<:-? Hard Disk Drive Tech Overview +++++++++++++++++++++++++++++ | MAGNETIC HARD DISK DRIVES | +++++++++++++++++++++++++++++ Significant improvements in manufacturing techniques and technologies has led to the availability of physically small, high storage capacity disk drives. In 1967 a state of the art disk drive had 50 rotating surfaces, platters 24 inches in diameter, an access time of 600mS and a total capacity of 5MB! In contrast, todays most popular products are 2.5-3.5 inches in diameter with inbuilt intelligence and capacities in the hundreds of Mega-Bytes. The advances in technology that has enabled the physical size to be reduced to todays dimensions have included: MATERIALS ========= - Very powerful magnets, made from rare metals - Control of chemical outgassing/contamination - Improved adhesives with predictable characteristics - Careful choice of materials to ensure compatibility DESIGN ====== - High level of intelligence built in through use of Micro-Processor controllers - VLSI and SMC to provide very compact electronics packages - Advanced signal processing techniques for data recording - Power requirements minimized by careful circuit design and choice of components - Huge increase in reliability through predictive modelling, component choice and component testing - Innovative mechanical design to reduce tolerance dependence - Magneto resistive and thin film heads improve the signal to noise ratio and increase areal densities MANUFACTURING ============= - Early involvement in product design ensures ease of assembly - Consistent quality through rigid process control limits - Close cooperation with suppliers to maintain high quality - Closed loop failure analysis for defective units ensures continuous process improvements +++++++++++++++++++++++++++++++++ | BASIC PRINCIPLES OF OPERATION | +++++++++++++++++++++++++++++++++ HEAD POSITIONING ================ Many disk drives have used stepper motors to control the position of the read/write heads. But as the capacity of a drive increases, the spacing between data tracks invariably decreases. This means that the heads have to be positioned more accurately. Stepper motors can no longer cope with the very small positional changes required to maintain the data head over the center of the track--it is an open loop system; ie, there is no feedback signal to provide a fine positional adjustment. Voice coil actuators are used to position the heads in high capacity drives because they form part of a closed loop system. As the head nears the data track which contains the desired information, a signal is generated from prerecorded data, which tells the positioning system how far the head is away from the center of the track. The servo loop moves the head until the position error signal has decreased to zero and keeps it at zero until a new access to a different data track is required. DATA RECORDING ============== There have been numerous techniques used for recording data, from the early Non Return to Zero (NRZ) to Modified Frequency Modulation (MFM) which was used on many drives during the late 1970s. Most of the current popular recording techniques use a form of data encoding based upon a pre-determined set of bit patterns. The encoding techniques is known as run length limited (RLL) and it can be implemented in various forms. The one used in IBM OEM drives tends to be the 2,7 RLL code, this means that the data pattern follows a protocol that allows for a bit stream of 2 zeros (minimum) of 7 zeros (maximum). This encoding method provides superior performance in terms of data recovery but does increase circuit complexity somewhat. In a typical drive the very low amplitude signals, picked up when the read circuits are activated, can easily be affected by external events such as mechanical shocks or vibration or electrical noise. These phenomenon usually show up as random read errors which become apparent when long data streams are being read. To ensure that the read signals are not corrupted by external means, it is important to examine carefully the operating environment that the drive will be used in. Errors can be introduced into the recorded data as it is being written in exactly the same way as described above except that when the data is read back it is most likely to show up as a hard error. Careful evaluation of the mechanical mounting arrangements, grounding system and the power supply will minimize problems of this sort. PACKAGING ========= During the development of IBM OEM disk drives the level of shock and vibration that each product will withstand is carefully established to ensure that the product will continue to function to its full specification for all of its expected life. If disk drives are subjected to shocks or vibrations which are in excess of the levels permitted, errors can be expected to occur. OPERATIONAL SHOCK occurs while the disk drive is powered on, depending upon the severity of the shock (or vibration), and what the drive is doing at the time will determine the damage suffered. For example, if a screwdriver handle is accidentally dropped onto a drive that is performing a seek operation, it will quite likely cause the head(s) to momentarily touch the disk(s), resulting in a small area of surface damage which in turn will cause data errors. The disk enclosure in higher capacity drives usually has a large mass, which must be allowed to move freely in any direction. Shock mounts are designed in provide a high level of isolation between the disk enclosure and the frame. If any of the shock mounts becomes jammed (for example, perhaps a mounting screw which is too long has been used during installation) and vibrations, due to cooling fans etc, will be transmitted through the body of the drive and will affect the ability of the data heads to follow the track properly--this will again give rise to data errors. On smaller drives, the disk enclosure, in fact the whole file, has so little mass that it is not necessary to fit shock mounts. NON OPERATIONAL SHOCK occurs when the drive is powered off. Damage sustained in this way is usually caused by inadequate packaging or poor handling disciplines. The fragile, high tolerance, mechanical components found in the modern disk drive can be permanently damaged by someone carelessly dropping a drive onto a work surface, even from a height of a few inches. Likewise, inadequate packaging, which has not been designed to withstand the rigors of road/rail transportation will not protect the drive against the shock levels experienced when a delivery man throws the box containing the drive into the back of his van. Once the parcel delivery service has got the box, it will be well tested by the time it reaches the customer! Modern disk drives will work for many years, trouble free, if they are handled with care and respect. Damaged sustained by poor handling techniques is a pure waste of your money! To avoid using inadequate packaging, it is possible to purchase single pack boxes from IBM OEM, or have your packaging tested to ensure it will meet the required performance. POINTS TO REMEMBER ================== - Do not remove drives from the box until they are needed. - Do not stack boxes containing drives too high. - If you are not sure how good your packaging is, have it tested. - Conductive rubber mats should be fitted to all surfaces where unboxed drives are stored. - Only use qualified packaging when transporting drives. ELECTRO STATIC DISCHARGE ======================== State of the Art electronic design has given us circuits which require extremely low power level to function. It is not unusual to find circuits which will work for years with only a tiny battery to power them. When electronic designers are confronted with the various problems of how to make circuits physically smaller, one particular aspect occurs time after time; ie, how to get rid of heat. The more amps and volts that a circuit consumes, the more heat there is to dissipate. By reducing the thickness of the semiconducting layers, which each transistor is made up from, it is possible to use less voltage and current to perform a given function. The individual links between each circuit element can be reduced in size because of the overall reduction in power requirements and so integrated circuits keep getting smaller, or the complexity in a given chip size keep increasing. Is everyone happy?...No, we have a problem! If you build a circuit that is both physically tiny and of very low power consumption, and then subject it to a very high voltage discharge, it will suffer damage. The individual connections between the various components within the integrated circuit will suddenly be required to conduct much more power than they were designed to. Power generates heat and the device will develop hot spots: areas where the power is concentrated in several components. Heating circuit elements beyond the design point reduce the reliability of the components and the interconnections which join them together. So, by applying a high voltage discharge to a modern electronic circuit we are most likely going to damage the long term reliability; the individual components that were stressed will fail long before those that were not. If a particular components absorbs a lot of energy from a discharge it will literally blow apart on a microscopic level. The individual layers that make up a transistor or diode can be punctured by the high power flowing in the component during the discharge. The industry has investigated failures due to ESD damage. Post mortems have been carried out to see what physical damage has been done and in most cases it is very similar to what happens during a lightning strike--except of course we are looking at a microscopic level. REMEMBER, when YOU pick up a modern disk file, from the bench or out of its' box, you are handling a device which probably has close to 250,000 internal circuit connections. If you have walked across the workshop, you are probably charged up to 8000-20000 volts. What do you think the chances are that you can avoid damaging something in the drive?