Before you buy a board, it is worth knowing whether your own board can be fixed — because on the ST150, a large share of sudden deaths are simple power-circuit faults that a real board-level technician replaces in an afternoon. This page explains the trade honestly: who can do it, what it fixes, and where its limits are.
Replacement is the loud answer to a dead board; repair is the quiet one. Chip-level repair replaces the failed component instead of the whole board, and on server boards the failed component is very often in the power delivery circuitry — the hardest-worked and most repairable area of the design. This page will not promise your board is fixable; no honest page can, before testing. It will show you how to find the people who can test it, how to read their verdict, and how to weigh their quote against the board market.
👆 Chip-level repair is excellent at local electrical faults and poor at everything else. A good shop will tell you which side of the line your board is on — before charging you.
Chip-level (component-level) repair means a technician works on the motherboard itself with a multimeter, a bench power supply, a hot-air station and a microscope: tracing where the power stops, finding the failed component — a MOSFET, a driver IC, a capacitor, a resistor network — and replacing that component rather than the whole board. On server boards like the ST150's, the most common curable disease is in the power delivery stages: the circuitry that converts the supply's output into the exact voltages the CPU needs takes the hardest electrical life in the machine, and it is also the most repairable part of the board.
Not every repair shop — and the honest ones will say so. Look for:
Set expectations the way a good technician would. A board that died suddenly with a power-stage fault — shorted MOSFET, failed VRM phase, damaged charging or standby circuitry — is a good candidate, and these faults make up a large share of sudden server deaths. A board with a dead chipset/PCH, burnt internal layers, heavy liquid corrosion across multiple areas, or several unrelated failures is a poor candidate: the labour and parts approach a replacement board while the remaining life stays doubtful. Any shop promising near-certain success on every board is selling, not diagnosing. The professional answer is a probability after testing, and it should sound like one.
Chip-level repair normally quotes well below a replacement board for simple power faults, but it spends a different currency too: time. Diagnosis and repair take days, not hours, and a failed attempt returns you to the board market a few days older. If the server runs a business that loses money every day it is down, the used-board route often wins on total cost even at a higher part price — and some owners do both: fit a used board now, and send the original for repair as a shelf spare. That combination is not extravagance on a platform this old; it is how you make the next failure a one-hour event.
Protect yourself and your data with five minutes of discipline: photograph the board and note its part and serial labels; remove nothing you cannot refit (the shop needs the board as it failed); write the symptom history on paper — when it died, what it showed, any smell or event like a power cut or storm. A board that died during a thunderstorm tells the technician where to start looking. And confirm the server's data is safe elsewhere: the repair is about the board, but your backup should never depend on its success.
See the ST150 repair route discussed in the video, with the faults that fix well shown up close.
▶️ Watch Full VideoOften, yes — when the fault is local and electrical, such as a shorted MOSFET or failed power-delivery phase, which are common causes of sudden death. Dead chipsets, burnt board layers and widespread corrosion usually cannot be repaired economically. A good shop tests first and gives you a probability, not a promise.
Quotes vary by city and fault, so check locally on the day — but for simple power-stage faults a repair normally quotes clearly below a tested replacement board. Agree the diagnosis fee, the no-fix terms and the repair warranty before handing the board over.
A board repaired at the component that actually failed, by a technician who then load-tests it, is a legitimate working board — the replaced components are new. Reliability doubts belong to boards with widespread damage, which honest shops decline. Ask for the repair warranty in months and a stress test before collection.