Hard Start Kit Assessment and Installation

Purpose

A start assist device is a diagnostic result, not an upsell and not a default add-on. Every symptom that sells one, a compressor that growls and trips, lights that dim across the house, a unit that starts every morning and refuses at four in the afternoon, is also produced by a weak run capacitor, a supply that sags under load, or a compressor on its way out. Fit the kit over any of those and you reach the same failure one step slower with a part number on the invoice, and the shop now owns the outcome.

Scope

Covers assessment and, where indicated, installation of start capacitor and potential relay assemblies and solid-state start devices on single-phase reciprocating and scroll compressors in residential and light commercial split systems and packaged units.

Does not cover inverter-driven or variable-capacity equipment, where a start device has no function and may void the warranty; three-phase compressors; run capacitor and contactor replacement; or compressor condemnation. It treats supply voltage problems as electrician work rather than something a capacitor fixes.

Roles and responsibilities

Procedure

1. Isolate, prove dead, and discharge before anything is opened. Open the disconnect, prove the meter on a known live source, read the line terminals, prove it live again per NFPA 70E-2021, 120.5 with work practices at 29 CFR 1910.333(b)(2), then bleed the run capacitor with an insulated resistor. Acceptance: under 1 volt at the line terminals and under 5 volts DC across every capacitor terminal pair. Wrong looks like voltage present at either check. Stop rule: trace the feed rather than assume, and if nobody on site is a qualified person under 29 CFR 1910.332 the electrical work goes to an electrician. Hazard: the run capacitor holds a charge with the disconnect open and every later measurement in this SOP puts a hand past it.

2. Clear the three look-alikes before a start device is on the table. Measure the run capacitor against the tolerance printed on its can, meter each compressor winding pair and to the shell, and check the contactor faces. Acceptance: capacitor inside its printed band, run plus start winding equal to the start-to-run reading within meter tolerance, no continuity to the shell, and contact faces free of pitting. Wrong looks like a capacitor low on its band or a winding reading to ground. Stop rule: fix any of the three first and re-test the start, because each one on its own produces the exact symptom a start kit is sold for. Hazard: still isolated here, so the exposure is the compartment edge and a capacitor that must have been discharged at step 1.

3. Measure the start event itself with the unit energized. Restore power standing to the hinge side of the disconnect, clamp the compressor common lead, and record peak current, the time it stays there, and voltage at the compressor terminals during the draw and at the disconnect line side at the same moment. Acceptance, all three: peak current consistent with the nameplate locked rotor amps; the locked-rotor period under 1.0 second, which is this shop's gate, tuned if you prefer but written down; and terminal voltage during the draw at or above the minimum utilization voltage printed on the nameplate. Wrong looks like a compressor sitting at locked-rotor current for seconds, or terminal voltage falling under the nameplate minimum. Stop rule: a terminal voltage below minimum utilization voltage is a supply problem, so compare the disconnect's line side against its load side under the same draw and route anything upstream of the unit to a licensed electrician before a start device is discussed. Hazard: this reading is taken inside an energized cabinet at locked-rotor current with the fan turning, so it is qualified-person work under 29 CFR 1910.333(b)(2) and 1910.332, clamp on one hand, body clear of the contactor face, and the customer out of the enclosure area.

4. Decide whether a device is indicated, and confirm the equipment permits one. Put the three step 3 numbers beside the gate and check the equipment manufacturer's service literature for that model. Acceptance: a locked-rotor period over the gate with capacitor, windings and supply voltage all confirmed good, plus written manufacturer permission or silence rather than a prohibition. Wrong looks like a decision made from the customer's description rather than the numbers. Stop rule: no device where any of the three look-alikes is unresolved, and none at all on inverter-driven equipment, where it has no function and can void the warranty. Hazard: none at this step, it is a decision made at the truck with the panel closed; the harm it prevents is a compressor whose real fault gets a season of cover.

5. Select the device on the compressor's data, not on a tonnage sticker. Choose between a start capacitor with a potential relay and a solid-state device, then size it. Acceptance: for a relay kit, a start capacitor inside the microfarad and voltage range the compressor manufacturer's application data lists, with a bleed resistor across it, and a potential relay whose pick-up and drop-out voltages match that compressor's back electromotive force; for a solid-state device, a model listed for that horsepower with its stated minimum off time. Wrong looks like a universal kit fitted on a tonnage range alone. Stop rule: order against the compressor data rather than fit what is in the van. Hazard: none at this step; its risk is a mis-picked relay that never drops out and vents a start capacitor in the cabinet.

6. Install with power off and wire the relay so the start capacitor drops out. Isolate and discharge again exactly as in step 1, then mount the device and land the leads to its own diagram. Acceptance: device clamped clear of the compressor top rather than hanging on its leads, every lead on its diagram terminal, bleed resistor present across the start capacitor, and the existing run capacitor left in circuit. Wrong looks like a start device wired in place of the run capacitor, or a start capacitor with no bleed resistor. Stop rule: check against the diagram rather than reason from the old wiring, because the relay contacts are normally closed and a reversed connection leaves the start capacitor energized in run. Hazard: the compartment was re-energized at step 3, so it gets re-isolated, re-proved and re-discharged here, not assumed still dead.

7. Restore, re-measure the start, and prove the relay drops the start capacitor out. Close up, restore power from the hinge side, and repeat the step 3 measurement. Acceptance, all four: locked-rotor period now under the gate; running current at or below nameplate rated load amps; no voltage across the start capacitor once the compressor is running, which is the proof that the relay opened; and the disconnect pulled with the unit running to confirm it still stops the unit. Wrong looks like a start capacitor still energized in run, which is a relay wired wrong or picked wrong and will vent the capacitor inside days. Stop rule: a start capacitor that stays in circuit gets the device removed and the system left as found rather than run overnight. Hazard: this is the step that puts full line voltage and a starting compressor back with the customer present, so panels go on before power does, the customer is moved away from the unit, and nobody stands in front of the disconnect as it closes.

Exception path. A supply-side finding ends this SOP at step 3 with a referral, not a device, and the system is left in whatever state the customer authorizes with the finding written on the ticket. A customer who asks for a start kit anyway after a supply or compressor fault has been found gets a written decline, because fitting one over an unresolved fault transfers the next failure onto the shop. Where the equipment manufacturer prohibits a start device, the answer is no regardless of what the measurements show.

The record this produces

One line per assessment, filled where it is read: run capacitor nameplate, printed tolerance and measured value; the three winding readings and the shell check; peak start current against nameplate locked rotor amps; the locked-rotor period against the gate; terminal voltage during the draw and at rest, plus disconnect line side and load side under the same draw; the manufacturer's position for that model; and, where a device was fitted, its ratings and the step 7 drop-out proof.

These land against the unit serial. The locked-rotor period is the field worth trending, because a compressor whose start time grows across three seasons is failing whether or not it carries a start device, and a shop holding 0.6, 0.9 and 1.4 seconds has a conversation to have instead of a surprise.

Worked pass: 3.5 ton R-410A condenser, will not start on hot afternoons

Booking noted that the house lights dim on a start attempt and that the unit always runs in the morning. Step 1: disconnect open, 0.3 volts at the line terminals, meter proved live either side, capacitor bled to 1.6 volts DC. Step 2: run capacitor nameplate 45 microfarads at plus or minus 6 percent, so the band is 42.3 to 47.7; measured 44.1, inside it. Windings 1.2 ohms run, 2.6 ohms start, 3.8 ohms start to run, and 1.2 plus 2.6 is 3.8, so they agree with themselves. No continuity to the shell, contact faces clean. All three look-alikes cleared.

Step 3 FAILED on its third acceptance. Peak current 84 amps against a nameplate locked rotor of 88, held 1.4 seconds before falling to a running 15.8 amps, so the duration is over the 1.0 second gate and a device looks indicated on the first two numbers. The third stopped it: terminal voltage 238 volts at rest, sagging to 191 during the draw, against a nameplate minimum utilization voltage of 197. Under the stop rule that is a supply problem, so the disconnect was read on both sides during the same draw: 228 volts line side, 191 load side, a 37 volt difference across the disconnect itself. That is not something a start capacitor fixes, it is heat being made inside an enclosure.

The disconnect was re-opened, proved dead again, and one fuse clip was found blued and springless. It was replaced, which is inside this shop's scope for the equipment's own disconnect; anything further upstream would have gone to an electrician. Step 3 re-run: peak 82 amps, sag to 214 volts against the 197 minimum, locked-rotor period 0.9 seconds, under the gate.

Step 4: with the duration inside the gate and no look-alike unresolved, a device is not indicated. Steps 5 and 6 were recorded as not applicable with that reason rather than left blank, so the next tech does not read the gap as a device fitted and never documented. Step 7 still ran as verification: running current 16.1 amps against a nameplate rated load of 18.2, three consecutive starts inside the gate, and the disconnect pulled with the unit running to confirm it stopped the unit.

What the failure teaches: two of the three step 3 numbers said fit a start kit, and a shop that measures current and duration but not terminal voltage sells one on that evidence. It would even have worked, in the sense that the compressor would have started, because a start device shortens the time spent at locked-rotor current. The corroded clip would have kept heating, at slightly lower current for a slightly shorter time, until it opened or started a fire, over an invoice saying the starting problem was solved.

References

• 29 CFR 1910.333(b)(2), electrical safe work practices, with qualified-person definitions at 1910.332 and 1910.399; 29 CFR 1926.417 is the construction counterpart
• NFPA 70E-2021, 120.5, for the live dead live proving sequence used in steps 1, 6 and 7
• Compressor manufacturer's application data, for start capacitor range and potential relay pick-up and drop-out voltages used in step 5
• Equipment nameplate for locked rotor amps, rated load amps and minimum utilization voltage, and the equipment manufacturer's service literature for its position on start devices
• See related: Capacitor and Contactor Replacement Standard; No Cool Emergency Call Response; Outdoor Unit Pad and Clearance Correction