How a hard drive works and the basics of diagnostics using HDDScan as an example

Poor computer performance is not always measured by the age of the processor or video card. Other components can also affect the power of the assembly. For example, the responsiveness of a computer is highly dependent on the quality of the discs. Let it have at least a dozen cores – if the disk cannot “feed” so many mouths, then comfortable work in such conditions cannot be achieved. This problem was completely solved with the help of solid-state drives with high speeds, but the main segment of volumetric drives is still inhabited by good old hard drives. Their peculiarity is that over time they begin to “crumble” and significantly reduce productivity. We will rely on special software to catch the scoundrel and return the freshness to the workhorse. In our example, this is the HDDScan utility. At the same time, let’s see what she can do with an SSD.

The speed of the disk directly affects the performance of the computer. This is because not only photos and music are stored on the disk, but also thousands of small system files, which the processor endlessly accesses during operation. Accordingly, the faster he receives the necessary data, the happier the user.

In general, the specifics of file delivery from disk can be compared to the work of a courier. If the road is flat, empty and without traffic jams, then the goods will reach the buyer instantly. When a courier hits a red traffic light every time or just physically cannot go faster because of pits and bumps, the delivery of pizza or a video card is delayed.


So it is with the speed of the disk: the fewer obstacles there are on the way to the RAM and the processor, the faster the computer works. Only instead of traffic lights and roads, there are nuances here, and such a primitive comparison makes it easy to understand how drives work, and why you need to check disks for bad sectors. Let’s start with the classics.

How a hard drive works
Let’s bypass the jungle of radio electronics and superficially consider the design of the hard disk. An ordinary hard drive consists of four main parts:

The plates contain information in the form of magnetized sectors. Each sector can contain from 512 bytes of data. He is in conjunction with others on the track. There are also several tracks on the plate, their number depends on the density. To read information, magnetic “heads” are used, which move with lightning speed over the entire surface of the plate and read sector by sector.
















2 TB Seagate Video 3.5 HDD [ST2000VM003]
In ideal conditions, the head should sequentially read each sector in one track and smoothly move to the next as it reads information, as it happens when playing vinyl records. But the fact is that the information on the disk is scattered over the entire surface, which greatly complicates the speed of access to certain sectors.

Let’s say you want to collect 100 apples. There are 100 trees in the garden with a distance of 100 meters between them. The condition is this: one person picks apples only from one tree, another picks only one from each. Of course, the first will pick up the required amount several times faster, because it will not waste time running between trees.

The same thing happens in the hard disk – only instead of people there are magnetic heads, instead of apples – sectors, and the track is taken as a hundred-meter distance. However, it is better to watch the work of the disc live:

The disk management system works like a real computer: the more powerful the processor and the more RAM a disk has, the faster it processes data from magnetic platters. Accordingly, such devices have failures and hardware problems.

500GB 2.5 “SATA Drive Samsung 860 EVO [MZ-76E500BW]
How a hard drive breaks down
Can’t be dropped

Of course, physically breaking a disk in half is not so easy, but it is easy to damage some parts during operation. The first thing a hard drive is afraid of is a blow or fall during operation. The reading head is located directly above the magnetic plate and, with a sharp impact, will definitely touch its surface, because the gap between them is smaller than a human fingerprint:

And the plate at this time rotates at a speed of 7200 rpm. After such a “touch” the disc can be thrown away:

To exclude cases with “gash” on the plates, the manufacturers taught the heads to park. Now the reader, in the absence of a read and write task, drives off to a safe place and does not “hang” over the spinning disk. And then hit, kick – the disk doesn’t care (just kidding).

Number of starts and spindle seizure

The uptime of the disk is also influenced by the number of spindles that rotate the magnetic plates. When the motor is turned on, it draws higher currents compared to the operating state, so the driver that controls its speed can easily fly out from overload.

This is unlikely to threaten a new drive, but it can easily cripple an older drive. Therefore, for those who have “served” it is recommended to turn off the energy saving and sleep functions, so as not to provoke equipment with increased consumption.

The flip side of this approach is heating. If you do not monitor the operating temperatures of the hard drive, you can bring it to the handle and overheat. Because of this, the thermal gap in the moving parts of the engine decreases and, as a result, the lubricant is squeezed out. Running without oil and cooling will cause the spindle to jam.

The disk crumbles, of course, not in the literal sense, the data sectors just fail. Those that contain 512 bytes of information and are located in the tracks. The cause of damage to the sectors can be a physical impact on the platters – cuts from the heads or the ingress and spread of dirt on the disks. Also, the integrity of the magnetic surface is affected by the temperature regime of the drive and simply the number of operating hours.

When reading information, each sector needs time to magnetize or demagnetize. Fresh and nimble sectors (blocks) do it very quickly, so a new disk always works noticeably faster. When the response of the blocks to state changes decreases, the time it takes to fully read information from the sector increases. And behind it the speed also decreases.

Let’s modify our apple example. fruits hanging on the lower branches are easier and faster to pick, and those that are ripe at the top are difficult to get. Accordingly, the faster the picker reaches the apple, the faster the required amount will be typed. And the one who climbs to the top for fruit will slow down the whole process.

If blocks do not respond to requests from the read heads, they are considered bad or bad (short for bad block). Such blocks appear on all hard drives, without exception, and even come across on new disks from the factory. This is the cost of magnetic storage technology. However, this is also the only frequent breakdown that you can find yourself at home. For this, there are special utilities, one of which we will turn to for help.

Find and eliminate BADs
To search for “bad guys” on the surface of magnetic plates, different software is used. Some programs only run on DOS, others can be run on Windows. One of such utilities is HDDScan . This is a very simple program and it is designed for fast disk run and search for bad sectors, which is what we will do. For this we need volunteers – this is a 3.5 hard drive for desktop computers and an old laptop 2.5 disk. Let’s see which of them got “bad guys”.

Then you can go to the SMART tab to find out about the state of the disk, which is monitored using the self-test program in the disk firmware. To do this, click the first button in the program and look at the output.

Green dots – everything is cool. Yellow exclamation marks call attention to defects in work. It is in this disk that the SMART system speaks about the excess of reassigned sectors and multiple errors in reading information. The error codes are 005, 197 and 198.

What is a remapped sector is an area that fell out of the magnetic plate and was reassigned to an additional place, where the manufacturer leaves a certain number of spare sectors for such cases. Their supply is limited, so when the limit is exhausted, the disk swears at being exceeded.

Catching bad blocks is fun, but very time consuming. Moreover, the duration of the surface check will depend on the volume of the hard disk. It’s good that for the tests we got into our hands models with a small volume, and one disk is also beaten to hell. In other words, what you need for visual testing. So, select the desired drive in the program and click on the “Tests” button.
Butterfly – similar to the Read test, but the blocks are read in pairs: the first block of the section and the last, and so on until all sectors on the track are checked;
Erase – the name speaks for itself: during testing, the blocks will be overwritten with zeros (this is necessary to completely format the disk without the possibility of data recovery).
For our experiments, Verify is enough . We click and be patient: sleep, walk, work, play.

During the test, the program will draw squares. These are blocks. Each is filled with a specific color based on the speed of response: the faster the block, the less time it takes to respond. The recall time is measured in milliseconds and is indicated on the right in the program window. The total number of blocks is also indicated there. This disk is healthy and the blocks in it are quite responsive, most of them run faster than 10 milliseconds. Three are the fastest and three more respond in 20ms. Such a disc should not fall apart.

Now let’s check another drive that was removed from the laptop due to poor performance. We turn on the same test:

Here they are, the bad guys. Literally from the very start, badges rained down. And also a bunch of colorful blocks. These are the laziest points on the surface of the disk, which react for a very long time to commands and will soon turn into Bads – blocks that are completely out of order and are actually holes on the surface of the plates.

For comparison, here’s what the 5-year-old SSD shows in the same test:

Almost all blocks are responsible for 5ms or less. This is not surprising, since SSDs are solid-state drives and do not have magnetized platters. They are less prone to degradation from physical stress and do not react to dirt. But they have memory microcircuits, which will certainly begin to lose “banks” after overcoming the factory limit for rewriting. For some disks, this is 100 terabytes, for some more. This SSD is crumbling due to the high mileage:

Other possibilities

The program can show the temperature of the drives. During testing, the hard drives work for wear and tear and noticeably heat up, so it is imperative to monitor the temperature and create good conditions for cooling the drives:

In the Tools -> Features section, there are several functions for fine tuning:

Automatic Acoustic Management – Allows you to set the speed of movement of the heads to reduce noise at the expense of performance.
Advanced Power Management is the same, only it adjusts the spindle speed.
Power Management – the time after which the disk will go to sleep.
Spindle Control – forced control of the spindle (motor).
These features are dependent on the drive and some of them may not be available for adjustment.

When to look for a replacement
The program allows you to quickly check the status of drives, not only HDDs, but also modern SSDs. This is not yet available to all utilities, and HDDScan is convenient and intuitive in this regard. Moreover, the utility runs from a running system and does not require creating a boot disk.

The obvious question from the reader is how to understand that the disk will soon begin to crumble and when to start looking for a replacement. To do this, we focus on the number of multi-colored blocks: