KD - The WSJT-X FreqCal process should reliably get you very close unless your rig is drifting badly. You might try FreqCal at cold start, after 1 hour on receive, and after some high-duty-cycle transmitting to get your rig hot, and see how much they differ.
You can use some abbreviated version of the FreqCal process, maybe even just looking at WWV at the highest frequency you can hear well, as a sanity check. I have a relatively new TS-590SG with TXCO and can clearly see the temperature drift from cold start this way, but it's not bad enough to produce the biggest errors you describe. I believe you can safely assume that the error scales with frequency, so if WWV on 15 MHz appears off by 5 Hz, your frequency error on 2 m will be around 50 Hz.
Just let me know if you want further details.
John AA5PR