I am working on a university Smart Tire research project. We collect acceleration and strain data from sensors inside vehicle tires using an NI USB-6356 DAQ while the vehicle is driving.
We need a GNSS receiver to record the vehicle’s location and provide a timing signal so the sensor measurements can be matched with the correct time and road location.
Would a lower-cost SparkFun GNSS receiver be sufficient for this application, or would you recommend the ZED-F9T?
I would also appreciate clarification on:
whether correction data is needed for timing accuracy, positioning accuracy, or both
how the 1 PPS signal would connect to the NI USB-6356
how latitude, longitude, and timestamp data are transferred to the laptop
which antenna and cable would be required
The F9T does not use/require correction data, it provides super-accurate (~5 ns) timing output by tracking multiple satellite constellations at once on its own. It’s kind of a special-use GPS device
I haven’t used a NI USB 6356 but the PPS output would go to an input that then registers its timing
You create logs with the GNSS/GPS device(s), the GIS software you use will determine the logging parameters. You can also add a logger like the Openlog Artemis that automatically detects and logs Lat/Long, Altitude, Velocity, SIV, Time, and Date
F9T is accurate to 2m, which might not be enough for your application…we do sell the F9R that might work a bit better, as it provides heading and dead-reckoning as well (this is it in a kit with an antenna included, here’s the hookup guide that explains its use a bit. Accuracy ~1-1.5m without RTK/corrections, ~1cm + 1 ppm with, but the main thing is the vehicle attitude fusion that should make your project go more smoothly). If the un-corrected accuracy is fine you’re good to go, if you want to add a correction source I recommend using this bluetooth transceiver (it’s a newer version compared to the one in the guide but functionally the same) and providing correction with a phone in the car via BT.
How accurate do you need your time stamps to be? Even our low-cost GNSS receivers can be very accurate to about 60ns. If you need better accuracy than that, the F9P will get down to around 50ns, the F9T down to around 5ns. Note for the F9x to obtain this higher grade timing, you have to provide the device with corrections (~$15/month or if you have a base setup near by, that works as well).
whether correction data is needed for timing accuracy, positioning accuracy, or both
No if you’re using a simple GNSS receiver. Yes, for the extreme accuracy of the ZED-F9T.
how latitude, longitude, and timestamp data are transferred to the laptop
Any GNSS receiver, including ZED-F9x output a serial stream called NMEA. I recommend you start with the SAM-M8Q (it’s cheap and reasonably accurate), plug it into your laptop and walk outside. From there, you can learn how to parse out the lat/long/time from the serial stream.
TBH the application probably doesn’t need better than 1us accuracy, and there are probably delays in cables, interface and buffering into the 100’s ns
The uBlox M8 should be better than 21ns RMS, and it does report the quantization error via UBX-TIM-TP for the 48 MHz timing clock. Not sure I can recommend the SAM-M8Q due to it’s very limited antenna implementation. You’d do much better with an (external) antenna you can place on the top roof surface of the vehicle.
The ZED-F9 series all use a 128 MHz timing clock, but only the F9T reports the quantization error in the edge placement. The F9T also assumes a static dynamic, for a vehicle you’d need to change the dynamics setting. I think they all use the same spec TCXO, and the timing inside the receivers is materially very similar. The F9R, as part of sensor fusion, is aligning the timing and motion, via measurements from the IMU, and the IMU data is measured at 100 Hz. Reported positions, interpolated from RTK and IMU can be provide at 30-50 Hz.
I’d anticipate any system you build would require the ability to post-process and align measurements from different sub-systems and sensors.
Do any other sensors have TIMEPULSE inputs, or TIMEMARK (EVENT) outputs?
To get sub-metre you’re going to need to be using RTK methods.