: February 06, 2014, 09:20:28 pm »
By the way, I thought I'd never heard of the embedded LEON3 SPARC-V8 cpu that NavSpark is using.  But maybe I've been using one without realizing it.

Today I read the description for the (rather expensive) SDK for developing embedded applications on the Roving Networks WiFly (standalone wifi module- now owned by Microchip). It says:  "Based on GCC, GNU Make and the LEON BCC Compiler System" making me think they've got a Sparc-LEON2 or LEON3 in that module, running their wifi and TCP/IP stack.  Just for what it's worth.  http://www.microchip.com/stellent/groups/sitecomm_sg/documents/devicedoc/jp557986.pdf
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf