A struggling hard drive rarely stops working — it works a hundred times slower, in bursts. The benchmark graph of a hidden-fault drive has a shape you can learn to read in ten seconds: mostly normal, with craters where the head is fighting.
Here is the lived experience of a hidden drive fault: the PC freezes solid for ten or twenty seconds — mouse dead, video stalled — then everything resumes as if nothing happened. Open Task Manager during an episode and you see the paradox: disk active time pinned at 100 percent, transfer speed near zero. The drive is not idle and it is not working; it is retrying — reading the same weak spot again and again until an attempt finally succeeds. Software checks that only ask 'did the read work' miss this completely, because eventually it does. The measurements that expose it are about time: response latency, and the speed graph across the whole surface. The video names HD Tune among the tools for this; here is how to read what it draws.
Learn to spot the pattern in Task Manager's Performance tab. A healthy busy drive shows high active time with high transfer rates — work is flowing. A retrying drive shows high active time with transfer rates in the kilobytes and a graph that looks like static. Open the drive and copy a large file while you watch: if active time slams to 100 percent and stays there during every freeze, storage is your prime suspect. This costs two minutes and no tools, and it already separates a storage stall from a CPU or RAM stall, where the disk graph would sit quiet while something else pegs.
For precision, open Resource Monitor with resmon, go to the Disk tab and watch the Response Time column while the PC is misbehaving:
This is the same event Windows logs as a retried IO operation. A drive can pass every "did it work" test while failing this "how long did it take" test completely — and how long it takes is what you actually feel at the keyboard.
Run the read benchmark in HD Tune (the video's toolset) or a similar utility and look at the shape, not the average. A healthy mechanical drive draws a smooth staircase: fastest at the outer edge of the platters, stepping down towards the centre. A hidden-fault drive draws the same staircase with craters — sudden plunges to near-zero speed, then recovery. The craters are the head re-reading weak regions. Two details matter: craters that appear at the same position on repeat runs are surface damage, while a single shallow dip once can be Windows interrupting. A graph full of random small jitters on an old drive under a running Windows is normal noise; deep repeated craters at fixed addresses are not.
The full (not quick) error scan reads every block and maps the damaged ones in red. It is the most thorough surface evidence available — and this page must be honest about its price: on a large drive it runs for hours, working a possibly weakened drive at full effort the entire time. So the order is strict. If the drive holds data you cannot replace, clone it first and scan the original afterwards. If the data is already safe, the scan is an excellent verdict tool: scattered isolated bad blocks suggest contained damage; a growing cluster in one region suggests a head or surface failing progressively — either way, the map ends the guessing.
Before the verdict, one cross-check protects you from convicting an innocent drive: latency craters can also be manufactured by a bad cable forcing renegotiation, so if the SMART row C7 (UDMA CRC) is climbing, do the cheap cable swap first. Likewise, a PC that freezes with a calm disk graph, low response times and a clean benchmark is telling you the fault lives elsewhere — honour that evidence too. The goal of this diagnosis is not to blame the drive; it is to find which component's time measurements break during the freeze. When graph craters, response-time spikes and freeze timestamps all line up on the same drive, the case is as closed as home diagnosis gets.
Watch the video — the full behaviour check, from slow boots and hangs to the physical test.
▶️ Watch Full VideoThe classic cause is a drive retrying a weak spot. The read head attempts the sector, fails, recalibrates and tries again — sometimes dozens of times — while Windows waits and the whole PC appears frozen. When an attempt finally succeeds, everything resumes normally, which is why the episode leaves so little trace. Check disk response times during a freeze: values in the hundreds or thousands of milliseconds confirm the retry pattern.
A smooth descending staircase. Mechanical drives read fastest at the outer edge of the platters and slow steadily towards the centre, so the graph steps down in a fairly even curve. Deep narrow craters that plunge towards zero — especially at the same positions on repeat runs — are not part of that shape. They mark regions where the head is retrying, and they are the visual signature of hidden surface damage.
It is thorough but demanding: hours of continuous reading across every block. On a drive whose data is already backed up, it is a fine verdict tool. On a drive holding the only copy of important data, clone the drive first — a mechanically weak drive can fail during the scan itself, and the scan's red-block map is no consolation for files that were never rescued.