Works On Bypass But Not On Control Decision Tree

Why this matters

You jumper the controller out, the load runs. You reconnect the controller, the load will not run. This is the classic "load is fine, control is faulted" pattern. The diagnosis from here is to figure out which control element in the chain is broken. This decision tree gives the isolation method, applied so it works for any control circuit: HVAC, refrigeration, plumbing controls, irrigation, gate openers, elevators, conveyor systems.

The method is sectional isolation, the same approach NFPA 70B describes for control circuit troubleshooting. Bypass proves the load side is healthy; then you reintroduce the control side one segment at a time.

Symptom presentation

• Compressor pulls in when you jumper the contactor coil but not when the controller calls for it
• Pump runs when you direct-wire it but not when the timer calls for it
• Burner ignites when you jumper the gas valve but not when the thermostat calls
• Gate motor runs when you press the wall override but not from the keypad
• Fan runs when manual contactor is pushed in but not when temperature controller closes
• Solenoid valve operates when you energize directly but not on the system call

In every case the proof of load health is the bypass running. The fault is upstream of the load.

Quick checks

1. Confirm the bypass actually runs the load in normal operation, not just briefly. A bypass that runs for two seconds then drops out is a different problem than a bypass that runs steady.
2. Identify the control chain. List every element between the demand input and the load: signal source, transformer, fuse, switch, relay, contactor coil, interlock, safety, control board, output stage.
3. Identify the type of control signal. Line voltage? Low voltage 24V? Dry contact closure? Network message? PWM? 4-20 mA? Each type fails differently.
4. Find the demand-side end of the chain (thermostat, controller call, network message) and the load-side end (the contactor coil, valve coil, or driver). The fault is somewhere between.
5. Note the power source for the control circuit. Loss of control power is the most common cause of this symptom.

Isolation tree

Apply sectional bisection: find the midpoint of the control chain, test for the expected signal there, and decide which half contains the fault.

Stage 1: Control power

Confirm the control circuit has power. A blown low-voltage fuse, a tripped control transformer, a failed line-to-low-voltage transformer, or an opened secondary protection is the single most common cause of works-on-bypass-not-on-control. Read the control circuit at its source. Should be at nameplate voltage. Anything else: stop and fix the supply first.

Stage 2: Demand signal present at the source

At the demand-side end of the chain, confirm the demand actually closes/sends/asserts. A thermostat reading 1.2 mA when it should be closing fully is a demand fault. A network controller showing "send" on its diagnostic but with no actual change at the wire is a demand fault.

If demand is not asserting, the fault is in the demand source (thermostat, controller, sensor input that is preventing the call). Stop here and fix that.

Stage 3: Demand signal absent at the load coil

If demand asserts at the source but the load coil sees nothing, the fault is in the chain between. Bisect: read at the midpoint of the chain.

Common midpoints:

• Safety chain (series of interlocks: high pressure, low pressure, flame proving, limit switch, door switch, end-of-travel)
• Relay between low-voltage logic and line-voltage load
• Control board output that switches the coil
• Sequencer or staged controller
• Pilot relay or interposing relay

If signal is present at the midpoint but absent at the load coil, the fault is downstream of the midpoint. If absent at the midpoint, the fault is upstream.

Stage 4: Safety chain open

A safety chain in the open state silently kills the call. Walk the chain and identify which interlock is open. Common culprits:

• High-pressure switch tripped from a temporary condition that has not been reset
• Limit switch out of adjustment
• Door or panel safety microswitch out of contact
• Float switch in fault state
• Smoke detector or duct detector in alarm
• Lockout flag still set from a prior fault

Do not bypass the safety chain to "test." Identify the open device, understand why it is open, and address that cause.

Stage 5: Output stage failed

If the chain is clean to the output stage and the output stage does not energize the coil, the output stage is the fault. Triac, relay, MOSFET, mechanical contactor pilot, all of these fail in known ways. Replacement or board replacement is the resolution depending on whether the output is field-serviceable.

Confirming the diagnosis

Single-variable confirmation: with the identified faulted element removed or replaced, the system runs on demand without the bypass. If it does not, you have not found the only fault. Keep going.

Beware of multiple faults in series in the same chain. Replacing one open safety can reveal a second open safety that was masked. Walk the full chain after each fix.

Next steps

Document the failed element and why it failed if you can determine that. A high-pressure switch that opened because the condenser was dirty is a maintenance issue; replacing the switch without addressing the dirty condenser is a callback waiting to happen.

Re-verify the full control chain after the repair. The fault you found may have been a downstream effect of an upstream condition. The customer paying for the call deserves you to spend the extra five minutes confirming end-to-end function.

References

• NFPA 70B Recommended Practice for Electrical Equipment Maintenance, control circuit troubleshooting
• ISO 13379-1 Condition monitoring and diagnostics of machines, data interpretation
• OSHA 29 CFR 1910.147 Control of hazardous energy (lockout/tagout for safety chain work)
• OSHA 29 CFR 1910 Subpart S, Electrical safety
• ACCA Standard 4 Maintenance of Residential HVAC Systems, control circuit verification