If an identical model iPad produces reliable results, this might suggest an issue with the iPad. However, I'm curious to understand why you might substitute a working iPad for an identical model unless the original was itself perhaps faulty.
Otherwise, if you experience difficulties with GNSS/GPS when used in the cockpit, two logical possibilities exist:
1) Physical positioning of the device is blocking GPS signals. Remember, GPS requires a clear “line of sight” view of the sky. You shouldn’t rely upon reflected GPS signals - as these are unreliable and can significantly reduce accuracy of the computed position solution (i.e., Dilution of Precision, DOP, is degraded).
2) Something in the environment is creating an interfering signal. Even if not on the same frequency, interfering signals can swamp the very low-level signals received by your device. As you are doubtless aware, the typical GA cockpit is filled with avionics and other radio devices.
Speaking of personal experience, I suggest that you consider using an independent GNSS receiver positioned on the combing. Bad Elf, Garmin GLO and GNS 3000 are well regarded.