I also had a bad AVX, which I sent back and got a new replacement (haven't tested it yet).
But you had several issues with your PhD2 setup which could account for a fair share of the problems (the pixel size for one).
I can well understand your not wanting to fiddle with the mount (I got a Mach1 exactly because I got sick and tired of fiddling). But my problems were very different, and unfixable. Your problems look fixable--your graphs look normal, it's just that the swings are too high.
So I'd say, give it one more try, and this time be methodical about it.
0) Polar align well, and balance well (to start--see also below).
1) Put in the correct pixel size.
2) Use ASCOM rather than ST-4. Calibrate on a star near Meridian and Equator (a bit north of equator). That calibration will be good sky-wide (if using ASCOM), don't recalibrate.
3) Use a 2" refresh interval in PhD2.
4) Use 75% aggressiveness.
5) Run guide assistant and set the MinMo it tells you, but *no more than 0.15*.
6) Set MaxRa and MaxDec to 1000.
7) Pick a target somewhere between 30 and 70 degrees in elevation.
8) Mount should be slightly east-heavy, and *very slightly* camera-heavy. Emphasis on slightly, and why you have to have balanced well.
9) You may want to only guide in one direction in DEC, but if so, you'll need to switch to Auto as you approach the Meridian and to the other direction once past it (or if on a target on the other side).
10) If you are on concrete, *don't step around your mount or do it lightly*.
Do all those things. Give it some time. If you *must* change a setting, change only one, then restore it before changing another (else you can never tell what works and what doesn't).
If it is better, consider also doing PEC (AVX can load and replay PEC natively, it's not a hard process at all).
If it's not, send it back for a replacement or repair.