Short answer: land every shield 360° within 1 in [25 mm] of where the cable enters. Clamp the braid straight onto a grounded backplate, or pass it through an EMC gland in a metal wall. A shield carried through a connector pin and a jumper is a pigtail of about 25 nH, which is 1.6 Ω at 10 MHz, roughly 300 times what a metal clamp gives you. The reactance sits exactly where drive and switching noise lives, so the pin route deletes your shield above a few megahertz. You can carry a shield through a connector only when the metal shell grips the braid all the way round. Slow discrete wiring can live with a short tail; a drive motor cable lands its braid at both ends.
Where the shield current has to go
A braided shield stops interference by catching it on the outside of the cable and walking it to ground along a path with almost no impedance. The current that does the work is high-frequency, and high-frequency current treats any length of wire as an inductor. A straight run of 22 AWG wire carries about 1 nH of inductance per millimetre of length. The figure comes from the straight-wire inductance formula, and wire gauge moves it only slightly.
One number settles most shield arguments you will meet at the bench. Coupled noise on shielded cables climbs through a band reaching about 25 MHz, the range an IEEE EMC presentation on inductive shield coupling highlights. In that band a tail of that length stands in the path with ohms of reactance, while a braid pinched under a metal clamp measures milliohms.
A foil shield changes the hardware but not the rule. Its drain wire is a pigtail by construction, so land the foil itself under a clamp ring where the entry hardware allows it. For which conductor lands on which terminal at the entry, the enclosure-entry bonding article sorts the shield, drain and protective-earth side out.
The numbers behind the pigtail
The case is a drive-controlled axis with a shielded control cable entering the enclosure. The shield lands on a connector pin, then hops to a ground stud through the length of tail a pin connection always leaves you with. The inputs:
- Tail inductance: about 25 nH, from the straight-wire inductance formula for 25 mm of 22 AWG wire.
- Noise band: interference up to about 25 MHz, the coupled-noise range the IEEE EMC presentation shows for shields.
- The rule: a wire this short behaves as a pure inductor, so its impedance to shield current is 2π times frequency times inductance.
The arithmetic, written out:
- At 1 MHz: 2π × 1,000,000 Hz × 25 nH = 0.16 Ω
- At 10 MHz: 2π × 10,000,000 Hz × 25 nH = 1.6 Ω
- At 25 MHz: 2π × 25,000,000 Hz × 25 nH = 3.9 Ω
A 360° clamp path measures milliohms, so at 10 MHz the pin route runs roughly 300 times worse, before you count the pin contact's own resistance. The same arithmetic extended to the other tails you might be tempted to make:
| Tail from braid to ground | Inductance | At 1 MHz | At 10 MHz | At 25 MHz |
|---|---|---|---|---|
| 0.2 in [5 mm] clamp tail | 5 nH | 0.03 Ω | 0.31 Ω | 0.79 Ω |
| 1 in pin jumper | 25 nH | 0.16 Ω | 1.6 Ω | 3.9 Ω |
| 4 in jumper to a ground bar | 100 nH | 0.63 Ω | 6.3 Ω | 15.7 Ω |
The decision comes straight off the table. If the fastest noise the shield must stop sits below 1 MHz — relay logic and dry contacts well away from any drive — a stud tail of that length costs 0.16 Ω and works. Drive edges and fast serial links put energy into the cable from about 1 MHz up past 25 MHz; there the pin route is closed, and even a clamp tail stays under the 5 mm limit from the table.
The three terminations that work
Three landings hold the shield through the whole band; everything else is a variation on the pigtail.
Backplate clamp. Fan the braid out under a saddle clamp and pinch it flat against the plate: 360° contact, a tail under 5 mm, and it works in any enclosure, plastic included. Keep the plate surface under the clamp bare metal.
EMC gland. Spring fingers inside the gland body press the braid as it passes through the wall: 360° contact with zero tail, but the fingers need a metal wall to work through. Order the insert sized to the cable OD so the fingers actually load the braid.
Connector shell. A circular-connector backshell or a D-sub hood that clamps the braid as it enters does the same job, and shielded M12 and RJ45 connectors reach the shell internally. The shield ends at the shell and never becomes a wire.
The versions that fail are the pin jumper at 1.6 Ω, the braid folded back and taped, and the drain wire run to a distant terminal bar at 6.3 Ω. Each one passes a continuity check at the bench and still hands you a noisy axis. When a molded connector has to pass through the wall, getting a connector through a cable gland without leaks covers the hole sizing and the seal around it.
One end or both ends
Drive motor cable takes 360° termination at both ends. The high-frequency common-mode current a drive feeds into the motor cable has to come back through the shield, which is why drive-cable maker guidance says terminate both ends. Left open at one end, the braid becomes an antenna for exactly that current. For the drive-side hardware, the VFD motor cable entry article compares the locking connector with conduit.
Low-level analog loops are the classic exception. Land the shield direct at one end and through a 10-100 nF ceramic capacitor at the other. The DC ground loop stays broken so the signal does not hum, while the RF path holds.
Star point wiring keeps the rest honest. All the clamps land on one bonded plate close to the entries, and shields never daisy-chain through terminal bars. The enclosure grounding and bonding guide lays out the star point and what bonds to what.
What is tested and what is not
IEC 60529 defines the IP code, and its tests are dust and water ingress. Shielding appears nowhere in that standard, so an IP66 seal tells you nothing about how the gland behaves as a shield. An EMC gland carries two separate claims: the spring contact for shielding and the seal for water, and the finished assembly keeps the lowest-rated part. A plain gland with no spring contact shields nothing at all, whatever its sealing claim; what watertight and gas-tight gland ratings actually prove separates the two. Shielding claims on EMC glands are quoted as screening attenuation across a frequency range, not as a single number. Enclosure and gland ratings are manufacturer ratings. Ours are rated NEMA 4X / IP66 by the manufacturer; not UL listed.
Plastic box or metal box
A plastic wall gives the gland spring fingers nothing to conduct through, so the 360° point moves inside the box onto the bonded backplate described above. The shield stops at the plate, and the signals carry on through short unshielded leads kept apart from the power wiring. For plate material and standoff spacing, the enclosure backplate material and mounting article covers what to use.
A metal wall lets you terminate at the entry itself, with EMC glands or clamps, and you group the holes along one edge so the clamped shields share a short earth path. The wall also shields the components inside: the article on ambient RF fields inside a shielded enclosure puts limits on how much survives the seams and the holes. Whether the box itself should be metal for low-level signals is a separate trade; the grounded vs floating metal enclosure article weighs it.
What to put on the order
- One row of gland or clamp holes along a single entry edge, each position dimensioned on the drawing, holes sized to the gland thread.
- A backplate bonded to the supply earth at your star point, with bare metal under every clamp.
- Saddle clamps and EMC gland inserts sized to each cable's outside diameter.
- A sealing plug for every spare gland, so an open entry never sits unsealed.
Isolate the panel and verify absence of voltage before you cut a hole or fit clamps. Then check your cables against this table:
| Cable | Where the 360° lands | Tail limit |
|---|---|---|
| Drive motor cable | gland or clamp at both ends | under 0.2 in [5 mm] |
| Encoder, stepper control, fast serial | backplate clamp at entry, connector shell at the device | under 0.2 in [5 mm] |
| Analog sensor loop | direct clamp at one end, 10-100 nF capacitor at the other | under 0.2 in [5 mm] |
| Slow discrete I/O away from drives | stud tail acceptable | up to the entry-to-plate distance the intro allows |
We cut the entry row, the clamp holes and any other opening to your drawing in 304 or 316L stainless. Order a fully custom stainless body with factory cutouts or a custom box built to your own dimensions; it is delivered in 10 business days from drawing approval.
Catalog dimensions to compare
Compare these catalog dimensions with your component layout. Open each enclosure for its rating, material and usable space, then check those specifications against your application.
| Enclosure | Material | Outside H x W x D / display | Rating |
|---|---|---|---|
| 12 × 10 × 6 in | 304 stainless | 12 x 10 x 6 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 16 × 14 × 8 in | 304 stainless | 16 x 14 x 8 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 20 × 16 × 8 in | 304 stainless | 20 x 16 x 8 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 60 × 36 × 16 in | 304 stainless | 60 x 36 x 16 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
Questions
Can I carry a cable shield through a connector pin?
Electrically it conducts, but the pin and its jumper are a pigtail: about 25 nH, which is 1.6 Ω at 10 MHz. Above a few megahertz the shield stops doing anything useful. Keep pin connections for slow discrete signals; anything near a drive or carrying fast data needs a metal shell that grips the braid all the way round, or a gland and a backplate clamp.
How long can a shield tail be before it hurts?
Keep a braid tail under 0.2 in [5 mm] when it ends in a clamp. A 1 in tail is already 1.6 Ω at 10 MHz, and a 4 in jumper to a ground bar reaches 6.3 Ω, which blocks the shield across the whole band where drive noise lives.
Should a cable shield be grounded at one end or both ends?
Drive motor cable takes 360° termination at both ends, so the high-frequency common-mode current returns through the braid. Low-level analog loops take one end direct and the other through a 10-100 nF ceramic capacitor: the DC ground loop stays broken while the RF path holds.
What changes for shield termination in a plastic enclosure?
The wall cannot host the termination, so land each braid on a metal backplate bonded to the supply earth. The shield stops at the plate; from there the signals continue on short, unshielded leads kept away from power wiring.
Do short drive cables still need the shield terminated?
Yes. A 10 ft [3 m] motor cable still carries the drive's switching edges, and two clamps cost minutes to fit. The distance that adds work is tens of feet: past about 50 ft [15 m] the motor end needs its own 360° landing on a bonded junction box, not a drain wire to a terminal.
