Fanuc aiPS Alarm 05 on Both Units Traces to Shared Supply

Two Fanuc αi power supplies on this 31i-B5-controlled NTX trip on alarm 05 (pre-charge failure). The aiPS-26-B feeds the sub-spindle amplifier and the aiPS-30-Bfeeds everything else. They trip singly or together, at idle or in light cuts, never in a heavy cut. The deciding fact is that both units fault. Two independent rectifier or capacitor failures are unlikely, so look for a fault in the supply path both units share.

Repairs that leave alarm 05 in place

Each of the usual first responses either treats one unit or treats a symptom. None of them addresses a fault common to both supplies.

• Replacing one power supply. The same alarm appeared during commissioning, and swapping one PS unit made it go away for about a year. The service technician was not confident that the unit was the root cause. Now that the aiPS-26-B and the aiPS-30-B both fault, a second swap would only reset the clock.
• Raising supply voltage. Moving a transformer tap up about 10 V to gain margin is a reasonable idea when taps exist. This transformer has a single output and no taps. The measured 210-214 V at the PS inputs is inside the 200-240 V rating on the unit label, and it matches the other Fanuc machines in the shop.
• Power cycling. This clears the latched alarm but does nothing to the cause. The shrinking run time is itself a symptom of progressive degradation.
• No-load wiggle tests. Tapping contactors and pulling on cables at idle found nothing. A high-resistance joint, such as a fuse holder, a disconnect pole or a contactor contact, can read normal voltage with no current flowing and then collapse under load. Loaded testing is what exposes it.
• Writing it off as mains quality. Storm damage and flickering lights make the grid an obvious suspect. However, older machines on the same feed that are normally sensitive to power quality stayed quiet. The first occurrence also happened a year earlier, when the power was clean. Mains sags may lower the trip threshold, but they do not explain the pattern alone.

DC link pre-charge sequence and what alarm 05 measures

When the CNC enables the servo and spindle system, each αi PS closes its external magnetic contactor (MCC) through an internal relay output. The three-phase input then charges the DC link capacitor bank through a current-limiting path. The PS watches the DC link voltage rise. If the link does not reach its charged level within the permitted window, the PS declares a pre-charge failure and shows 05 on its status display.

The alarm therefore measures a time-to-voltage condition. It is not tied to a specific failed component, and several things can prevent the link from charging on schedule:

• a missing or high-resistance input phase
• a supply sag during the charge window
• an MCC that bounces or drops out while charging
• a loose DC link short bar between the PS and its amplifiers
• degraded capacitors or a failed charge path inside the PS

Loose DC bus links are a known cause of alarm 5 right after a power supply has been replaced. That makes the DC link bars on the unit swapped at commissioning worth a torque check.

The load pattern separates heat faults from logic faults here. The trips do not track cutting load: none in heavy cuts, several at idle. That rules out thermal overload of the power stage. The recovery after a 30-minute cool-down, combined with steadily shorter run times, points to a different kind of heat fault. It is typical of an electromechanical part whose margin shrinks with temperature. A contactor coil gains resistance as it warms, which lowers its pull-in force. Worn contacts also run hotter and less consistently as they degrade.

Why simultaneous alarms on both supplies point to shared hardware

A single failing diode module or capacitor bank produces alarms on its own unit only. When one module alarms by itself, inspect the diode modules and capacitors first. When both units alarm, even at random, go upstream to anything both units depend on:

• the transformer secondary and the conductors from it
• the main disconnect and the fused isolator
• any magnetic contactor that feeds both PS inputs
• the enable chain that energizes the contactor coils (door interlock, safety circuit, CNC enable)
• the 200 V control power the PS units draw

Confirm how the PS inputs are wired. They may be in parallel from a common point, or daisy-chained so that one unit's input terminals carry both units' current. If they are daisy-chained, check that the conductors and terminals to the first unit are sized for the combined full-load current of both drives. Heavy-gauge line wiring does not rule out a single undertorqued lug at the chain point.

The current prime suspect is one contactor that clatters several times before it seats, at the first door open-close after power-up. That door cycle is required to enable the machine. After the first cycle it closes cleanly. Two different faults produce this clatter:

• A weak or hot coil, or a sticking armature. This is a hardware fault in the contactor.
• A steady coil being dropped out by an upstream signal. Here the contactor is only responding to the door interlock, safety relay or enable logic cycling its coil. This is a logic or wiring fault upstream.

A scope on the coil terminals tells these apart (see the procedure below).

Symptom-to-cause map for this NTX

Observed symptom	Most likely cause	What confirms or rules it out
05 on both PS units at the same time	Shared input: disconnect pole, isolator fuse holder, common contactor, supply sag	Three-phase logger at the PS inputs with sag and dropout capture
05 on one unit only, repeatable	That unit's diode module, capacitors or DC link bars	Swap-test history; torque on DC link short bars
Contactor clatters at first enable, then holds	Weak coil, or enable chain cycling the coil	Scope coil voltage during door open-close
Trips at idle, not in heavy cut	Not thermal overload of the power stage; timing or supply event	Correlate alarm timestamps with logger events
d9 pulse error on the main spindle alongside 05	Side effect of the supply event disturbing the spindle drive	Alarm history order: 05 first means the spindle alarm is secondary
Overspeed then 05 after spinning a spindle by hand	Known behavior on this machine type; not the fault	Reproduced on another NTX during the commissioning investigation

Supply margin and the quantities to log

The one quantity that settles this case is the phase-to-phase voltage at the PS input terminals during the exact second alarm 05 latches. Spot readings with a multimeter at idle cannot capture a sub-second phase dropout or contact bounce. You need an event-triggered recording.

Quantity	Value or limit from this installation	Where to read it
Transformer primary	420 V	Fused isolator upstream of transformer
Machine nameplate input	220 V	Machine rating plate
PS rated input range	200-240 V	Label on each aiPS unit
Measured PS input	210-214 V (about 215 V at the machine, varying a few volts between phase pairs)	PS input terminals, all three phase pairs
Margin to lower limit	10-14 V, about 5-7% of 200 V (derived)	Compare logger minimums against 200 V
DC link charge state	Not specified here	PS status display and the CNC servo/spindle diagnostic screens; limits in the Fanuc αi series maintenance manual
Voltage drop per disconnect or contactor pole	Not specified here	Millivolt reading across each closed pole under spindle load

A 5-7% margin is thin when the local network is disturbed. A grid sag that other equipment rides through can still pull one phase below the PS threshold for the length of the charge window. If the logger shows sags that coincide with trips, the cure is at the supply: the utility, the transformer rating, or dedicated conditioning. If the logger shows one phase dropping while the other two hold, the fault is a contact or joint inside the plant.

Isolation procedure for the shared supply path

1. Pull the CNC alarm history after every trip. Record which PS showed 05, whether d9 or an overspeed alarm appeared, the order of the alarms, the time, and whether the machine was cutting or idle.
2. Hire a power quality logger and connect it at the PS input terminals, not at the bus bar. Set it to capture sags, swells and phase loss at half-cycle resolution, and leave it in place for a full cycle of production plus idle periods.
3. Put a second channel, or a scope, on the coil terminals of the clattering contactor. Power up and do the door open-close enable cycle.
   • If the coil voltage itself chatters, trace the door interlock, the safety relay and the enable wiring.
   • If the coil voltage is steady while the contactor chatters, replace the contactor or its coil.
4. With the spindle loaded, measure the millivolt drop across each pole of the main disconnect, the fused isolator holders and every contactor in the power path. A pole reading clearly higher than its neighbors is a high-resistance contact. Follow up with an infrared scan after an hour of running. On machines with this history, a weak spring or pitted contact on one disconnect pole has caused exactly this kind of load-dependent random fault; cleaning the contacts and restoring spring pressure cleared it.
5. Confirm whether the PS inputs are in parallel or daisy-chained. Check terminal torque at every junction, and check conductor sizing against the combined full-load current if the inputs are chained.
6. With power isolated and the DC link discharged per the PS display and manual, check the torque on the DC link short bars between each PS and its amplifiers. Start with the unit replaced at commissioning.
7. Check connections outside the main power path as well. The 200 V control supply, the interlock circuits and the signal cables between the PS and the drive modules can all corrupt the enable sequence. A single loose limit-switch connection has been known to cause repeated faults that looked like feedback hardware failures.

The machine is under warranty, so coordinate opening the power path with the machine builder's service organization. A paid full health check is a sensible way to get the contactor examined under warranty terms. Flag the clattering contactor and the logger data to the technician.

Overspeed and d9 alarms riding along with 05

Spinning any spindle by hand faster than a set speed triggers an overspeed alarm, followed almost immediately by 05. The same behavior was reproduced on another NTX during the original commissioning investigation, so it is a trait of this machine type, not evidence of the fault.

It does show how sensitive the charge supervision is. Rotation of a drive motor with no command can disturb the DC link enough for the PS to declare a charge fault. Avoid hand-rotating the spindles while diagnosing, or you will add false entries to the alarm history.

The d9 pulse error on the main spindle drive clears on a power cycle and leaves the spindle working normally afterwards. Treat it as a consequence of the supply event unless it appears without an accompanying 05. Use the alarm history timestamps to establish which alarm came first.

Proving the fix on an intermittent fault

This fault has already gone quiet for days at a time. After one morning trip, the machine ran two full days without an alarm. A clean week proves nothing on its own. Before you call it fixed, check all of the following:

• The contactor seats on the first enable command with no clatter, cold and after several hours of running.
• The per-pole millivolt drops are even across all three phases under load.
• No 05 or d9 alarms occur over a period at least several times longer than the longest clean interval seen before the repair. The early history was one or two trips per week, so the observation window is weeks, not days.
• The power logger stays connected for the whole observation period. If an alarm returns with no logged supply event and no contactor bounce, the fault is inside a PS or its DC link hardware, and that data justifies a unit-level replacement.

FAQ