Servo amplifier fault: repair or replace — a buyer checklist
A servo amplifier has failed — or something is wearing its clothes. The decision rewards the bench's order of work: record before removal, read the alarm pattern, separate the amplifier from its imitators, then weigh repair against replacement with the real variables on the table. A prepared buyer arrives with that record already made — alarm codes themselves belong to the bench, not to the buying decision.
From the Spareaxis blog — guides for buyers of FANUC spare parts.
Record before the unit comes out
The cabinet, before anyone touches it, holds evidence that removal destroys. Record the alarm exactly as the control displays it. Record which axis stopped and whether the spindle kept running. Record when it happens: at power-up, at first move, under cut, after an hour warm, after a power event. Record what the neighbouring units show — a cabinet where several amplifiers alarm together is telling a different story from a single silent axis, and the story changes the suspect list entirely.
Then photograph the failed unit's label in place and transcribe the full number — suffix included — into the record. The A06B table shows why: groups like 6222 contain a servo module and a spindle module separated only by suffix, and the time to discover which one is fitted is while it is still in the cabinet, not from a bench photograph of a unit already shipped.
Read the pattern before blaming the amplifier
Three patterns recur. All axes disturbing together points upstream — the shared power supply feeds every amplifier, and a supply in trouble writes its signature across the whole chain. One axis alarming under load but behaving at idle points at the drive chain for that axis, with the motor and its cabling as live suspects beside the amplifier. One axis dead in all conditions, with clean neighbours, is the pattern that most honestly convicts the amplifier itself.
Feedback wears amplifier clothes especially well. An encoder dropping its signal produces axis alarms that blame the drive; the amplifier is merely reporting the loss. A disciplined buyer therefore writes the pattern down, not a conclusion: the pattern is evidence, the conclusion is the bench's job, and confusing the two is how good amplifiers get replaced for bad cables.
The repair route, honestly stated
Repair begins with diagnosis on the test bench: the unit is run, the fault located to a stage, and the repair priced against that finding. The virtues of the route are real — the exact fitted unit returns to its cabinet, no pairing questions arise, and a stage-level fault is cheaper than a unit. Its costs are also real: the machine waits for the bench's verdict and the repair's completion, and damage spread across several stages can price a repair up toward the shelf alternative, at which point the desk will show both numbers.
Repair suits a unit whose fault is located and bounded, a machine that can wait, and a cabinet where the fitted unit's exact identity matters — older machines and uncommon groups especially. It suits less well when the machine earns money daily and the same number sits tested on a shelf. Buyers can read how an intake runs from their side on the repair page before a unit is sent.
The replacement route, honestly stated
Replacement buys time and certainty of a tested unit: the shelf amplifier — new, new-surplus or used as the condition guide defines them — passes the bench gate and ships. Its disciplines are identity and evidence: the full number including suffix, the condition of the specific unit stated plainly, a test before dispatch, and a test video requested by name. A replacement ordered without those four is a parcel, not a plan.
Replacement suits a stopped machine, a fault spread across a unit, and any case where the failed unit's own future is doubtful — and it leaves a question behind: the old unit. Sending it for assessment after the machine is running again converts a dead amplifier into knowledge, and sometimes into a repaired spare, which is the correct ending for a machine that has now proved it can stop.
The decision in one pass
Weigh four variables. Downtime cost per day, against the bench's turnaround for a located fault. Fault shape: located and bounded argues repair; spread or unknown argues replacement while diagnosis continues. Identity scarcity: an uncommon number argues repair of the unit in hand; a common shelf number argues replacement. And the value of a spare: a machine that has stopped once for an amplifier has made the case for holding its twin, and either route can end with a spare on the shelf if the buyer plans it.
Whichever route is chosen, the inquiry that starts it is the same document: the recorded alarm, the pattern, the full part number, the machine, and a statement of whether the machine is stopped. Send that to the repair desk or the contact page and the first answer can address the actual decision instead of reconstructing the evidence.
Keep the record with the machine
Whichever unit ends up in the cabinet, close the loop on paper. File the recorded alarm and the pattern notes with the machine's own maintenance record, together with the full number of the unit that came out and the full number of the unit that went in. Cabinets outlive the people who stopped them; the next stoppage, perhaps years later and under different hands, starts from whatever the file holds. A file that shows the same axis alarming twice, with an amplifier changed between the two events, redirects suspicion toward the motor, the cabling and the feedback line before any money moves — and that redirection is worth more than the minutes the filing takes.
What a repair quotation should separate
A useful repair quotation separates three things that are often bundled into one number. Diagnosis is the bench work that names the fault, and it has a price whether or not a repair follows. Repair is the work on the buyer's own unit, priced against the diagnosed fault and bounded by what the bench found. Exchange is a different transaction altogether — a stocked unit against the failed one — and it should be quoted as its own line, with the condition of the stocked unit stated, so the buyer can compare unlike things knowingly.
Ask also what evidence closes the job. A repaired amplifier should come back with the test it passed, stated plainly: run on the bench, under load where the bench allows, with the original fault shown cleared. If a quotation cannot say what its unit was tested on, the price is carrying a risk the buyer cannot see. The same plainness applies to time: a repair time quoted from the day the unit reaches the bench is a commitment; a repair time quoted from the day of the first email is a hope.
The returned paperwork deserves the same separation. For a repaired unit it should name the fault found, the work done and the test passed, in that order; for an exchanged unit it should name the stocked unit's condition and the test it passed before dispatch. Paperwork that merges the two transactions into one vague assurance leaves the buyer unable to say, a year later, which promise stands behind the unit in the cabinet — and that is exactly when the question gets asked.
Related pages
Send an inquiry
Send the part numbers you need, or the number of a unit you want assessed, with the machine each belongs to. One desk answers for the shelves and the bench.
Open the inquiry form