Rackmont Computer Shock and Vibration
Acceleration – rate of change of velocity with time. Usually along a specified axis, usually expressed in “g” or gravitational units. It may refer to angular motion.
Amplitude – the maximum displacement from its zero value position.
Compression – when specified as a direction for loading – a deformation caused by squeezing the layers of an object in a direction perpendicular to the layers.
Damping (c) – the mechanism in an isolation system which dissipates a significant amount of energy. This mechanism is important in controlling resonance in vibratory systems.
Disturbing frequency (fd) – the number of oscillations per unit time of an external force or displacement applied to a vibrating system. fd = disturbing frequency.
Durometer (hardness) – an arbitrary numerical value which measures the resistance to the penetration of the durometer meter indenter point; value may be taken immediately or after a very short specified time.
Fragility – is the highest vibration or shock level that can be withstood without equipment failure.
“G” level – an expression of the vibration shock acceleration level being imposed on a piece of equipment as a dimensionless factor times the acceleration due to gravity.
Isolation – the protection of equipment from vibration and/or shock. The degree (or percentage) of isolation necessary is a function of the fragility of the equipment.
Load deflection curve – the measured and recorded displacement of a mounting plotted versus an applied load.
Natural frequency (fn) – the number of cycles (expressed as Hertz or cycles per second) at which a structure will oscillate if disturbed by some force and allowed to come to rest without any further outside influence.
Random vibration – non-sinusoidal vibration characterized by the excitation of a broad band of frequencies at random levels simultaneously.
Resonance – A vibratory system is said to be operating at resonance when the frequency of the disturbance (vibration or shock) coincides with the system’s natural frequency.
Set – is the amount of deformation never recovered after removal of a load. It may be in shear or compression.
Shear – when specified as a direction for loading – a deformation caused by sliding layers of an object past each other in a direction parallel to the layers.
Shock Pulse – a shock pulse is a transmission of kinetic energy to a system, which takes place in a relatively short length of time compared to the natural period of this system. It is followed by a natural decay of the oscillatory motion. Shock pulses are usually displayed as plots of acceleration vs. period of time.
Spring rate – is the force required to induce a unit deflection of spring. A steel spring has a very linear relationship between force and deflection. Elastomeric springs may or may not be linear depending on the amount of deflection due to the load.
Static deflection (ds) – the deflection of the isolator under the static or deadweight load of the mounted equipment.
Transmissibility (T) – is a dimensionless unit expressing the ratio of the response vibration output to the input condition. It may be measured as motion, force, velocity or acceleration.
Introduction
You should not shock mount disk drives! This tutorial explains why.
One of the great marketing gimmicks on the part of all the industrial computer manufacturers is shock mounting disk drives. This is a waste of money and actually exposes the drives to higher peak loads than if the drives were simply screwed directly to the chassis. Drive manufacturers have made great strides in bullet proofing their devices and the g-limits are usually very high, on the order of 100G’s. The purchase price for shock mounts is very low so an attitude of “why not” is easy to understand. After all, when your competitor advertises “Shock Mounted Drive Cage”, you had better add it to your specs to stay competitive. However, the true cost to the user in added labor to install a drive (4 mounts, 4 washers, 4 inserts, 4 special screws) or replace a missing grommet or insert is much higher than the simple cost of the components. In addition, the resonate frequency of a shock mounted drive is right in the middle of most chassis manufacturer’s vibration specifications (67Hz natural frequency versus a chassis spec of 5-100Hz). Ask your chassis vendor for their testing data or engineering analysis for their shock mounts. I doubt they will be able to provide it to you.
So why do all industrial chassis provide shock mounts. Every industrial chassis is very much the same. Same layout. Same power supplies. Same look. This is like the lemonade stand. Everybody sells lemonade. How can your lemonade be different than your neighbor’s? So they all compete on price alone. Until somebody adds a cherry. Now his lemonade is different. He gets to charge more. Same with chassis. Everybody is looking for the cherry to add. Soon, they all have cherries. Same chassis. Same specifications. Same shock mounts.
Shock is defined as a single-pulse event such as dropping a package onto a concrete floor. Vibration is defined as a continuous sine-wave motion subjecting the product to continuously varying g-loads along one or more axis.
Shock affects components differently than vibration and must be looked at separately. A screw will not come loose because a chassis is dropped once. However, plug-in I/O cards or connectors may be knocked loose from the high impulse g’s from the shock. Low level vibration may not unseat the plug-in boards, but a capacitor may be resonate and break the leads as it vibrates back and forth. Shock mounts work by isolating the chassis or the mounted component from the motion of the surrounding structure. The best mount would be made from a very soft material, allowing inertia to hold the device in place while the surrounding structure moves. If you have a CD player in your car, notice how much the inner workings move in relation to the outer housing. Any bumps cause CD’s to skip so they try to allow the car to bounce around while holding the CD mechanism steady. But there must be room for the relative motion between the case and the inner parts. If you hit a big bump and bottom out the inner mechanism, you will get skips in your music.
Industrial shock mounts work the same. They provide a soft interface between the drive and the chassis. Or between the chassis and the rack. Or between the rack and the vehicle. But the only way these work is to allow room for the surrounding structure to move relative to the protected device. This leads to a fine balance of limited protection versus limited relative motion.
Shipping, by far, is harder on chassis than anything they will be subjected to in service. Shock is of most concern during shipping. A researcher once shipped a several recording G-meters to determine the load packages are subjected to in shipment. Several came back broken!
Rarely are computers subjected to significant shocks after they are installed. The first place to start with shock protection is with the packaging. Properly engineered foam inserts in a heavy cardboard box will provide the best protection. Styrofoam `peanuts’, crushed paper, or foam-in-place are not adequate and will not protect the chassis. Every city has packaging vendors that can provide a box and packing material suitable for the chassis dimensions, weight, and shipping method. Shock is attenuated over distance. Therefore the packing material has to have a certain give to allow the chassis to move within the confines of the box. Too much give and parts of the chassis poke holes through the box and get damaged. Not enough and too much of the shock is transmitted to the chassis. Part of a package qualification is a drop test with g-sensors mounted on the chassis to measure actual packaging performance.
The next step is to make sure nothing is packaged inside the chassis. Often there are manuals, cables, etc., which must be shipped with the system. These are often put inside a cardboard box and tied or taped inside the computer. This simply becomes a loose missile the first time the chassis is subjected to any kind of shock, then merrily careens around inside the chassis as the truck bounces down the road.
The two biggest problems from shock are connectors coming loose and I/O cards coming unplugged. Both these problems are easily solved as discussed below. Modern disk drives are very shock resistant, especially in a non-operating mode. While disk drives should not be dropped or mistreated, the non-operating shock specification for modern 3 ½” drives is 75G’s for Seagate and IBM and 150G’s for Western Digital. Operating shock specifications go to 10G’s. A chassis will not be exposed to these kinds of shocks if properly packaged. Shock mounts will attenuate some of this load. However, a chassis may also be subjected to vibration and the shock mounts will increase the loads seen by the drives. This is explained in the following section. With the capability of today’s drives, does shock mounting provide any benefit?
Most commercially available I/O cards and drives do not include locking connectors. A simple alternative is to glue the connectors in place with electronic compatible (no acetic acid) RTV adhesive. Hot melt adhesive also works well. Card hold-down brackets can provide a seating force on I/O cards to assure they stay fully seated in the motherboard connectors. Obviously the screws used to secure the cards in the chassis should be installed and tightened. Cables should be tied down so they can’t pull on the connectors. Unused drive connectors should be secured so the exposed pins can’t accidentally touch any components when the chassis is powered up – 12 volts is very hard on 5 volt circuits. PC Power and Cooling (619-931-5700) sells covers for drive power connectors.
If the system is being shipped overseas, it is a good idea to remove the I/O cards and drives and pack them in separate boxes. The cards and drives can be easily reinstalled when the system is put in service at the remote site. This also works well domestically and provides additional assurance the product arrives at the customer’s site in working condition.
The typical shock mount used by most chassis manufacturers is a thermoplastic grommet manufactured by E.A.R. The material these are made from is different than rubber in that the material provides damping. Rubber can be thought of as simply a spring while an EAR shock mount provides a spring with a shock absorber built in. They allow some motion, but quickly damp out oscillations.
Other shock mounts are available including damped springs, simple rubber, wire mesh, etc. Very few selections are available for small loads and E.G.. has a good selection. For heavier loads, there are a variety of mounts available from different manufacturers.
The selection criteria for a shock mount is based almost entirely on the weight of the mounted device and the anticipated vibration frequency to assure the natural frequency is well below the expected vibration frequency.
Problems with Shock Mounts and Shock
Shock can only be attenuated over distance but drives are typically mounted adjacent to each other in a cage. If drives are spaced apart to provide room to move, there are unsightly gaps between the drives the marketing people don’t like. These gaps also make good places to loose floppy disks into when you miss the drive. So the chassis manufacturers place the drives next to each other. They also place some kind of front panel around the drives, also mounted in close proximity. Shock mount grommets are available in with varying stiffness and are selected depending on the weight of the mounted device. However, there is a limited selection leading to several compromises in selection. If a lightweight drive is placed next to a heavy drive, they will move relative to each other during the shock event. If there is just a small gap between them, they will collide during the shock. This collision leads to a very high, very short duration shock directly input into the drives. Usually the lighter drive will suffer. Shock mount grommets are also very temperature sensitive. For one commonly used thermoplastic material (EAR C-1002), a base stiffness or a correction factor of 1 is rated at 70 deg F. At 45 deg F., the correction factor is 8.6 meaning they are 8.6 times stiffer. At 140 deg F., the correction factor is 0.15. A chassis being shipped in cold weather will perform very differently from a chassis installed in a warm environment and in operation.
Vibration will effect different parts of a chassis in different ways. If an electrical component (capacitor, inductor, etc.) is vibrated at its resonate frequency, it can fatigue the leads and break. This often happens to capacitors in the power supply. Any unsecured component will have a resonate frequency. The heavier a component is, the lower its resonate frequency. Cables can vibrate or sway and fatigue the conductors creating an open in the wire. In extreme cases, screws can come loose or boards may become unseated.
If a rackmount computer chassis is to be installed in an environment where vibration is a concern such as on aircraft, ships, trains, etc., or in a factory where the building or floor is being shaken by nearby machinery, then a systematic approach to vibration control will be most cost effective. There are obvious preventative measures such as securing the cards and cables and making sure all screws are tight. The power supply should be examined for loose components (capacitors, wires, inductors, etc.) that can be glued to other components to limit their motion. Plug-in I/O cards do not usually have components with long leads because of the limited space available between cards.
An often overlooked source of problems in a vibrating environment is the rack mount slides. Vibration analysis should include the rack and chassis dynamics and the slides in particular. Standard chassis slides have a lot of play in them. When the rack moves, the chassis is first suspended in space. Then the slide bottoms out, metal on metal, and transfers a high peak shock load to the chassis, instead of the slowly building sine wave of the surrounding rack. One company providing equipment to drilling rigs had to provide expensive custom slides with zero backlash to prevent system damage during vibration testing. In extreme cases, the slides can be ripped off the chassis from the high loads.
After these measures are accomplished, that leaves the disk drives to be protected. Several chassis manufacturers provide shock mounts for the drives. However, shock mounts were introduced to improve the chassis features list and not because rigorous engineering analysis dictated their use. Unfortunately shock mounts only subject the drives to higher loads in a vibrating system than if they were not installed. The following information will clarify the fundamentals of vibration when applied to shock mounted drives.
As you read this section, keep in mind a vibration damping system provides no or minimal benefit until the natural frequency is well below the vibration frequency, typically 1/3 or less. If the natural frequency of the mounted device is above the vibration frequency, the mounts are too stiff and provide no isolation for vibration, but do provide some shock attenuation.
Every flexible system has a preferred or sympathetic frequency known as the natural frequency where the system will freely resonate. This is the fundamental mechanism behind watch crystals and pendulums. Driving a system at or near its natural frequency will lead to high amplitude excursions limited only by some physical restraint. Disk drives mounted with shock mounts form a flexible system with a well-defined and easily calculated natural frequency. If the chassis is subjected to a vibration with a frequency near the natural frequency of a mounted drive, the drive `absorbs’ energy and will start moving until something physically limits the motion, usually adjacent drives or the surrounding structure. The suspension bridge over the Tacoma Narrows is a prime example of a system in resonance leading to failure.
EAR Corporation (317-692-1111) manufactures a broad line of highly damped Isodamp Grommets that are used by most chassis manufacturers in their shock mount systems. Their color is bright blue. These are not natural rubber, but are a specially formulated thermoplastic which is highly damped, exhibiting extremely low amplification at resonance and quick return to system equilibrium after shock or vibration input. EAR publishes a good guide “Designing with Isolators” which details these mounts and provides information for determining natural frequency. The formula for Natural Frequency is:
where K is a spring constant or stiffness and W is the weight. The stiffness K is derived from a series of table and curve look-ups from the EAR literature and is dependent on the load on the mount, the material the mount is made from, and its temperature. You can see that either a lower stiffness (K) or a higher weight (W) will lead to lower natural frequencies. As an example, consider a typical 1Gigabyte 3-1/2″ hard drive weight about 1.3 pounds. The drive will be mounted using four EAR mounts. The appropriate mount is the G-427-1. This mount provides for a recommended maximum static load of 1.5 pounds. Going through the calculations yields a natural frequency at room temperature of:
The next factor to consider is temperature compensation. The above calculation is for a standard temperature of 70 deg F. The EAR mounts are highly temperature sensitive with a correction factor ranging from 8.58 at 45 deg F to .16 at 125 deg F. The temperature correction factor is simply multiplied by K, the spring constant, to arrive at a corrected stiffness. This leads to the following curve:
This curve shows the natural frequency for various temperatures. Note that the natural frequency varies from a low of 67Hz at 125 deg F to 493Hz at 45 deg F. As a reminder, most chassis manufacturers specify a vibration limit of 5-500hZ.
If the calculations are done for a large drive weighing 6 pounds, the natural frequency ranges from 23 to 174Hz over the same temperature range. Note the mounts have a recommended maximum static load of 1.5 pounds each. This would put the resonant frequency right in the middle of several chassis manufacturer’s vibration specifications. At room temperature, the driving frequency would have to be in excess of 450 Hz before these mounts provide much benefit. The most common problem with vibration is resonant vibration of electronic components, usually capacitors, when they are mounted parallel to the chassis motion. Similar to shock protection, a simple fix is a generous application of some kind of adhesive to brace the component against movement. The best approach is imperial; shake the chassis until something breaks, then secure those components and shake again.
Drive Dynamics as Related to Shock Mounts
As you can well imagine, a lot happens inside a hard drive to allow reliable writes and reads. The platters are spinning at 10,000 and higher rpm. The read/write head is zipping back and forth writing to the drive and reading data back. In all, a dynamic control nightmare. Part of the magic in a drive is in the control algorithms. These algorithms compensate for the flexibility of the actuator arm, vibrations in the platters, shock and vibration imparted on the drive housing, and the signal latency. If you change any of the assumed physical factors in the equations, the control algorithm falls apart and the drive stops working.
The drive manufacturer makes an assumption the drive is rigidly mounted and the housing is not moving. If you mount the drive in shock mounts, you change the operating conditions of the drive in a manner the drive engineers could not predict. The housing is now able to move and is moving!
There are two problems affecting drive reliability; actuator arm induced movements and self induced vibrations.
As the actuator arm moves, the drive will tend to move in the opposite direction. The control algorithm expects the drive to be rigidly mounted and does not expect this motion. Thus there is no compensation for the media displacement. The drive servo puts a force on the actuator arm for a determined time to move the head to a certain location in space, expecting the desired track to also be there. Unfortunately, the track has moved because the housing moved. Thus the heads are not in the right location on the platter when they read or write. To complicate matters, the worst motion you can impart on a drive is a housing rotation around the spindle because that changes the rotational position relative to the heads. The heads will be reading or writing from a position either too early or too late on the track instead of at the expected spot.
In addition, the platters are not perfectly balanced and will cause some vibration. AT 5400 rpm, this works out to a 90Hz vibration frequency. From the above discussion on shock mounts, you can see that 90Hz is potentially a fundamental frequency of the drive at normal operating temperatures. As such, a very small vibration amplitude can be amplified into relatively large motions leading to the same failures as discussed in the previous couple of paragraphs.
An example of these effects can be seen by simply setting a drive on a neoprene mouse pad. There is a very good chance the drive will start generating errors because it is now able to move. You can actually feel the drive moving as it works.
Where Vibration Control Should Be Implemented
The two variable factors in vibration control are mounting stiffness (or give) and weight. The place where both can be applied to greater effect is to the entire rack. There is generally room for a rack to move, and there is more weight to work with. A variety of high load mounts are available from Korfund Dynamics (201-838-1780) and they have a very effective catalog. Three axis shock and vibration control can be implemented by using these mounts to secure the rack to the surrounding structure. This type of mounting scheme will attenuate vehicle motion limiting the loads seen by the installed computers and other electronic equipment. In these installations, no additional work may be needed to allow the electronics to survive the environment.
An alternative to shock mounts where hard drive failure is a concern, is to use 2-½” drives as used in notebook computers. These are very tough drives with high shock and vibration specifications and control algorithms optimized for mobile systems.
To summarize, existing shock mounting technology as employed in industrial computers is poorly implemented. While adding shock mounts to the drive cages is easy and makes for a great `feature’, Chassis Plans has elected to remove this option. You, the client, are much better served by providing system shock mounting, isolating the chassis rack, designed for the particular rack weight and environmental dynamics. For those clients in high-shock or vibration environments, Chassis Plans can provide designs for shock mounted components appropriate to the specific situation.
/>