Post
by whiic » Wed Aug 05, 2009 3:45 am
MikeC (about Shamgar's problems): "If the suspension was really loose -- like the ride in an old 60s Caddilac -- then I could see some problems"
I don't think that'd be a problem. From HDDs perspective, laying the HDD on foam or rubber suspension is all the same, no matter how soft the suspension is. This is because all decoupling methods are VERY soft to begin with. Seek that lasts 10ms duration is just that: a 5ms acceleration pulse followed by 5ms of decelarion. This 0.005 seconds is used to wind the suspension. During 0.005 seconds of deceleration, this stored energy could affect where the seek ends. But how much energy can be stored to soft decoupling with a displacement of less than 1 millimeter and a spring force that is a fraction of a gram... Not much.
Any effective decoupling has a vibrational frequency far below that of seeking (which is in range of 50-200 Hz depending on seek distance). Because soft suspension cannot really store vibrational energy of this high frequency, the biggest potential problem could be lack of mass attached to HDD cast. HDDs may rely on cast remaining stationary when arm is moved to perform the seek. If cast accelerated to the opposite direction, seek distance could end up overshooting (if HDD designed relied on spring force to "bounce" HDD cast back to it's original position before seek ends). This would be a problem even if HDD was magically floating in mid-air with no spring force of decoupling.
There was a study, made by Seagate (if I remember correctly), that concluded that semi-rigid mounting is far worse than soft-mounting or rigid mounting. Since "rigid" in that study meant bolting it into a few kilograms of metallic mass, this mounting isn't really realistic for computer cases...
The worst case scenario is a semi-rigid mount, like "hard-mounting" on 0.5mm thick steel case. Yes, the mounting which we consider secure, is the worst-case scenerio for seek stability. The fact that majority of computer cases absolutely love to resonate at 120Hz doesn't mean the mount has 120Hz vibrational frequency, merely that the side panels, or any pieces coming into contact with them, do. Still, even though being able to hear non-decoupled vibration is not conclusive evidence of HDD mount itself amplifying 120Hz, it think it's still more reasonable to assume hardmount's vibrational frequency is clores to 120Hz than softmount's vibrational frequency which is probably in range of 1...50Hz.
Less than 1 Hz for extremely loose rubber band "cradle". The type we don't really use because HDDs are so loose they may actually fall down accidentally.
Up to 5 Hz for decoupling with tightened rubber bands.
10 Hz with foam.
Up to 50 Hz with rubber softened semi-hard-mounts.
These are of course just guesstimates of mine. Place the HDD in it's suspension and push it sideways or rotate it... and let it unwind. Especially with rubber band decoupling, you can see it bounce back and forth repeatedly in very low frequency (that doesn't correspond to anything induced by drive itself). Of course, if you do such testing, don't do it with computer powered on.
Conclusion: "Hard"-mounting HDD in computer case = about equally bad.
Small rubber pieces for decoupling purposes = ineffective (may reduce vibration-borne noise a bit, but may actually worsen the HDD's own vibration more than proper decoupling and more than "hard"-mounting).
The best possible solution for random access performance would be 20 kilograms of lead bolted to HDD. With that much metal, decoupling it like sandwich would be impractical but also unnecessary as 20 kg really can't vibrate enough to be audible. For practical silencing, any solution having vibrational frequency <20 Hz should have very little wind-up effect. And to reduce risk of overshooting seeks due to HDD casts free rotation (at least in microscopis scale, of course it can't rotate 360 degrees in any suspension), sandwich it with <500 gramms of metal... (<100 gramms for laptop HDDs). More than that would be definite overkill.
Did it make any sense? I kinda started ranting...
Anyway, I don't think it's vibration issue. If decoupling with rubber cause bad access time compared to decoupling with foam, I'd probably say it's due to WHERE the HDD was mounted (HDD or CD bay compared to computer case floor) and EMI (ElectroMagnetic Interference) between computer components. Decoupled HDDs aren't properly grounded with metal screws and all grounding is done via power connector. Adding additional grounding strip from HDD cast to computer case may help.
Another EMI related explanation could be related to Faraday cage. If your HDD works ok inside HDD tray for example, it may be that the tray is shielding the HDD from external electric, magnetic and electromagnetic fields, as opposed to decoupling in CD bay which have no metallic material blocking EMI coming from direction of motherboard. You could try decoupling the HDD again, but placing a metal grille around the HDD, making sure the grille is properly grounded to computer case. Also verify that your PSU is properly grounded to computer case!
I honestly doubt it's anything mechanical. It's probably not due to mechanical decoupling but about electrical decoupling. Also, if the drive is failing, it may be due to some component tolerances that make them more vulnerable to EMI than working HDD specimens. But before you conclude HDD in any way faulty, check grounding issue, run SMART tests, etc.
If problem is really ONLY with seeks, then it could possibly be mechanical as well. Then I'd like to hear more about how tight you made the rubber decoupling, etc. Also: to conclude it really is related to seeks, do synthetic random access tests. You can start the test with HDD decouple with rubber, and let the test run... then grab the HDD by your fingers trying to reduce vibration and see if seek time improves. This should keep EMI and grounding about the same as your hand don't significantly affect either. (Just make sure you're not carrying any static electricity while touching a HDD, especially when it's running.)