Defective z-axis or PID needs calibration?

[ednisley]

short of running autotune again

IIRC, you tuned the PID coefficients without the fan, then turned the fan on during the print. Autotune sets the control coefficients for best heat control under the conditions it sees, so you must do the tuning with the fans running if that's how you'll be printing.

Although better tuning may get you closer to the goal, remember you can see surface finish effects you cannot measure. Simple things like a vertically repeating infill pattern (changing the spacing of the perimeter thread), periodic filament diameter variations (due to control loop hunting on their extruder!), a slightly eccentric drive gear (tiny speed variations every 38 mm of filament), or a flaw on the filament spool ID (snagging slightly on each revolution) can produce barely visible changes in the extruded thread.

You're fighting the process, which generally ends in heartache & confusion …

[zemlin]

You're fighting the process, which generally ends in heartache & confusion …

which was going to be part of my follow-up post.
If you're looking for perfection, you've chosen the wrong process for your 3D printing. There are too many variables that are simply out of your control. IMHO, you are, to an extent, on a witch hunt, lighting your test parts so they look as bad as possible.

I applaud you for working hard to achieve the best results possible, but with FDM, there are limits.
Here is a photo of the part where I see waves not unlike what you have posted.

To the left is the wavy end. The split is about mid point in the image. The right side of the image still shows surface imperfections, but they are not cyclical.

Here is the same part lit at 45 degrees. Personally I think it's a pretty good looking print. This was run with a .35mm nozzle, and I think the layer thickness was .17mm.

Just for reference, here is another ABS print I ran recently. There are internal features which impact the outer surface which result in the ridges you see on the right side of the image. I lean toward slight over-extrusion to improve perimeter and layer bonding strength and that probably makes things a little worse. This part was run with a .5mm nozzle and .25mm layers.

and here is a close-up of a PETG print. This was run with .3mm layers and a .5mm nozzle. Pretty clean, IMHO - this image was also lit at a low angle to highlight surface irregularities.

If you're looking for a more perfect surface finish, plan to invest in an SLA machine.

[BBID]

I did a little research before running PID autotune the first time. The effect of the cooling fan on the PID values was heavily debated. Also, if the effect was pronounced, that means the PID values would be completely wrong when you switch to a filament that doesn't require cooling (ie. ABS). To test this quickly, I printed a cube with the fan off (closer to PID tune conditions), but the lines were still evident; pics are posted in this thread. I don't think you can run autotune while printing, can you? Because that would give you ideal conditions for the PID tune. I could try to run autotune again with the fan ON to see for myself. I guess that's going to look this this:

*Jog bed up close to extruder around printing position*
M106 S255
M303 E0 S200 C10

I'm going to give this one more shot and then throw in the towel. I have no control over filament diameter changes, eccentric drive gears, filament spool snagging, etc. Running a PID tune, calibrating the extrusion multiplier, and making sure I'm using quality filament with an unobstructed path to the printer are the only things I have really control over, and I've already gone after all those things.

If you're wondering why I care about lines in calibration cube print walls, it's because the other three FDM printers I've used (Flashforge Creator Pro, Ultimaker 3, and Markforged Mark II) all printed cleaner (no PID loop lines) 20mm calibration cubes at faster speeds (80 mm/s). I'm trying to figure out if this is a hardware limitation or there is something wrong with my M2. You can imagine my surprise when I printed a calibration cube on the M2 for the first time and realized I had to run the printer at half my typical print speed (40mm/s vs 80mm/s) to get a decent calibration cube without warping that STILL didn't look as good as the one I printed on the cheaper, chinese-made FFCP with smooth rods. It just didn't make sense to me that the M2, which cost me twice as much, is made in the US, and uses precision linear rails would have much worse cooling, slower usable prints speeds, and lower quality prints than the FFCP. Maybe it seems like I'm nitpicking here, but it's mainly due to my experience with other commercial desktop printers.

[BBID]

@ Zemlin (thanks for those photos!)

IMHO, you are, to an extent, on a witch hunt, lighting your test parts so they look as bad as possible.

Hahah. I used low angle lighting not because I want my prints to look bad, it's because I'm trying to show the issue more clearly. I have no idea if this is a hardware limitation or if this is an issue that needs to be fixed on my M2. What I want to take away from this thread is figuring out if this PID loop "issue" is representative of ALL M2s and not just MY M2. If this is an M2 hardware limitation, so be it.

Thanks for your input guys!