@samlogan87 here's a guide to how I do tuning for closed loop mode:
Use a custom tuning move in the closed loop plugin. I use this or similar: G91 G1 H2 X40 F15000 G1 H2 X-40
Set data collection to about 1000 samples/sec. 500 samples will cover the above move.
Make sure that the print head is in a suitable place. The move I use returns to the original position so that I can run it repeatedly.
Adjust the V and A terms to get the P term closer to zero on average. This applies to full closed loop mode only, not to assisted open loop mode.
Here's a plot from the plugin with A0 V0 on my test system:
ac6cbdea-313b-422c-b98e-3a90e9cb69ed-image.png
The PID control signal is almost entirely made up of the P term (the only other term present is D), which of course means there has to be a significant error present to generate the control signal.
Here's a plot from the plugin with A200000 V700:
1ab00dee-0858-45e4-9dce-d5c0bbd12175-image.png
You can see that the A and V terms are doing most of the work, with the P term doing the fine tuning. The peak error (difference between measured and target motor steps) has reduced from about 3 to about 1 full step. To see the error more clearly, enable the display of Current Error and disable the display of Measured and Target Motor Steps so that the scale expands.
You can use the D term to supress oscillations in the P term, but if you make D too high then the motor noise may become unpleasant. In both plots I had P40 D0.1. Acceleration in M201 was set to 5000. I get similar results if I increase acceleration to 10000 but then the mechanics sounds clunky.