A Sighthill customer brought in a nearly-new build last week — carefully specced, neatly cabled, and absolutely refusing to produce a single character on screen. Three fans spun, the RGB on the memory lit up, and the monitor stayed black. By the time he brought it to us he had tried a different graphics card, swapped the kettle lead, and reseated the memory twice. The real fault was a brass motherboard standoff left over from a previous case he had reused: a two-pence part putting a short across the back of the mainboard every time he screwed everything down. His case had seven standoff holes; the board he had bought now only needed six.
That is how most first-boot failures end on the bench — a small, honest mistake that looks exactly like a dead CPU or a bad GPU until you work through the list in the right order. This is the order we actually use, written for a confident home builder in Sighthill-Wester-Hailes, Dalry-Gorgie, Corstorphine or anywhere else on the west side of the city who wants to narrow a dead build down before they start posting a parts receipt back to the shop.
Before Anything Else, Define “Won’t Boot”
“It won’t turn on” means at least four different faults depending on what the PC is actually doing when the power button is pressed. On the bench we sort every new-build no-boot into one of these buckets before touching a screwdriver, because each one points at a different half of the machine.
- Completely dead. Press the button, nothing happens. No fans, no LEDs, no case lights, no twitch. The fault is almost always in the power path: the PSU switch, the wall socket, the 24-pin, the front-panel header, or a shorted board.
- Spins for a second, then stops. Then does it again. Fans twitch for about half a second, the board tries to post, and it short-cycles. This is nearly always a POST protection trip — usually RAM, sometimes a loose CPU power cable or a stuck reset switch.
- Fans run, lights are on, but the screen stays black. The machine is powered but is not completing POST. This is the biggest bucket, and the one where motherboard debug LEDs earn their keep.
- POSTs, then hangs on the motherboard logo or the Windows installer. Different problem entirely — the hardware is fine, you are now in BIOS, boot order or an operating-system install conversation.
Writing down which of those four descriptions matches your build is the first step, because the fix for a dead-flat tower is nothing like the fix for one that short-cycles.
Completely Dead: Walk the Power Path
Confirm the obvious ones in order: the switch on the back of the power supply is set to “I” not “O”, the kettle lead is seated at both ends, and the wall socket actually works (an undermounted socket in a Sighthill-Wester-Hailes tenement that has lost one of its wires looks exactly like a dead PSU). Then open the side panel. Check that both the 24-pin main cable and the 8-pin EPS cable near the CPU are fully clicked in — the EPS is the single most forgotten cable in first builds, because it hides under the top shroud of most modern cases and the PSU often ships with two very similar-looking 8-pin cables.
If everything is seated, bridge the two power-switch pins on the front-panel header with a flat screwdriver for half a second. If the PC leaps to life, the fault is in the case’s power button or the wires behind it, not the board. If it still does nothing, pull the PSU out of the case and run the paper-clip test on the 24-pin to confirm the PSU itself is alive. We keep an old-stock known-good 500 W unit on the bench specifically for this test; swapping to it takes two minutes and settles the question for good.
Short-Cycling: It Is Almost Always RAM
If the build turns on for a second, stops, and keeps trying, open the manual that came with the motherboard and look up the correct DIMM slots. On almost every modern board with four slots, a two-stick kit must go in slots A2 and B2 — the second and fourth from the CPU — not slots A1/B1. Fit the sticks the wrong way and the board will fail to train the memory, trip, and start again. Press each stick down firmly until both end clips snap closed; a stick that looks seated but has one clip not quite over the notch is the single most common first-build fault we see.
If the right slots are confirmed and both clips are closed, try booting with a single stick in slot A2 only. A kit sold as DDR5-6000 will often not boot at the advertised speed on first power-up because the board defaults to a conservative profile and the memory expects an EXPO or XMP toggle before it is willing to run that fast. One stick at the board’s default speed almost always posts; from there you re-enable the memory profile in BIOS and add the second stick back. If you want the fuller planning context behind kit selection before any of this happens, our walkthrough on planning a first custom PC build covers how to pick memory that will train on your specific board.
Fans Run, Screen Black: Read the Debug LEDs
Almost every mid-range motherboard sold in the last five years has a Q-LED block somewhere near the 24-pin: four small lights labelled CPU, DRAM, VGA and BOOT. During a normal POST these flash on in sequence and go off within a few seconds. A light that stays solid is telling you exactly where POST is stalling.
- Solid CPU light. Either the CPU power (EPS 8-pin or 4+4-pin at the top of the board) is not plugged in, the CPU is seated badly, or a bent pin in the socket is making one of the pads unreliable. Pull the cooler, inspect the socket under a bright torch, and reseat. A bent pin can sometimes be straightened on the bench with a plastic card; we would always rather do this under a microsoldering scope than you doing it at the kitchen table.
- Solid DRAM light. Go back to the previous section — wrong slot, half-seated stick, or incompatible kit at advertised speed.
- Solid VGA light. No graphics output path. Either the GPU is not clicked fully into the PCIe slot, the PCIe power cable is not fitted (the lights on the back of some cards will fool you — the 8-pin is required even when the LED on the shroud is lit), or the monitor is plugged into the motherboard’s HDMI on a CPU that has no integrated graphics. Most Ryzen non-G chips and most of Intel’s F-suffix chips have no iGPU: a monitor on the motherboard output will stay black until the cable moves to the graphics card.
- Solid BOOT light. POST finished and the board cannot see a boot device. Check the M.2 drive is screwed down, the SATA cable is on both ends, and the drive shows up in BIOS before worrying about anything else.
The BIOS Older Than the CPU
If the DRAM light stays solid even with memory you have swapped twice, there is one more thing to try before swapping anything else. Many motherboards sitting on shop shelves were manufactured a year or two before the CPU you have just paired with them, and their shipped BIOS does not yet speak the new chip’s microcode. The symptom is a dead-looking board that may or may not light a debug LED and will never POST, no matter what memory you fit.
The fix is the BIOS Flashback button on the back I/O: a small button, usually next to a dedicated USB-A port, that flashes a BIOS file from a FAT32 stick without a CPU or memory being installed. Download the latest BIOS from the board manufacturer’s support page, rename it as the manual instructs, drop it onto a formatted stick, press the button, and wait for the LED to stop blinking. On the right board this turns a dead build into a working one in six or seven minutes. On a board without Flashback the realistic options are either an older-generation CPU borrowed from a friend long enough to run the BIOS updater, or a trip to the bench where we keep a boot CPU on hand for exactly this.
The Standoff That Shorts Everything Out
If the board is powered correctly, the memory is in the right slots and the debug lights give no hint, there is one more mechanical fault worth considering before condemning parts. A single brass standoff left in a case position that the current motherboard does not use will press against the back of the board and short a trace to the case metal. The symptom is anything from a dead board, to one that short-cycles, to one that posts intermittently depending on how tightly the lid is screwed down.
The sanity check takes five minutes. Unscrew the board, lift it clear of the tray, and count the standoffs. Lay the motherboard upside down beside the tray and look at the screw holes: every standoff must line up with a hole in the board. Any orphaned pillar either moves to a position that has a hole, or comes out entirely. On the Sighthill customer’s build at the top of this article, that one check saved him a motherboard replacement he had already added to a basket.
Common Front-Panel Traps
A couple of smaller things show up often enough to deserve naming. The power and reset wires on the front-panel header are easy to swap by one pin, and a reset button wired into the power pair will reboot the PC the second it tries to post. A reset switch stuck in the pressed position (the plastic cap misaligned on cheap cases) does the same thing. If the build keeps restarting cleanly rather than freezing, pull the whole front-panel header, bridge the power pins by hand, and see if the behaviour changes.
USB headers plugged into the wrong block are harmless but will stop the case’s front USB-C from doing anything useful, and a case fan plugged into the CPU_FAN header can cause some boards to refuse to post because they think the CPU cooler is missing. If you have moved a fan, check that the one on the processor is still the one feeding the CPU_FAN connector.
When the Hunt Is Honestly Over
If you have confirmed the power path, the EPS cable, correct DIMM slots, both debug-LED routes, the standoffs and the front panel, and the build still refuses to boot, the realistic next step is a second opinion rather than another parts order. On the bench we have known-good power supplies, memory kits, boot CPUs and GPUs that we can swap in one at a time to isolate the dead part — a diagnosis that would need you to buy and return a shelf’s worth of components at home.
The broader planning context for a custom build — sizing a case, choosing a cooler, matching parts to a specific room — is in our Tollcross piece on planning a build for a small flat, and if the dead build turns out to have a part you cannot clear, our guide to checking a used gaming PC covers how we stress-test the salvageable bits before anyone writes them off. For brand-new builds that need parts verified, hardware upgrades and custom PC builds on our main site set out the diagnostic turnaround, and the custom build service for Sighthill-Wester-Hailes has local collection and lead-time detail.
Bringing a Dead Build In
If you are in Sighthill, Wester Hailes, Dalry-Gorgie, Corstorphine, Livingston or anywhere on the west side of Edinburgh and you have a new build that will not boot, bring it in with the parts list. We will walk the full diagnostic sequence — power path, cables, debug LEDs, BIOS flashback where the board supports it, standoff check and bench-swap of any part the LEDs point at — and tell you honestly whether the fault is a reseat, a BIOS flash or a dead part before any component is replaced.
Book a diagnostic online or get in touch and we will arrange a convenient time to take a look — usually same day for anything in the EH11, EH14 or neighbouring postcodes.
Last updated: 10 October 2026