Workflow · no power

Welder No-Power Diagnostic Workflow

Follow this sequence when the welder is dead, the 24 V control rail is missing, or the auxiliary supply ticks and restarts. Every step defines the reference ground, the evidence expected and the next branch.

Fast triage: decide where the diagnosis starts

No DC busInput fuse, inrush path, bridge and mains wiring.

DC bus but no VCCStartup resistor, VCC capacitor, controller loading or bias clamp.

VCC but no output pulseVREF, RT/CT, current sense, COMP/feedback or controller.

Output pulse but no secondaryGate path, MOSFET, transformer, rectifier or shorted load.

Choose the correct reference ground before probing

Primary-side measurements: reference the KA3845/UC3845 ground pin or the documented primary return. Secondary rail measurements: reference the corresponding isolated secondary return. Chassis earth is not automatically the signal reference. Use isolation-rated instruments and never connect a grounded oscilloscope clip directly to a live primary unless the test setup is designed for it.

Mark the primary and secondary sides on a board photo before applying power. A correct numeric reading taken against the wrong reference is still invalid evidence.

Measurement order from input to regulated rails

Input continuity: verify fuse, switch, connector, NTC/inrush resistor and obvious carbon tracking with power removed.
Bridge and bus: confirm the rectified DC bus exists under controlled power.
Startup feed: measure the voltage before and after the high-value startup resistor chain.
Controller VCC: capture the ramp at the controller VCC pin relative to controller ground.
VREF: confirm the reference appears after VCC crosses the start threshold and is not pulled down externally.
RT/CT: verify oscillator activity at the timing pin.
Output pulse: check the controller output and then the MOSFET gate, one node at a time.
Primary switching: verify the MOSFET and transformer primary transfer energy without abnormal current.
Secondary rectification: check each winding, diode and filter capacitor.
Regulated rails: measure 24 V, 15 V, 12 V and 5 V only where that board actually provides them.
Feedback: verify TL431/KA431 divider, optocoupler LED current path and primary feedback response.
Loads: reconnect fan, relay, driver and control boards one at a time.

Decision branches after the first six measurements

DC bus missing

Do not move to the PWM section. Repair the input side first.

VCC never reaches startup

Test the startup resistor path, VCC leakage, shorted controller and bias clamp.

VCC reaches startup then falls

Check VCC capacitor ESR, auxiliary-winding handoff, output overload and feedback instability.

VREF present, no oscillator

Check RT/CT components and contamination before replacing the controller.

Oscillator present, no output pulse

Inspect current-sense noise, COMP/feedback clamp and device damage.

Output pulse present, rails absent

Trace gate resistor, MOSFET, transformer and secondary rectifiers.

Interpret rail combinations instead of judging one number

Rail evidence	Likely boundary	Next measurement
All secondary rails absent	Primary switching, transformer or common secondary return.	MOSFET gate/drain evidence and transformer continuity.
24 V low, 5 V normal	24 V winding, rectifier, filter or external 24 V load.	Disconnect fan/relay and measure diode/filter ripple.
24 V normal, 5 V absent	5 V regulator or logic-load branch.	Regulator input/output and rail-to-ground resistance.
All rails pulse in rhythm	Supply hiccup from VCC handoff, overload or feedback.	Correlate VCC, output pulse and secondary rise on the scope.
Rails normal unloaded, collapse loaded	Weak supply or excessive downstream current.	Reconnect loads individually and monitor current/ripple.

Load isolation sequence

Disconnect only documented low-voltage connectors, label them, and preserve protection paths. Start with external loads such as the fan and relay, then the display/control board, and finally driver-board loads. A rail that returns after one connector is removed identifies the faulty branch; it does not prove the disconnected board is safe to replace without further resistance and current checks.

Do not defeat feedback or protection to “make the rail appear.” The purpose of isolation is to locate the load, not to force the converter to run outside regulation.

Staged power-up after repair

Recheck primary and secondary shorts with power removed.
Power the auxiliary supply with the approved limiter and minimum safe load configuration.
Confirm one clean VCC startup, stable VREF, oscillator and output pulse.
Confirm secondary rails and ripple before connecting the power-stage driver.
Reconnect each low-voltage load and watch for rail collapse.
Only after control power is stable, reconnect the full inverter stage and verify protection inputs.
Perform a short controlled output test, then inspect temperatures and restart behavior.

Stop conditions before another power-up

The bus, primary return or isolation boundary is not identified.
A switching device or rectifier remains shorted.
A secondary rail measures near short to its return.
The current limiter remains bright or VCC collapses with no clear load isolation.
The output pulse is abnormal while the MOSFET and full power stage are connected.
Feedback or protection has been bypassed for testing.

Technical sources

Texas Instruments UC3845 product page — pin functions, UVLO, VREF, current limit and maximum duty-cycle reference.
UCx84x current-mode PWM controller datasheet — controller-level limits; it does not replace the actual board schematic.
Texas Instruments TL431 product page — secondary feedback reference when the board uses TL431/KA431 regulation.
WelderData internal technical records — used for measurement sequence and repair-pattern organization; exact readings remain board-specific.

Related repair map

Follow the evidence path from symptom to measurement, circuit and repair case.