Old Armored Cable and the Ground Path It May Not Provide

Why this matters

A furnace, a range, a disposal or a garage door opener gets connected, the plug-in tester shows the two lamps that mean "correct," and the ticket closes. In a house wired before about 1960 with armored cable, that tester may be reading a ground path that will not carry a fault. The metal case of the thing you just installed is then bonded to seventy-year-old steel strip, and if a conductor ever chafes against it, the case can sit energized while the breaker waits. Nobody gets a callback for this. Somebody gets a shock.

What the cable is, and when the armor counts as a ground

Armored cable is two or more insulated conductors inside a steel (later also aluminum) strip wound in an interlocking spiral. The trade calls all of it BX, a name from an early product that stuck to the category. Today it is sold as Type AC (armored cable) and Type MC (metal-clad cable), and the National Electrical Code treats the two differently.

The armor was designed as protection first: it stops a nail, a staple, a chewing rodent. Whether it is also the equipment grounding conductor, the path that carries fault current back to the panel so the breaker opens, depends entirely on how the cable is built.

• Modern Type AC has a thin aluminum bonding strip running the full length, inside the armor and touching it. The strip shorts across every turn of the spiral. NEC 250.118, as adopted by your jurisdiction, lists Type AC armor as an equipment grounding conductor on the condition that it has that construction, which Article 320 describes.
• Standard interlocked Type MC is not a ground path by its armor alone. It carries a green or bare grounding conductor inside, and that conductor is the ground. Some MC variants are listed with an aluminum grounding conductor in contact with the armor; the listing is what qualifies them, not the metal jacket.
• Early armored cable, the kind in houses wired from the 1920s into the 1950s, has neither. No strip, no grounding conductor. The armor is the only metal path back to the panel.

The bonding strip became part of the listed construction around the late 1950s. Treat any armored cable in a house wired before 1960 as suspect until you have seen the strip with your own eyes. (When a house was built and when it was rewired are different dates; see related: Dating a House by What You Find in It.)

Why a spiral is a poor conductor

Current trying to travel the length of the cable through the armor has two choices. It can follow the strip around every turn of the helix, a path many times longer than the cable itself. Or it can jump from turn to turn where the edges of the convolutions touch. When the cable was new and the steel was clean, those contacts were acceptable. Decades later the touching faces are oxidized, turns have loosened wherever the cable was bent around a joist, and the connector at each box is a setscrew biting into rust.

Three consequences follow, and each one defeats a shortcut a tech might take:

• The resistance is unknown and uneven. One run can test acceptably and the run beside it can be open at a single rusted coupling. You cannot infer one cable from another, or one box from the next box on the same cable.
• A coil opposes fault current more than it opposes a meter. The helix behaves partly like an inductor, so at fault-current levels it offers more opposition than a low-current resistance reading suggests. A good ohmmeter number overstates how well the path will clear a fault.
• Every box is a joint. The path is only as good as the worst connector between your equipment and the panel, and most of those connectors are inside walls you will never open.

What a weak ground path does during a fault

A breaker does not open on a ground fault because it senses ground. It opens because the fault draws enough current. A thermal-magnetic breaker has two regions: a magnetic (instantaneous) region, where it opens in a fraction of a second once current reaches some multiple of its rating, and a thermal region below that, where it can take seconds to minutes. Where the instantaneous region begins is printed on the manufacturer's time-current curve; for common residential breakers it falls somewhere from several to roughly ten times the rating. Read the curve for the breaker in front of you rather than carrying a rule of thumb.

Here is the arithmetic laid out with illustrative values, not measurements from any real house:

15 A breaker, 120 V circuit
assume instantaneous trip at about 10 x rating = 150 A

loop must be at or under: 120 V / 150 A = 0.8 ohm

say old armor + rusted connectors make the loop 2.0 ohm
fault current: 120 V / 2.0 ohm = 60 A
60 A / 15 A = 4 x rating  -> thermal region, slow trip

say one coupling is open
fault current: near 0 A   -> breaker never trips

At 60 amps the breaker opens eventually, but "eventually" can be longer than a person holding a furnace door stays in contact with it, and for that whole time the cabinet sits at a voltage set by how the loop divides. At an open coupling there is no clearing at all: the cabinet holds line voltage until someone touches it and a grounded water pipe together.

This is also why the plug-in receptacle tester misleads. Its lamps light on a few milliamps through a high-resistance circuit. A path that passes a few milliamps can fail completely at sixty amps. "Correct" on that tester means the ground pin connects to something. It says nothing about whether that something will clear a fault.

Telling old from modern at an open box

Safety before anything is opened. Only a person qualified for the work opens a box, with the circuit de-energized at the panel, locked and tagged, and proved dead with a meter checked on a known live source before and after, per NFPA 70E Article 120 (120.5 in the 2021 edition, renumbered in 2024), with work practices at 29 CFR 1910.333(b)(2) for general industry (construction electrical lockout is 29 CFR 1926.417). If nobody on site is qualified, the box stays closed and the call goes to an electrician. Old rubber insulation in these boxes cracks when flexed, so look with a light and do not bend conductors aside to see behind them. Cut armor edges are sharp; wear cut-resistant gloves.

Then read what is there:

The anti-short bushing is the small fiber or plastic sleeve pushed into the cut end of the armor so its sharp edge cannot bite the insulation. Its absence does not prove age, but it tells you how carefully the original install was done, and a missing bushing over brittle insulation is exactly where a conductor eventually touches the armor.

Two traps. A bonding strip that was snipped off flush during the original install is worth no more than no strip, so look for it actually folded back under the connector, not a stub. And seeing modern cable at your box proves nothing about the run upstream: additions and repairs often splice new cable onto old at a box you cannot see.

What a non-electrical trade does with this

Recognition is not diagnosis, and the full rule for legacy materials lives in its own card. For armored cable it comes down to four plain behaviors:

• Do not use old armor as the ground for new equipment. If the equipment needs a grounding conductor and the only path back is unidentified armor, the connection waits for an electrician.
• Do not upgrade a two-prong receptacle on the strength of an armor-fed box. The options for an ungrounded receptacle have their own card.
• Do not re-bend, pull or re-terminate the cable to reach your equipment. The conductor insulation inside is the more fragile part, and moving it is how a sound circuit becomes a faulted one.
• Write what you saw, not a verdict. "Appears to be early armored cable, no bonding strip visible at the furnace junction box" is a record. "Ground is bad" is a diagnosis you did not perform.

The electrician's route is a ground-path test with a loop impedance tester, or simply a new circuit pulled with its own grounding conductor, which on a single appliance circuit is often the quicker decision.

Worked example: one gate, two furnace swaps, opposite answers

The shop's gate, applied by any tech connecting equipment whose metal case must be grounded: before connecting, identify the ground path back to the panel; if that path is armor, it must be identified as Type AC with a bonding strip at your box, or an electrician must have verified it, or the connection waits.

House A, a 1954 ranch. The tech, qualified under the shop's policy for the furnace's line-voltage connection, de-energizes the 15 amp furnace circuit at the panel, locks and tags it, and proves the junction box dead with a meter checked on a known live source before and after. Inside: two cloth-braided conductors, no bonding strip, no anti-short bushing, a rusted setscrew connector. The receptacle tester on a nearby outlet on another old run had shown "correct" an hour earlier, which this rule ignores. Gate result: the path is armor, it is not identified Type AC, nobody has verified it. Stop. The new furnace is set and vented, the line-voltage connection is not made, the circuit stays off and tagged with the shop's number, and the electrician is called.

The failure in House A is not the armor. It is that the armor was discovered on installation day in January. The electrician's earliest slot was the next morning, so the family spent a night on the old furnace's absence. A legacy walk at the quote would have booked the electrician for the same day as the swap; see related: How to Do a Legacy-Materials Walk Before Quoting an Older House.

House B, a 1971 split-level. Same steps, same proving sequence. Inside the box: thermoplastic conductors, an anti-short bushing, and a thin aluminum strip folded back under the connector. Gate result: identified Type AC with a bonding strip at this box. The tech confirms the connector is tight and the box bonded, then connects.

What would flip House B: the strip turns out to be a snipped stub, or the circuit is known to pass through an older section (a remodeled basement, a panel relocated in the 1980s). The record therefore says what was and was not seen: "Type AC with bonding strip at furnace box; panel end and intermediate boxes not inspected."

Putting House B back in service carries its own check. Close and secure the junction box cover before the breaker goes back on, stand to the side of the panel rather than square in front as the breaker closes, run the furnace through a call for heat, then meter from the furnace cabinet to a separately verified ground and confirm no voltage before anyone opens the cabinet door.

How to check you got this right

• The ground path was identified before the connection, not inferred from a tester. If your note cites the plug-in tester as evidence of grounding, the rule was skipped.
• The photo shows the open box with the conductor insulation and the presence or absence of the strip visible. A photo of the closed cover proves nothing a later reader needs.
• The record names what was not inspected. "Panel end not inspected" protects the next tech from assuming you checked it.
• The circuit's state at departure is written down. On and verified, or off and tagged with who to call. A circuit left off with no tag is how the customer, or the next trade, turns it back on.

References

• NFPA 70, National Electrical Code, 250.118, Article 320 (Type AC) and Article 330 (Type MC), in the edition your jurisdiction has adopted.
• NFPA 70E Article 120 (120.5 in the 2021 edition, renumbered in 2024); 29 CFR 1910.333(b)(2) for general industry and 29 CFR 1926.417 for construction electrical work.
• The breaker manufacturer's time-current curve for the breaker installed.
• See related: Recognize, Record, Route: The Rule for Legacy Materials; Two-Prong Outlets and the Ungrounded House; Cloth-Sheathed and Early Plastic Cable.