Hard disk drive electronics from 40MB to 10TB

TechEkspert 06 May 2024 20:28 26 4149 Cool? (+9)

📢 Listen (AI):

TL;DR

Hard-disk-drive electronics are compared across models from 40MB to 10TB, showing how PCB layouts, interfaces, and head/motor control circuitry evolved over time.
Early 40MB drives use many discrete parts: parallel interface, 32k SRAM, ROM, and separate power circuits for the platter motor and heads.
A 40GB ATA133 drive provides transfers up to 10x faster than the 4GB predecessor, and the 80GB line moves to SATA with further integration.
The 10TB drive shows a drastically smaller PCB with three main circuits—controller, SDRAM, and power—raising the question of why the chips remain separate.
Spinning HDDs are increasingly displaced by FLASH memory, while helium-filled drives are sealed like a can of canned food.

Generated by the language model.

As part of the device interiors section in this material you will find pictures of the electronics boards of several hard drives, we will see how HDD electronics have changed over the years.

The first board is from a disk with a capacity of 40MB .
Many discrete elements can be seen.

1) parallel interface circuit
2) RAM -. SRAM 32k 8b
3) ROM
4) "power" circuit to control the platter motor
5) "power" circuit to control the heads

Image of a circuit board from a 40MB hard drive.

On the other side of the board you will find no electronic components, only the contact fields for the platter motor and the contacts for controlling the head arm and communicating with the head. This is how it will look like for all other disks.

View of the back side of a hard drive PCB with PATA interface, showing traces and contact points without visible electronic components.

Data is stored on one platter by two heads.

Inside of a hard drive with visible platter and head arm.

In the Ultra ATA drive 4GB a greater degree of integration can be seen, such drives used to achieve transfers in the order of 4MB/s which is nowadays a bandwidth often less than the speed of an internet connection...

1) interface circuitry and arguably the drive control
2) RAM - SRAM 64K x 16b
3) ROM
The rest of the circuitry is the control of the motor and head arm and the processing and generation of signals for the heads.

Hard drive circuit board with marked integrated circuits.

Two platters and four heads.

Image of a hard drive interior with a visible platter and read/write arm.

W 40GB drive with ATA133 interface we see a further increase in the scale of integration. Such drives provided transfers up to 10x faster than the 4GB predecessor.

1) interface chip and disk controller
2) RAM - SDRAM 512k x16b
3) power chip

Printed circuit board of a hard disk drive with three main components labeled with numbers.

One platter two heads.

Interior of an open hard disk drive showing the platter and head.

One disc.

Disk 80GB changed to serial SATA interface.
1) interface circuit
2) disk controller
3) RAM - SDRAM 1M x 16b x 4 banks
4) 'power' circuit.

Hard disk drive circuit board with four marked integrated circuits.

One plate and two heads.

Interior of an open hard drive showing the platter and read/write arm.

.

Drive 80GB Newer generation SATA, further scale-up of integration is evident. Gone is the separate interface chip, now the disk controller chip is connected to the SATA interface. In this version, the electronics have moved to the side where the contact fields for the platter drive and head arm and the signals for the heads are located.

1) controller
2) RAM - SDRAM 1M x16b x4 banks
3) 'power' circuit

One platter and one head.

Interior of a hard drive with visible platter and read/write arm.

Disc. 10TB a drastic reduction in the size of the electronics board can be seen.

Why are there still 3 main circuits on the PCB?

PCB from a SATA hard drive featuring various integrated circuits and connectors.

I suspect, that the "power" chip is made in a different process/lithography than the disk controller, as is the RAM, and it still pays to put two separate chips. RAM in a separate module is perhaps an easy production model with a different amount of cache?

Today most of the media we use do not contain spinning platters, HDD are displaced by FLASH memory.

Have you come across any interesting media designs
or any design details of interest to you in the pictures in this material?
If you prefer footage, below are attempts to identify the circuits used in the first of the 40MB HDDs:

And the operation of the newer HDD with SATA interface:

The helium-filled drives are sealed like a can of canned food

About Author

TechEkspert wrote 7170 posts with rating 5539 , helped 16 times. Been with us since 2014 year.

Comments

Add a comment

Mateusz_konstruktor 06 May 2024 22:07

Yes: DVD-RAM. [Read more]

andrzejlisek 06 May 2024 23:30

It may not be shown, but in my opinion, one thing hasn't changed: Every disk, different electronics. In the case of a working device, it doesn't matter much, but in the case of a hardware fault, this is... [Read more]

DJ_Opornik 07 May 2024 00:56

I also see a low noise LM358 on the first board. My guess is that it amplified the head read signal. There is also an LM324, surely it amplifies something, maybe it even generates the pad current.... [Read more]

prosiak_wej 07 May 2024 06:21

And interestingly - after 20 years it is still functional - it is possible to read files from it, freely delete and write. I have such a disc, which in my high school days was used as a "memory stick",... [Read more]

ArturAVS 07 May 2024 07:16

In the years of my first contact with PCs on the hardware side, I got my hands on a few MFM disks where the head drive was a simple stepper motor. The disks were produced by IBM and had a "staggering"... [Read more]

DJ_Opornik 07 May 2024 23:59

Good point, the 358 hums low, but it is slow. The question is whether it is too slow on archaic 40MB disk electronics, where the read speed was in kB/s, i.e. the stream from the head did not exceed a few... [Read more]

mmm777 08 May 2024 12:37

I remember disks that had a knob on the outside connected to the head drive. When the drive was running, it was spinning amusingly :) I think I also remember drive electronics that had two(?) Z80s in... [Read more]

LEDówki 08 May 2024 15:59

And I have a working 40MB conner drive. It sits in a working 486 40MHz and it barely works. Before ATA, SATA interfaces there were other weird and incompatible interfaces and also some of the control electronics... [Read more]

kaleron 09 May 2024 20:26

- PCB interchangeability is much greater than head interchangeability. - not true - PCBs from other related models often fit, and if the PCBs are not interchangeable (e.g. have a different signal processor... [Read more]

TechEkspert 10 May 2024 17:55

Separating the hard disk from the controller were, for example, the ST225 https://www.os2museum.com/wp/seagate-st-225-just-wow/ with interfaces like the ST-506 MFM https://en.wikipedia.org/wiki/ST506/ST412.... [Read more]

sq3evp 16 May 2024 11:16

I have also seen magneto-optical disks and so-called ZIP drives. [Read more]

prosiak_wej 16 May 2024 17:02

DVD-RAM discs came in a cartridge version, requiring dedicated drives. Later they were as loose discs, and could be handled by ordinary DVD recorders. An example of a magneto-optical disc would be... [Read more]

speedy9 22 May 2024 10:51

In Amiga it was used (Amiga VBS). Such a cassette held between 340 and 520MB of data, which for the second half of the 20th century was not a bad value, especially as cassettes were widely available. [Read more]

CMS 22 May 2024 17:57

And then there were dedicated devices for backing up data on magnetic tape called Streamers. I even repaired a few back in the old days. Mainly banks used them. [Read more]

kaleron 22 May 2024 18:04

- they still do. [Read more]

prosiak_wej 22 May 2024 18:25

And I have an HP streamer on USB and some cassettes physically probably even identical to DAT, but with what to bite it...? Because ZIP floppy disks (like DVD-RAM) are handled like regular removable d... [Read more]

CMS 22 May 2024 19:03

This surprised me. I was sure these devices were already extinct. But since they haven't, that can only mean one thing - they are reliable and the carrier is long-lived. [Read more]

LEDówki 22 May 2024 19:16

They use a library with LTO3, LTO4 tapes. Copying drags on mercilessly, especially on old disk arrays crammed with millions of files. [Read more]

stachu_l 22 May 2024 21:27

Maybe Unix/Linux and the tar command. Generally this command originates from the days of such tape memories with 1/2" tapes on spools much larger than in amateur reel-to-reel tape recorders. [Read more]

FAQ

TL;DR: From 40MB boards full of discrete parts to compact 10TB PCBs, HDD electronics moved toward tighter integration, yet one expert point remains: "the motor controller will probably never make it into the processor." This FAQ helps repair-minded readers, retro-computing fans, and storage learners understand why HDD controller design changed, why PCB swaps fail, and why tape still survives. [#21076170]

Why it matters: If you repair, recover, collect, or compare storage hardware, the thread shows which HDD changes were evolutionary, which were compatibility traps, and which old media still make technical sense.

Example from the thread	Interface or format	Electronics trend	Notable detail
40MB HDD	Parallel	Many discrete components	SRAM 32k 8b, separate ROM, separate power stages
4GB HDD	Ultra ATA	Higher integration	Around 4MB/s transfer mentioned
40GB HDD	ATA133	Fewer main chips	Up to 10x faster than the 4GB predecessor
80GB HDD	SATA	Interface merged into controller	Separate SATA chip disappears in newer generation
10TB HDD	SATA	Much smaller PCB	Still keeps controller, RAM, and power chips

Key insight: HDD PCBs shrank, but they did not become generic. Modern drives still separate signal processing, cache memory, and motor power because those functions have different electrical and manufacturing needs. [#21072338]

Quick Facts

The thread traces HDD electronics across 40MB, 4GB, 40GB, 80GB, and 10TB designs, showing a steady move from many discrete parts to three dominant chips on modern boards. [#21072338]
The 4GB Ultra ATA example is described as reaching about 4MB/s, while the 40GB ATA133 drive is said to deliver up to 10x that predecessor's transfer rate. [#21072338]
One 40MB board stores data on one platter with two heads; the 4GB example uses two platters and four heads. [#21072338]
LTO tape is still presented as practical archival media: one post states LTO-9 holds 18TB native or 45TB with compression. [#21105742]
A forum example of VHS-based backup on Amiga reports 340MB to 520MB per cassette, showing how large-capacity removable storage predated cheap flash media. [#21091687]

How did hard disk drive electronics evolve from a 40MB parallel-interface HDD to a 10TB SATA drive?

They evolved from many discrete chips to a compact, highly integrated PCB. The 40MB board had separate interface logic, 32k×8 SRAM, ROM, a platter-motor power stage, and a head-control power stage. The 4GB Ultra ATA board integrated more logic, the 40GB ATA133 board dropped to three main chips, and newer 80GB SATA boards folded the interface into the controller. By 10TB, the board became much smaller, but still kept three core functions separate: controller, RAM cache, and motor power. [#21072338]

Why do modern 10TB HDD PCBs still use three main chips for the controller, RAM cache, and motor power stage?

They stay separate because the jobs are electrically different. The thread explains that the motor controller handles relatively high currents, creates heat, and injects noise, so it is better kept away from the signal processor. RAM also remains separate because it is a different memory function and may support cache-size variants. As one expert put it, “the motor controller will probably never make it into the processor,” because mixing power switching with sensitive read-channel processing hurts design margins. [#21076170]

What is ST-506 MFM, and how did drives like the Seagate ST-225 work with an external controller?

“ST-506 MFM” is an HDD interface standard that connects early hard drives to an external controller, separating the drive mechanics from much of the control electronics. In the thread, the ST-225 is cited as a 20MB Seagate MFM drive. Participants compare it to a floppy drive model, where the drive needed a separate controller card rather than exposing a fully standardized high-level data interface on the drive itself. The thread also argues that integrating the controller into the HDD later accelerated hard-drive development. [#21077314]

Why can't I usually swap a hard drive PCB just by matching the model number, and when does the ROM chip also need to be transferred?

You usually cannot swap by model number alone because HDD electronics are often drive-specific. One post states that even two disks from the same production batch may fail after a plain PCB swap. In those cases, the ROM must also be moved because it carries individual programming data tied to that drive. The simple “same sticker, same board” rule stopped being reliable about 20 years ago, especially once per-drive adaptation data became critical for signal decoding. [#21072594]

In what situations can HDD PCBs from related models be interchangeable, and what label details matter most?

PCBs from related models can work when the underlying hardware family matches, not just the retail model name. The thread says boards from other related models often fit, but you must read the label carefully because one sticker can hide technically different designs, or identical hardware can ship under different names. The most important detail is whether the board uses the same signal processor family and matching firmware context. If not, a ROM transfer is still required even when the board looks physically compatible. [#21076170]

What is the Soft UnderLayer (SUL) in perpendicular magnetic recording, and why does it matter for signal processing?

“Soft UnderLayer (SUL) is a magnetic layer used in perpendicular recording that closes the field lines from the write head, but also adds electromagnetic interference that the read channel must handle.” The thread says SUL became important after perpendicular recording arrived, because it increased the need for individually programmed signal-amplification parameters, called adapters. That is one reason modern drives need per-drive tuning data and why careless PCB swapping fails even when the mechanical model appears identical. [#21076170]

How do ATA, Ultra ATA, ATA133, and SATA HDD interfaces differ in controller integration and transfer speeds?

The thread presents them as a progression in both bandwidth and integration. The 4GB Ultra ATA drive is described at about 4MB/s and still uses a separate interface/control area plus SRAM and ROM. The 40GB ATA133 example is said to be up to 10x faster than that 4GB predecessor and already collapses the board into three main chips. SATA then moves further by first using a distinct interface chip and later integrating SATA directly into the disk controller, reducing chip count again. [#21072338]

What changed inside HDDs when manufacturers moved from separate interface chips to a single integrated SATA controller?

Manufacturers removed one whole chip class from the PCB and pushed more logic into the main controller. In the first 80GB SATA example, the board still has a separate interface circuit, controller, SDRAM, and power chip. In the newer 80GB SATA generation, the separate interface chip disappears, the controller connects directly to SATA, and the electronics move to the side with the motor, actuator, and head contacts. That cut board complexity while keeping RAM and power handling separate. [#21072338]

How were early MFM hard drives with stepper-motor head positioning different from later linear-motor designs?

Early MFM drives used simpler mechanical positioning and later designs switched to more precise linear actuation. A participant recalls IBM 10MB to 40MB MFM disks where the head drive was a simple stepper motor. In slightly newer drives with linear motors, he saw a DAC0832 generating the head-position signal for the linear motor driver. Another user even remembered units with an external knob linked to the head drive, showing how visibly mechanical early designs could be. [#21072710]

Could LM358 or LM324 op-amps really be used in old 40MB HDD read/write electronics, or were dedicated head amplifier ICs required?

Dedicated head amplifier ICs were required for the actual read/write path. One user first suspected LM358 or LM324 devices on the 40MB PCB might amplify head signals, but another participant rejected that explanation because those op-amps were too slow. He states that special multi-channel read/write amplifier circuits were used for HDD heads. The later follow-up keeps the doubt alive only as a question, not as evidence, so the thread’s clearest answer favors dedicated head-channel ICs. [#21072710]

What is DVD-RAM, and how is it different from DVD-R and other writable optical discs for long-term use?

“DVD-RAM is a rewritable optical disc format that supports repeated file deletion and writing, and some versions were sold in protective cartridges.” In the thread, users contrast it with common DVD-R discs and note that a 20-year-old DVD-RAM still remains readable and writable. They also say early DVD-RAM media needed dedicated drives when sold in cartridges, while later loose discs could work in ordinary DVD recorders. That makes DVD-RAM sound more like removable storage than one-time-burn media. [#21084798]

How were Labelflash, DiscT@2, and LabelTag actually used in practice compared with LightScribe?

The thread raises them as niche labeling features, not mainstream storage workflows. One user asks who actually used Labelflash, DiscT@2, or LabelTag in practice and explicitly excludes LightScribe from the question. No one in the discussion then provides a real-world success case or workflow for those three systems. The practical takeaway is simple: unlike DVD-RAM, they appear in the thread as curiosity features for disc marking rather than important long-term data formats. [#21072679]

ZIP drive vs DVD-RAM vs magneto-optical disc vs MiniDisc — which storage format was better for removable data archiving?

No single format wins in the thread; each served a different niche. DVD-RAM is praised for surviving about 20 years with continued read, erase, and write capability. ZIP disks are described as removable drives handled like regular media. Magneto-optical formats appear as notable archival designs, and MiniDisc is cited as a still-liked Sony magneto-optical audio format. If you want the thread’s strongest archival vote, DVD-RAM gets the clearest durability endorsement, while ZIP gets the easiest-drive-handling comment. [#21084798]

How do I read data from an old HP USB streamer or DAT-like tape cartridge on a modern Linux or Unix system?

Use a tape-aware workflow rather than treating it like a normal USB disk. 1. Connect the HP USB streamer to a Linux or Unix system. 2. Use tape-oriented tools; the thread specifically suggests Unix/Linux and the tar command. 3. Expect streamer-style access, unlike ZIP or DVD-RAM, which behave more like regular removable drives. The key limit is usability: having the cartridge and drive is not enough if you lack the right tape workflow and software expectations. [#21092521]

Why are LTO tape libraries still used at places like banks, CERN, and CI TASK instead of relying only on HDDs or SSDs?

They remain useful because tapes are cheap per terabyte, durable in storage, and consume no power on the shelf. The thread says banks still use streamers, and another post cites CERN and CI TASK as examples of tape-library deployments. LTO-9 is quoted at 18TB native or 45TB compressed, which makes tapes attractive for large cold archives. The trade-off is speed and equipment cost: one user says copying can drag badly on old arrays with millions of files. [#21092549]

Generated by the language model.