I had a lot of questions and concerns about the GPS battery, and tried to get answers, refine theories and get a better understanding of what it does.
I downloaded all the info on the uBlox website concerning the NEO-6Q GPS module
after pouring through a lot of info that I didn't completely understand, I started putting together some questions and posed them to someone familiar with the engineering aspects of the module.
First off, the NEO 6q module was a good choice to base the GPS board around . It uses a TCXO (Temperature Compensated Crystal Oven) and BBRAM (Battery Backed RAM) to enhance time base accuracy and speed up TTFF (Time To First Fix).
I was concerned that the TCXO was draining the backup battery, and found that the battery only backed the RAM and the TCXO was controlled from the main power supply (lipo battery).
I was also concerned that the battery was too small in terms of reserve current to keep the last GPS fix in RAM. I was Informed that I could compare the battery specs to the current requirements of the module to determine suitability.
I was concerned that a low battery would cause slow reacquisition of satellites in the event of a GPS glitch. The specs for the module show that battery backup is optional. If not used, the contacts were to be bridged. If used, the battery is effective above the minimum 1.8 VDC requirement. The battery is a 3 volt battery of undetermined composition. I don't know if it is supposed to be rechargeable. I found that the presence of resistance across the BBRAM terminals of the GPS module could cause deterioration of the module's operation.
I questioned a DJI repair tech, and found they do not check the GPS backup battery voltage as a function of repair.
So, I still have questions, but found some of my original concerns weren't valid.
The NEO-6Q, with a working backup battery has a 1 second TTFF; very good. This is due to correct almanac data and last fix already being present when the Phantom is powered up. But, what if the backup battery is below the 1.8 volt minimum and is seen as resistance? If the Phantom is in flight, the BBRAM is bypassed; the module feeds off the main battery. So, what could be the problem of having a low battery? From what I have been able to guess; none - other than slower TTFF due to the possibility of stored GPS data becoming corrupted between uses. That would trigger a new download of almanac data, when a flight battery is connected and the GPS acquisition routine searches for new data.
I don't know if any of the above is useful or not.
I will throw another variable in the mix....
Hovering at the same motor speed with no wind could cause the vibrations of the props to start affecting other pieces of the Phantom. Sympathetic Resonance. Changing speed, small maneuvers, etc. tend to break the consistency of inherit vibrations.
Under optimum sat lock, there is still a 2.5 meter horizontal margin of error. That would be an 8 foot square box to hover within and still be as good as it is expected to get.
I have no answers here; just trying to share a little to think on.