SparkFun Forums

[StuckintheCity]

Hi, I'm a new visitor to this board, which I found because I'm trying to do some research for a product prototype that would require GPS integration. Seemed like there's some good info flowing around here, so I thought I'd raise my hand with a question. The product I'm looking to build needs gps capability, but only in a very limited way. Its basically a remote weather station that someone will set out in the woods. It only needs to look it's location up once each time it's set, and it will remain in the same location for many hours or even many days at a time. Most of the gps chips I've been reading about are touted for their accuracy, which is good (pinpoint accuracy is important for this product). But that accuracy is generally framed in terms of accuracy while moving, which is unnecessary for this product, and therefore it seems like advanced chips would be unnecessarily costly. Does anyone have a recommendation for the least expensive GPS chip set needed to take an accurate location reading once (taking as long as it needs to to do so)? Note that the product would often be located in forests, so it should pickup eventually under tree cover. Thanks in advance for any help!

[mbavaro77]

It is not true that GPS receivers are only tuned for dynamic scenario: they often have different settings of their Kalman filter depending on the navigation mode, e.g. static, pedestrian, automotive, marine, flight, etc.

IMHO the MCU in your weather station should monitor when a position is achieved with a required level of DOP (TDOP, GDOP, HDOP, VDOP?), register that position, and then simply shut down the GPS for example driving high the gate of a P-mosfet switch.
Sometimes receivers have a GPIO input working as interrupt generator to wake them up from deep sleep up for as long as a fix is achieved: ON/OFF pin on GSD4e (e.g. GP-2106), INT pin on uBlox, etc..

The cheapest and lowest power receivers are arguably the CSR GSD4e (for 1.8V systems) and the Mediatek MTK3339 (for 3.3V systems).

Having said that, I would ask one question. Does the weather station have none, inbound, and/or outbound communication?

case NONE
To my knowledge, the best receiver to achieve high accuracy without external assistance is the uBlox NEO-6P. In fact, PPP is perfectly suited for quasi-static applications and will achieve <1m accuracy in about 10-15 minutes. Glonass enabled receivers -like the NVS NV08C, the Skytraq S2525G2F and the STM ST8088 (fastrax IT600 amongst others) can instead benefit of an additional constellation which can make a difference achieving a fix in an environment where navigation signals are significantly attenuated.

case IN
In this case one could benefit of assistance (A-GPS) not only to speed up the acquisition process, but also to provide more accurate Space Vehicle (SV) ephemeris/clock information to the receiver in order to minimise the contributes to the UERE coming from the space segment. But being under foliage the receiver will likely suffer from attenuation and multipath, which usually are worse anyway. So one should choose a receiver accepting external assistance (most do), RTCM input capability, and perhaps also Glonass support (to increase ranging signals availability).

case OUT
In this case one could benefit of snapshot positioning technology to minimise the power consumption, but I suppose that it's not a major driver in this case. A raw measurement enabled receiver or a GNSS digitiser like the MAX2769 or the SE4120L would be needed. This solution has significantly higher complexity... let us know if you want to know more about it.


Hope this helps,
Michele

[StuckintheCity]

Michelle, this is hugely helpful. Thank you!

There are two scenarios I'm considering.

First is the "Case None" scenario. In this scenario a user will set the device in the woods, use it to record conditions (specifically wind speed and direction), then retrieve the device and download this information into a PC/laptop (via USB?), which would then upload the data to a remote server for processing.

Second is the "Case Out" scenario. In this scenario the device would send geo-tagged wind speed and wind direction data to the remote server via some sort of uplink, on a regular schedule (say, every 5 minutes). This scenario would be ideal, but I'm concerned it would make the device too costly for the market I'm looking to target. As a complete neophyte in the hardware arena (I'm more concept than execution) I'd love to know more of your thoughts on the hardware components needed for this scenario, and also any suggestions about the lowest-cost ways to achieve that uplink in remote locations.

Again, thanks for taking the time to craft such a detailed response.

Nate

[mbavaro77]

Hi,

This looks more like an integration task.. with little to do with GPS.
Perhaps I am not the most qualified to answer.

I would say that for a standalone solution you need to control the weather station with a simple MCU (ARM7 or Cortex-M3 will do the job) which wakes up periodically to store the data on the flash card and then puts everything to sleep again. I could suggest a STM32F102 or a LPC134x to accomplish the task. The BOM should be around 80 USD including the GPS.

For a connected solution - if a simple cellular modem works - I would suggest to use something like a Telit GL865 or equivalent (Sparkfun has nice breakout boards). The module can be programmed with python to do the data collection and go to sleep again. Some cellular modems also embed GPS so you can save some money there, but anyway this case will have a BOM twice more expensive than the above.

Of course a proper answer needs a thorough requirements definition... I don't think mine will bring you very far but I gave it a shot.

Best wishes and good luck with your project!

Michele

[calbertson]

Michelle, this is hugely helpful. Thank you!

There are two scenarios I'm considering.

First is the "Case None" scenario. In this scenario a user will set the device in the woods, use it to record conditions (specifically wind speed and direction), then retrieve the device and download this information into a PC/laptop (via USB?), which would then upload the data to a remote server for processing.

Second is the "Case Out" scenario. In this scenario the device would send geo-tagged wind speed and wind direction data to the remote server via some sort of uplink, on a regular schedule (say, every 5 minutes). This scenario would be ideal, but I'm concerned it would make the device too costly for the market I'm looking to target. As a complete neophyte in the hardware arena (I'm more concept than execution) I'd love to know more of your thoughts on the hardware components needed for this scenario, and also any suggestions about the lowest-cost ways to achieve that uplink in remote locations.

Again, thanks for taking the time to craft such a detailed response.

Nate

Just about any GPS will work. Put because the unit is stationary you can have as part of the setup procedure that the user plug in a hand held GPS and let the GPS send a bit of data. Most every GPS can send NMEA data and sends it's loation once per second. So you plug in the GPS and wait for the green "got location" LED then unplug.

This will save you some money as the hard part about a GPS in the forest is trees. You would need a god antenna. using a hand held GPS also allows the weather sensor to be located in a place where GPS could never work. With hand held the user can walk 20 feet away, get a fix then walk back to the sensor and download the fix. You might also have a keypad or some other way to manually key in the lat. and long and not even use an plug in connection.