What a Control Signal Type Implies About the Fault

Why this matters

Two identical faults on two identical air handlers produce completely different service calls, because one unit's damper runs on a 2 to 10 volt signal and the other's runs on 0 to 10 volts. On the first, a broken wire announces itself as an impossible value and the alarm points at the wire. On the second, the same broken wire reads as a perfectly legitimate command to close, nobody is alerted, and the complaint arrives weeks later as a comfort problem with no fault history behind it.

The signal type is not a wiring detail. It decides, before anything ever breaks, whether a break will be visible. This card is the field card for reading that, and it is organized as a survey you fill in for a specific unit rather than as theory.

The one property that separates the types

Does the value a broken path produces fall inside the type's live range or outside it?

Every signal path has a null: whatever the receiving end reads when the path is gone. If that null is outside the range of legitimate values, the receiver can tell a break from a reading, and the system can alarm on it. If the null sits inside the legitimate range, the receiver has no way to know, and a break becomes a silent, plausible, wrong number.

Everything else - accuracy, noise immunity, run length - is secondary to that single property when you are diagnosing, because it decides whether the fault will ever be reported at all.

The field card

Read down the last column and the pattern is plain: the types that announce their own faults are the ones whose designers deliberately moved the live range off the null. That is not an accident of the technology, it is the entire reason those ranges look the way they do.

Why the current loop is the odd one out

A two-wire current loop is a series circuit. The same current flows through every element in it, so conductor resistance and any difference in reference potential between the two ends do not change the value that arrives. That immunity is the reason it survives long runs and noisy buildings, and it is also why a break is unambiguous: current in a broken series circuit is zero, and zero is below the live floor.

The immunity has a stated limit and it is worth knowing where. Take a 24 volt loop supply, a transmitter that needs a minimum of 12 volts across itself to operate, and a 250 ohm sense resistor at the receiving end. That leaves 12 volts to be spent on everything else at the worst-case 20 mA, so the total external resistance budget is 12 divided by 0.020, which is 600 ohms. Subtract the 250 ohm sense resistor and 350 ohms remain for wire and any other series device. At 18 AWG copper, about 6.4 ohms per 1000 feet at 68 F and rising roughly 0.4 percent per degree Celsius above that, and counting both conductors of the run, 350 ohms is on the order of 27,000 feet of two-conductor cable. In a building, wire resistance is simply not the constraint.

Those numbers are derived under that specific set of conditions. Change the loop supply to a lower voltage, use a transmitter needing more headroom, or add a second 250 ohm sense resistor for a recorder in series, and the budget shrinks fast. Two 250 ohm receivers in the same loop leave 100 ohms for wire, and now the run length genuinely matters. The rule to carry is not the 27,000 feet, it is that you compute the budget from the supply voltage minus the transmitter's minimum, divided by 20 mA, and then subtract every series device.

Voltage signaling has no equivalent immunity to a shifted reference, because a voltage is only meaningful against the common it is measured to. That failure mode belongs to a sibling card on shared commons and is cited at the end rather than re-derived here.

The three ways a designer buys a live zero

Offset the range off the null. 4 to 20 mA, 2 to 10 V, 3 to 15 psi. The bottom quarter or fifth of the physical range is deliberately spent so nothing legitimate ever lands where a failure lands. It costs resolution and buys detectability, and it is nearly always the right trade on a signal that leaves the enclosure.

Supervise the circuit with resistance. A dry contact alone gives two states and a break looks like one of them. Put a defined resistor at the far end of the line and a second value in the contact branch, and the pair reads four distinguishable states: normal, operated, infinite for a broken conductor, near zero for a shorted pair. The specific values are set by the receiving panel's specification, so read them off the panel rather than assuming a standard number.

Require positive liveness. On a digital point there is no amplitude to be impossible, so detection moves into the time domain: a heartbeat that has to keep arriving, a minimum pulse rate, a timestamp the receiver checks for age. A bus point with none of these holds its last received value forever and looks healthy while the device behind it is dead.

Before you meter any of it

De-energize the line-voltage supply, lock and tag it, and prove dead before opening a panel to land a meter, under 29 CFR 1910.333(b)(2), with the live-dead-live sequence at NFPA 70E-2021, 120.5. Where a signal genuinely can only be read with the circuit live, that is the narrow troubleshooting allowance at 29 CFR 1910.333(a)(1), taken with meter and leads rated for the circuit's measurement category and available fault current plus the shock and arc-flash protection the assessment requires. Switched control voltage at line potential looks exactly like a 24 volt control wire in a panel and is not.

Two habits specific to signals. Reading a current loop in series means breaking it, and the loop goes to zero while your meter is out, so whatever that loop commands will act on a zero for the duration; if the loop drives a final element on a combustion, pressurized or refrigerant-bearing path, or anything serving a relief or protective function, do not break it - use a clamp-on milliamp meter instead. And any resistance measurement is taken on an isolated, de-energized circuit, because an ohmmeter on a live circuit reports a fiction and can be destroyed doing it.

Worked example: the signal survey for one air handler, filled in

One unit, six points, surveyed in about twenty minutes with a print and a meter. The column that matters is the last one.

Four of the six points cannot tell a break from a legitimate value. That is the finding, and it is available before any fault exists. Indistinguishable is not the same as unheard, and the two do not always travel together: of those four, three fail without anybody being told, and one, the fan run status, announces loudly as a fault that does not exist. Both are wrong; only one wastes a trip instead of a season.

Now read the consequences off the survey rather than guessing at them. The damper command at 0 to 10 V nulls to 0 volts, which the actuator reads as a legitimate command to its zero position. On this unit that is minimum outside air. A broken damper command wire therefore parks the unit at minimum outside air permanently, with no alarm, and the symptom that eventually arrives is a ventilation or economizer complaint months later, in a season nobody associates with the wire.

The valve command at 4 to 20 mA nulls to 0 mA, well below the 4 mA floor, so the receiving actuator can and usually does declare a fault and drive to its configured failure position. The identical break on the identical unit produces an alarm the same day.

The bus space temperature is the worst of the six, and it does not look like it. It nulls to whatever value last arrived, held indefinitely with no indication. A room device that dies at 71.0 F on a mild afternoon leaves the loop controlling to a permanent 71.0 F, and the loop will do that faithfully through a heat wave. The tell is that the value stops moving entirely, which nobody notices on a point that normally moves slowly.

Both dry contacts null to open, and the operational consequence is opposite. For the fan status a broken wire reports the fan stopped, which the receiver cannot distinguish from a real stop but which nobody can miss: a false alarm and a wasted trip. For the filter switch a broken wire reports a clean filter forever, which is the direction nobody hears at all. Same type, same null, decided entirely by which state the contact uses to mean trouble.

What this survey changes about a live call. Given a complaint on this unit with no alarm history, the three points that fail without announcing anything are your first suspects, not your last. The other three are close to cleared by that same absence: the two that declare a fault on a break would have declared one, and the fan status would have raised its false alarm. That inverts the usual order, in which a tech starts at the point that is easiest to reach.

Turning a silent point into an announcing one

Three changes are available without replacing controllers.

Re-range a zero-based voltage signal to an offset one where both the controller output and the receiving actuator support it. It converts a permanently silent break into an announcing one, and it shifts the actuator's whole span, so it is re-verified by sweeping the actuator against a fixed reference mark afterward, not assumed.

Reverse the sense of a dry contact so the condition you care about is the closed state and the null is the alarm state. Wired to close on a dirty filter, a broken wire is silent; wired so the healthy state holds the contact closed, a broken wire reports dirty. You trade a silent miss for an occasional false trip, which is the correct trade on anything protective.

Add an age check to a bus point, with the limit stated as a multiple of the point's normal update interval rather than as a fixed clock value, so a slow-updating point does not alarm continuously.

Where a survey turns up a silent point on a device serving a protective function - a limit, a pressure switch, an interlock - the change is not optional and not yours to defer. A protective device whose broken wire reads as healthy is a protection that has already failed, quietly.

References

• 29 CFR 1910.333(a)(1) - live parts de-energized before work, and the narrow conditions permitting energized troubleshooting
• 29 CFR 1910.333(b)(2) - lockout and tagging for work on electric circuits and equipment
• NFPA 70E-2021, 120.5 - process for establishing and verifying an electrically safe work condition
• Manufacturer documentation for the transmitter's minimum operating voltage, the receiver's input impedance and sense resistance, and the supervised panel's specified end-of-line resistance
• See related: Why a Shared Common Causes Symptoms That Make No Sense; What a Sensor Actually Reports; The Difference Between a Status and a Command