Well, the boards are back and the numbers don't lie. A pair of 1uF caps across the motor terminals is not as good as an X2Y capacitor. It's over twice as good. And it replaces a $0.60, hard-to-get capacitor, with $0.16 of easier-to-get parts. Because of the voltage rating, (50V), they are not pure jelly bean, but definitely in the Jelly Belly variety.
In short, the person that does this professionally knows exactly what he is talking about, and I, fell for magic beans.
Thanks for the coaching, Tim!! I owe you a couple beers, at least.
Note, this result also the outcome of wrapping copper foil tape around the exposed areas of the motor and grounding the PCB ground plane to the case of the motor, but its virtually impossible to get a good noise reduction on a brushed motor without filtering at the terminals. I included two control cases below showing the noise with just copper foil shielding, shielding and an unpopulated PCB, and a populated PCB with no foil shielding. There really is only *one* solution for the these cheap, brushed motors: a PCB with a filter in conjunction with shielding to cover the holes in the motor and ground the PCB ground plane to the motor's case.
These motors are improperly designed in several ways and almost seem intended to be spark gap radios.
Unfiltered output of a 24VDC brushed motor:
Filtered output using the reference design, 0.4uF, Johannsen X2Y capacitor:
Filtered output using 2X 1uF MLCC's across each motor terminal:
First control sample. This is a motor with an unpopulated PCB but grounded to the case as above with copper foil tape:
Second control sample. This is a motor with the 2X 1uF MLCCs across each terminal, but with no Cu foil tape. Once you put in the filter, taping over all the holes on the motor and grounding the PCB to the case are good for another 50mV of noise reduction (in this case).