Fig. 2 (p.18, MIL-STD-461G) is most likely the setup you require:
You don't mention an EUT [metallic] enclosure, or shielded cabling extending from such, but you do mention a line filter, which might be construed as an EUT enclosure boundary for these purposes.
Whatever the case, presumably you would have all applicable enclosures, shields, etc. attached together, to metallic frame members, etc., whatever is appropriate for the end use; and based on that end use, you choose the most appropriate/similar test configuration (i.e. bonded to ground plane).
If not bonded to surrounding metal, that's fine too, and Fig. 3 would be chosen.
Mind that, if there are any other connecting cables from the EUT, those need to be treated in suitable methods not addressed by this figure. Cables might be grounded, looped up, terminated in a CDN, terminated in auxiliary equipment, etc.
Note the visual language used in this figure:
Ground bonds are indicated by foil/sheet angles, joining the metal enclosure of EUT (if applicable) and LISN to ground plane, and plane to test chamber.
In practice, these might simply be copper or aluminum foil tape (specifically, EMI type with conductive adhesive and/or a corrugated texture to ensure contact), ensuring a wide, short, low-inductance join from enclosure to plane. It might also be brackets or braid bolted into place.
The unconnected circles on the LISNs most likely represent the BNC jacks, not shown connected here. They would be panel-mount types, i.e. bonded to the metallic enclosure.
Fig. 6 gives the LISN schematic:
The dotted line implies an enclosure (chassis) boundary. A ground symbol is not shown, but is implied by the bonding in the above diagram. The hoop around the signal wire indicates a coaxial connection, with the shield bonded to the enclosure. This is widely used notation (give or take variations); see these excerpts from ANSI Y32.2-1975 for example:
(p.33)
and
(p.45)
So the BNC cable should already connect to the LISN enclosure or bonding terminal(s) and does not need additional consideration. You can confirm this on inspection of your units, by measurement (the AC impedance will be low), etc.
Regarding the chamber: a sealed test chamber is used; a metallic room with absorption materials on the walls to reduce reflections. The metal walls provide easy ground connection, and serve as reference plane for signals passing through from the outside, or around within. That plane can be extended by bonding additional planes to it, such as the metal table diagrammed above.
Note the meaning of "grounding" here. "Ground" is a regrettably ambiguous word in EE. It means some defined voltage reference. In EMC work, the definition will be a reference plane that measurements are made with respect to. Which plane, and which point on that plane, follows where, in the system, that measurement is taken. For example, the "Power Source" or "Access Panel" connections can be considered ports with respect to the immediately surrounding chamber wall. The cables laid over the table's ground plane, can be measured with respect to the table; at the LISN, or EUT, it might again be measured at the point of enclosure penetration, or on-board, where a connector reaches the PCB ground plane, etc.
Note that voltage cannot be measured unambiguously in a real 3D-fields scenario: it is a path-dependent quantity, and the position and orientation of the measurement loop matters. By using a ground plane, by keeping cables a controlled distance apart (some cm from each other, and from the ground plane), and concerning ourselves with modest frequency ranges (i.e., well under 1GHz), we can minimize the ambiguity due to superposition/interference, reflection, and other inhomogeneity. Finally, we can eliminate it completely, by fixing a definition for our measurement. For example, we can take the shortest path between a point of interest (say, the voltage on a cable wire, or its shield), and the ground plane. (Such a path doesn't prevent path-dependency, it just defines a path, making the dependency moot -- for better or worse.)
Around or above 1GHz, we incur wave effects between cable and plane (i.e. wavelength comparable to height above plane), and conducted emissions/susceptibility (i.e., the signals carried by a LISN) cease to be useful or meaningful. Well below this range, of course, we switch to radiated testing instead (typically 30MHz+), and the LISN remains a representative impedance, even if beyond its calibration range.
Note, LISNs aren't magically ineffective beyond 30 or even 100MHz; they're just less effective at accurate and repeatable measurement, and to more setups (particularly, dimensionally large ones), as frequency goes up. Many LISN designs are still useful beyond their rated frequency range, and this can be useful in precompliance testing for example (where representative change is an acceptable alternative to absolute calibration).
In any case, you can think of a LISN as exactly that type of measuring device: a well-defined, zero-dimensional point, on a cable, with respect to the ground plane, that communicates between fields around the EUT cable, and signal in the BNC cable.
Yet another representation: a LISN is an application of a bias tee.
See also p.177,
A.4.3.5.1 (4.3.5.1) Metallic ground plane.
When the EUT is installed
on a metallic ground plane, the ground plane shall have a surface
resistance no greater than 0.1 milliohms per square. The DC resistance
between metallic ground planes and the shielded enclosure shall be 2.5
milliohms or less. The metallic ground planes shown on Figures 2
through 5 shall be electrically bonded to the floor or wall of the
basic shielded room structure at least once every 1 meter. The
metallic bond straps shall be solid and maintain a five-to-one ratio
or less in length to width. Metallic ground planes used outside a
shielded enclosure shall extend at least 2.5 meters beyond the test
setup boundary in each direction.
(my emphasis). This gives you some idea of what the bonding might look like (but, I don't actually see a definition in this document, which is a little annoying). Like I said, wide foil, sheet or braid is a typical choice.
And p.183,
A.4.3.8.2 (4.3.8.2) Bonding of EUT.
Only the provisions included in
the design of the EUT shall be used to bond units such as equipment
case and mounting bases together, or to the ground plane. When bonding
straps are required, they shall be identical to those specified in the
installation drawings. Bonding of the EUT to the ground plane shall be
verified to be in accordance with the installation drawings or
equipment specification before connecting cables and EMI testing. The
verification process and results shall be recorded in the EMITR.
In other words, as I said above: set it up, more or less, as it would be in the intended application.
The references in §2 may have further definition/explanation of terms and practices. Of them, ANSI C63.14 (American National Standard Dictionary of Electromagnetic Compatibility (EMC) including Electromagnetic Environmental
Effects (E3)) seems most likely.
Other comments:
Note that your brushed motor, itself, most likely needs to be part of the test. Brush noise is notorious.
Most likely you would use one (single-line) LISN each for + and - (or H/N or etc.) power wires. (It sounds like you understand this already; just to confirm that.)
You may need CDNs for other cables.
