Packaged Heat Pump Auxiliary Heat Verification

Purpose

Auxiliary heat on a packaged heat pump is invisible for nine months and load-bearing for about six nights a year. A dead element, a failed sequencer contact or an outdoor lockout set wrong produces no symptom in mild weather, because the compressor carries the load. It produces a cold building on the coldest night of the year, which is the night your phone is already full.

The other failure runs the opposite way: a thermostat left in emergency heat after a service call, or an aux stage energizing alongside a healthy compressor, burns resistance heat all season on a bill the customer eventually questions. Both are found by staging the heat deliberately and putting a clamp on it, and by no other method.

Scope

Covers verification of electric resistance auxiliary and emergency heat on packaged heat pumps and heat pump rooftop units: nameplate and staging data, per-stage current against computed values, temperature rise cross-check, defrost and lockout logic, and the safeties.

Does not cover the refrigeration circuit, defrost board diagnosis or charge verification, and does not cover gas heat on a dual-fuel unit, which is owned by the furnace and commercial boiler service SOPs. Does not cover isolating and proving the second electrical supply that strips usually carry, owned by the rooftop disconnect verification SOP.

Roles and responsibilities

Procedure

1. Read the configuration off the nameplate before you test anything. Total strip kW, number of stages and the kW per stage, nameplate voltage and phase, minimum circuit ampacity and maximum overcurrent device, and how the control is configured for balance point, outdoor lockout and staging delay. Acceptance: every one of those recorded, including whether an outdoor lockout thermostat is fitted and at what setpoint. What wrong looks like: testing a 15 kW bank as though it were 10 kW because the site file was written from a model number. Stop rule: no legible nameplate means you get the data from the manufacturer before energizing anything, not from a guess at the wire size. Hazard: none physical, a nameplate read on a closed cabinet.

2. Isolate, prove dead, then inspect the highest-current connections in the unit. Open every supply, including the separate feeder strips usually have, lock and tag each, and prove dead with the before-and-after instrument check per NFPA 70E-2021, 120.5, under 29 CFR 1910.333(b)(2). Acceptance: zero at every strip terminal, and terminals tight with no bluing, browning or melted insulation at the lugs. What wrong looks like: a discoloured terminal treated as cosmetic, when it is the signature of a loose connection that has been heating for a season. Stop rule: a damaged terminal or a scorched element bracket is repaired before any test current is put through it. Hazard: strip terminals carry the largest current in the cabinet, so this is a qualified-person task under 29 CFR 1910.332 and 1910.399, and both supplies are locked before a hand goes in.

3. Compute the current each stage should draw at the voltage you actually measure. A resistance element is a fixed resistance, so current scales linearly with applied voltage while power scales with its square. Take the stage kW, divide by nameplate voltage for the nameplate current, then scale that by measured voltage over nameplate voltage. Acceptance: an expected current per stage and an expected cumulative total, written down before the clamp goes on. What wrong looks like: comparing a measured current to the nameplate figure at a site running 8 V low and calling a healthy bank weak. Stop rule: a measured supply voltage outside the equipment's rated range is its own finding and stops the heat test. Hazard: none physical, arithmetic done with the unit still isolated.

4. Stage the heat one step at a time and clamp each step. Restore power, close the cabinet, put the stat in emergency heat so the compressor is locked out, and bring stages on one at a time, reading cumulative current at the supply after each. Acceptance: each stage adds its computed current within 10 percent, which is this shop's default where the manufacturer publishes no tolerance, and the sequencer brings stages on separated in time rather than together. What wrong looks like: a total that looks about right because two stages came on and the third was never separated out. Stop rule: a stage that adds materially less than computed is an open element or a failed sequencer contact, and the unit goes back to step 2 for isolation before anything is proved further. Hazard: current is read with a clamp on a conductor through an existing access with the compartment closed, never with a hand inside an energized strip compartment.

5. Cross-check the electrical measurement against temperature rise. With the compressor still locked out so the rise belongs to the strips alone, read supply and return dry bulb and compare measured rise to computed: measured kilowatts times 3,412 gives Btu per hour, divided by 1.08 times the unit's rated cfm. Acceptance: measured rise within 10 percent of computed and inside the manufacturer's published rise range. What wrong looks like: a supply probe in line of sight of glowing elements, which reads radiant energy rather than air temperature and reports a rise the air never had. Stop rule: a rise well below computed at correct current is an airflow problem, so it moves to filters, blower and duct static before anything electrical is touched. Hazard: the 1.08 constant assumes standard air near 0.075 lb per cubic foot at sea level and falls with density, so at altitude it must be corrected before the comparison means anything.

6. Prove the logic, not just the elements. Confirm emergency heat brings on every stage with the compressor locked out, that auxiliary heat energizes during a defrost cycle, and that any outdoor lockout thermostat blocks aux above its setpoint and permits it below. Acceptance: all three observed and the lockout setpoint recorded as found. What wrong looks like: aux energizing alongside a healthy compressor in mild weather, which is comfortable, invisible and expensive. Stop rule: a lockout found bypassed or set implausibly is recorded as found before it is changed, because that setting is often the explanation for a bill the customer has already complained about. Hazard: forcing a defrost drives the outdoor coil and reversing valve through a real transition, so keep hands clear of the outdoor fan and expect the unit to shift under you.

7. Verify the safeties without defeating any of them. Confirm the high limit and any thermal cutout have continuity cold, and that the blower interlock energizes the blower whenever the strips are called. Acceptance: continuity confirmed, interlock proven by watching the blower start before or with the strips, airflow present. What wrong looks like: a limit that has been opening replaced as the fault, when a limit opens because airflow failed and a new one will open too. Stop rule: never jumper a limit to keep a unit running, and never replace one until you have established why it opened; if you cannot, the unit stays off. Hazard: the limit sits in the supply air stream directly above live elements, so both supplies are open and proved dead again before the meter touches it.

8. Return every setting you changed and prove normal operation resumes. Thermostat out of emergency heat, original setpoint and any lockout restored, cabinet closed, both locks removed, disconnects closed from the hinge side after calling it out. Acceptance: the unit resumes compressor heating with the strips off at the current outdoor temperature, the blower interlock still starts the blower on a heat call, and no setting differs from step 1 except one you documented and told the customer about. What wrong looks like: a stat left in emergency heat, which runs the full bank continuously and turns a service visit into the customer's worst bill of the year. Stop rule: any setting you cannot restore is written on the ticket and reported the same day. Hazard: full supply energy returns to the largest load in the unit with the customer sometimes present, so the cabinet is closed and everyone clear of both disconnects before either handle moves.

The record this produces

One auxiliary heat block: total kW, stage count and kW per stage, nameplate and measured voltage, computed and measured current per stage side by side, cumulative totals, the temperature rise computed and measured with the cfm used, emergency heat result, defrost aux result, outdoor lockout setpoint as found and as left, limit and interlock results, and every setting changed with who was told.

Computed beside measured is what makes the record usable a year later. A lone current reading means nothing, because supply voltage moves; a measured value next to the value computed at that day's voltage is a percentage anyone can compare against next season. A stage that read 99 percent of computed last year and 94 percent this year at the same voltage is an element degrading, visible only because both numbers exist.

Worked pass: 4-ton packaged heat pump, tenant reports cold mornings in a January cold snap

Rooftop packaged heat pump over a small office. Nameplate: 15 kW auxiliary in three stages of 5 kW, 240 V single phase, rated 1,200 cfm. Complaint only appears when it is well below freezing outside.

Step 1 recorded the nameplate and found an outdoor lockout thermostat fitted at 40 F. Step 2: both supplies opened, locked and tagged, proved dead at every strip terminal, terminals tight and clean. Step 3: measured supply voltage 232 V. At nameplate, one 5 kW stage draws 5,000 divided by 240, or 20.83 A; scaled by 232 over 240, which is 0.967, the expected current per stage is 20.1 A, so expected cumulative readings are 20.1 A, 40.3 A and 60.4 A.

Step 4 FAILED at the third stage. Measured cumulative current: stage one 20.0 A, stage two 40.2 A, stage three 40.4 A. Stage three added 40.4 minus 40.2, or 0.2 A, against 20.1 A expected, which is not a tolerance question. Under the stop rule the unit went back to isolation. With both supplies locked and proved dead, each element was read cold: a 5 kW element at 240 V nameplate should measure about 11.5 ohms, since 240 squared is 57,600 and 57,600 divided by 5,000 is 11.5. Elements one and two read close to that and element three read open. The sequencer contact for that stage closed correctly, so the element is the failure.

Step 5 was run on the two working stages to confirm the rest of the bank: measured kilowatts are 40.2 A times 232 V, or 9,326 W, which is 9.33 kW. That is 9,326 times 3,412, or 31,820 Btu per hour, and 31,820 divided by 1.08 times 1,200 cfm, which is 1,296, gives a computed rise of 24.6 F. Measured rise was 23.8 F, a difference of 0.8 F, which is 3.3 percent of 24.6 and inside the 10 percent band. Sea-level site, so the 1.08 constant was used uncorrected.

Step 6: emergency heat brought on both surviving stages with the compressor locked out, aux energized during a forced defrost, and the 40 F outdoor lockout permitted aux below it. Step 7: limits had continuity cold, blower interlock started the blower ahead of the strips, airflow present. Step 8: stat out of emergency heat, setpoint restored, unit resumed compressor heating with strips off at 34 F outdoor.

What the failure teaches: this unit heated the space adequately into the low thirties on two stages plus the compressor, which is why earlier visits found nothing. The complaint appeared only when the third stage was called, and that stage had been dead long enough that nobody could say when. Nothing looked or sounded wrong at any point in the visit; the finding is entirely in 0.2 A against 20.1 A.

References

• 29 CFR 1910.333(b)(2) with the qualified-person definitions at 1910.332 and 1910.399, and NFPA 70E-2021, 120.5, for the before-and-after instrument check
• The unit manufacturer's data for rated airflow, published temperature rise range and staging configuration, which govern over any default in this procedure
• See related: Rooftop Disconnect and Service Receptacle Verification; No-Heat Emergency Call Response; Thermostat Replacement and Configuration Handover