The £100 Lesson: How a Faulty DVD Drive Destroyed Two Power Supplies

Some repairs go exactly the way you expect. You diagnose the fault, order the part, fit it, and the customer is smiling within the hour. And then there are the ones that humble you, the jobs that remind you that no matter how many machines you have had your hands inside, a computer can still find a brand new way to catch you out.

This is one of those stories. It cost me two power supplies, a couple of return trips across Manchester, and a fair bit of head-scratching to crack. But it is also one of the most useful jobs I have ever done, because it taught me something that has saved dozens of machines since: when a power supply dies, the real question is never "is the PSU dead?", it is "what killed it?"

If you have found this post because your own PC is shutting down after a few seconds, or you have had a burning smell coming from the back of the case, stick with me. By the end you will understand exactly what is likely going on, why cheap power supplies make a bad situation far worse, and when it is worth calling someone in.

The call-out: a PC in Withington that would not stay on

It started, as most of my jobs do, with a phone call. A client based in Withington, South Manchester, got in touch because her desktop computer was powering down after just a few seconds. You would press the button, the fans would spin up, maybe you would see the first flicker of the manufacturer logo, and then nothing. Dead. Press it again, same story.

I always try my best to diagnose the problem over the phone first. It saves the customer money and it saves me a wasted trip. Over the years you build up a mental flowchart: certain symptoms point to certain culprits, and a "powers down after a few seconds" complaint has a fairly short list of usual suspects.

My initial instinct was that this was either an overheating issue or a problem with the power supply unit, the PSU, which is the box at the back of your computer that takes mains electricity and converts it into the various low-voltage rails (3.3V, 5V and 12V) that every other component depends on. When a machine powers up and then immediately cuts out, it very often means the PSU is either failing to deliver clean, stable power, or one of its built-in protection circuits is tripping the moment it detects something wrong.

But, and this is the important bit, that is an instinct, not a diagnosis. A "powers off after a few seconds" fault can be caused by all sorts of things: failing memory, a dying graphics card, a fault on the motherboard, a struggling processor, even a corrupted BIOS. The list genuinely does go on. My job is not to guess; it is to prove it. And the way you prove it is with the most reliable tool in any repair engineer's kit: the process of elimination.

The process of elimination: stripping it back to basics

When I arrived at the client's house, I did what I always do with this kind of fault. I opened the case and started removing every component that the computer does not strictly need in order to power on and stay on.

Out came the hard drive, the DVD drive, the graphics card, the keyboard, the mouse and the speakers. The logic here is simple: a desktop PC will happily power up and sit at a "no boot device" screen with nothing but a motherboard, processor, a stick of RAM and a power supply connected. Everything else is a potential point of failure, so you take it all out of the equation. If the machine suddenly behaves itself, you know the fault was in one of the parts you removed, and you reintroduce them one at a time until the problem comes back.

So I stripped it right down to the bare minimum, and the computer still shut down after a few seconds.

Now, that is actually a useful result. It told me the fault was not hiding in the graphics card, the storage, or any of the peripherals, because I had removed all of those and the symptom had not changed. That left a much shorter list: the motherboard, the CPU, the memory, or the power supply itself.

The next logical step was to swap in a test power supply, a known-good unit I carry specifically for this purpose. This is one of the single most useful things any technician can own, because it lets you instantly rule the PSU in or out without ordering anything. I disconnected the customer's power supply, connected my test unit, and hit the button.

The result was a positive one. The computer powered on, and, crucially, it stayed on. At that point I was confident my initial diagnosis was correct. The customer's original power supply was the problem. Job more or less done, or so I thought.

Ordering the replacement (and where it all went wrong)

I ordered a new power supply to match the one already in the machine. This is where I would make my first mistake of the job, though I did not know it yet, and it is a mistake worth dwelling on because it is the whole reason this story is worth telling.

When you replace a PSU on a like-for-like basis, the natural instinct is to match the wattage and the form factor so it physically fits and electrically does the same job. What I did not sufficiently weigh up was the quality of the unit, and specifically the quality of its protection circuitry. More on that shortly, because it turned out to be the hinge the whole job swung on.

The following day I returned to Withington to fit the new power supply and put all the components back in. Hard drive back in, DVD reconnected, graphics card seated, everything plugged in. With the machine fully reassembled, it was time to power it up and put a smile on my client's face.

It was not going to be my day. This computer had other plans for me. As soon as I powered it up, I knew something was wrong. Within seconds the machine powered down again, and this time there was a smell of burning drifting from the back of the power supply.

I have experienced this with older power supplies over the years. But this was a brand new unit from a reputable brand, and modern PSUs really are not supposed to do this. I was genuinely puzzled, and more than a little concerned, because it is not normal for a 12V system to create that kind of mess. Something had gone badly wrong, and a new component cooking itself within seconds of being switched on is the sort of thing that makes you stop and think very carefully before you press anything again.

And then the client said something that turned the whole job around. She mentioned that she had noticed a similar burning smell in the days leading up to the original problem. Bingo.

The penny drops: something was shorting to ground

That one comment reframed everything. If there had been a burning smell before the first power supply failed, then the PSU was not the root cause at all, it was a victim. Something else in that machine was pulling far more current than it should, and almost certainly creating a direct short to ground.

Here is what was really going on, and why the choice of replacement PSU mattered so much. A direct short is what happens when electricity finds a path it was never meant to take, typically when a positive voltage line connects straight to ground with little or no resistance in between. When that happens, current spikes enormously. A well-designed power supply has a stack of protection circuits precisely for this scenario, the most relevant here being Over-Current Protection (OCP) and Short-Circuit Protection (SCP). The job of these circuits is to detect that abnormal current draw within milliseconds and shut the entire unit down before anything is damaged. You can read more about how these safeguards are supposed to work in the ATX power supply specification that every quality unit is built around.

The replacement I had fitted was essentially a budget copy of the original unit, and the original had failed in exactly the same way. Both, it turned out, had inadequate protection. So instead of detecting the short and cutting out cleanly, the new PSU just kept pushing current into the fault until it cooked itself. The same weakness, the same outcome.

And this is the lesson I want every reader to take away: had I fitted a properly engineered power supply with robust protection, the outcome would have been completely different. A good unit would have detected the short, refused to deliver power, and shut the system down before burning out. The PSU would have protected itself, and given me a far clearer diagnostic signal in the process. Cheap power supplies do not just fail more often; they fail destructively, and they hide the real fault while they do it.

The takeaway in one line: a quality PSU shutting down cleanly is doing its job. A cheap PSU burning out is taking the evidence down with it.

Back to elimination, and finding the real culprit

Of course, at this point I still had not actually found the short. I had worked out that something was shorting to ground, but I did not yet know what. So I went straight back to the process of elimination, but this time I changed my technique.

Previously I had been disconnecting components by pulling the cables from the motherboard end. This time I decided to disconnect them at the component end instead, starting with the DVD drive. Rather than unplugging its cables from the board, I removed them from the back of the drive itself. Straight away, I spotted the problem.

The power connector on the back of the DVD drive was completely melted, and so was the corresponding plug on the power cable feeding it. That was it. That was the short, sitting in plain sight the moment I looked at the right end of the right cable. A fault inside the optical drive had created a low-resistance path to ground, and every power supply I had connected had been dutifully feeding current straight into it until something gave.

This was the real culprit all along. The DVD drive had been quietly killing power supplies, and the burning smell the client noticed days earlier had been the very first warning sign.

So was my original diagnosis wrong? Yes and no. I was correct that the power supply was failing, that part was never in doubt. What I had missed was the reason it was failing. And that distinction is the entire point of this article.

Why this matters: the difference between a symptom and a cause

When components fail, it is not always possible to tell why they failed just by looking. A dead PSU might be a dead PSU, or it might be a perfectly healthy PSU that has been murdered by something else in the system. On this occasion, it cost me two power supplies to learn the difference.

If you take nothing else from this post, take this: a failed power supply is a symptom, not always the disease. Before you replace one, it is always worth asking what else might have caused it to die, particularly if there has been a burning smell, repeated failures, or a unit that fails again shortly after replacement. Swapping the PSU without investigating the cause is how you end up cooking the replacement too.

This is exactly why a methodical, eliminate-one-variable-at-a-time approach beats guesswork every single time. It is slower. It is less satisfying than a quick swap. But it finds the actual fault instead of just treating the most obvious symptom, and in the long run it saves the customer money. You can see the broader logic behind this kind of structured fault-finding in any decent guide to computer hardware troubleshooting, which is essentially the philosophy I have applied to every repair for years.

Warning signs your PSU (or something connected to it) is in trouble

Drawing on this job and hundreds like it since, here are the symptoms that should make you stop and pay attention. If you are experiencing any of these, it is worth getting the machine looked at before a small fault becomes an expensive one:

• The PC powers down after a few seconds and will not stay on, no matter how many times you try.
• A burning, acrid or electrical smell coming from the back of the case. This is never normal and should be treated as urgent. Switch the machine off at the wall.
• Random shutdowns or reboots, especially under load such as gaming or video editing.
• The PC will not power on at all, with no fans, no lights and no sound.
• Buzzing, clicking or coil-whine noises from the power supply area.
• Repeated component failures, particularly if a newly fitted part fails quickly, which can point to an underlying fault feeding the problem.
• Visible scorching, melted connectors, or bulging on any cable, drive or component.

A burning smell in particular is one you should never ignore or "wait and see" on. As the experts stress in their fire safety guidance on electrical equipment, an overheating electrical fault is a genuine fire risk, not just a hardware inconvenience. If you smell burning, power the machine off at the mains and do not switch it back on until it has been checked.

The real lesson: do not cut corners on your power supply

The power supply is, frankly, the most under-appreciated component in any computer. People will happily spend a fortune on a graphics card or a fast SSD and then fit the cheapest PSU they can find to feed the lot, which is a bit like buying a sports car and putting the cheapest possible tyres on it. The PSU is the one component that touches everything else in your system, and when it fails badly, it can take other parts down with it.

A quality, well-protected power supply does three things a cheap one often will not:

• It protects itself and your other components with proper Over-Current, Over-Voltage, Over-Power and Short-Circuit Protection, so when a fault like that melted DVD drive occurs, it shuts down cleanly instead of cooking itself and potentially damaging your motherboard, storage or graphics card.
• It delivers clean, stable power, which means fewer mysterious crashes, reboots and "it just does that sometimes" gremlins that are maddening to diagnose.
• It lasts. A good unit from a serious manufacturer typically comes with a long warranty, often five, seven or even ten years, because the company actually expects it to survive that long.

The table below sums up why that protection circuitry is worth paying for:

When a fault hits	Cheap, no-name PSU	Quality, well-protected PSU
Protection circuits	Minimal or missing	Full OCP, OVP, OPP and SCP
If a short occurs	Keeps feeding current, can burn out	Detects it and shuts down in milliseconds
Risk to other parts	Can take the board, storage or GPU with it	Protects the rest of your system
Power delivery	Noisy, unstable, random crashes	Clean and stable
Noise	Often louder	Very quiet
Typical warranty	1 to 2 years	5 to 10 years

This is exactly why, when customers ask me what they should fit, I point them towards properly engineered units. As an official be quiet! retailer here in Manchester, I am a particular fan of their Pure Power, System Power and Straight Power ranges. They are built with the kind of comprehensive protection circuitry that turns a potential disaster like this one into a clean, harmless shutdown, and they are exceptionally quiet too. You can browse the full range on the Manchester PC be quiet! shop, and a dependable mid-range choice for most desktops is the be quiet! Pure Power 13 M. If you are not sure which wattage or model suits your build, our PSU calculator will point you in the right direction, or just ask and I will happily advise.

I am not saying spend the earth. I am saying that the power supply is the last place you want to save twenty quid, because, as this Withington job proved twice over, the cheap option can cost you far more in the long run.

When to call in a Manchester repair expert

You can absolutely tackle some of this yourself if you are confident. Swapping a power supply is one of the more approachable PC repair jobs. But there are good reasons to call someone in, and this story illustrates most of them:

• You smell burning. Stop. This is a switch-it-off-at-the-wall-and-get-it-checked situation, not a DIY-it-this-evening one.
• You have already replaced a part and the fault came back. That is a strong signal there is an underlying cause that needs proper diagnosis, exactly as happened here.
• You are not sure whether the PSU is the cause or a victim. Working that out reliably needs test equipment and a methodical approach, and getting it wrong can be expensive.
• You would rather not risk it. There is no shame in that. A botched repair often costs more to put right than doing it properly first time.

If you are anywhere in Manchester, whether that is Withington, Didsbury, Chorlton, the city centre or further afield, and your PC is shutting down, refusing to power on, or worrying you with a smell it should not have, that is exactly what I am here for. Whether it is a desktop or a laptop repair, I will diagnose it properly, give you an honest, no-obligation quote, and only fit parts I would be happy to put in my own machine.

The bottom line

This job cost me two power supplies, but it gave me a lesson I have carried into every repair since: always ask what killed the PSU before you replace it. A burning smell is a warning, not a quirk. A part that fails again after replacement is telling you something. And a properly protected power supply is not a luxury, it is the cheapest insurance you can buy for everything else inside your computer.

The faulty DVD drive in that Withington machine was the real villain all along, hiding behind a symptom that looked, for all the world, like a simple power supply failure. Find the cause, not just the symptom, that is the whole game. If your computer is shutting down or worrying you, get a free, no-obligation quote and we will get to the bottom of it properly.