Hi

Just put in a WD20EADS into my HTPC+server box (weird combination I know, it will be split out at some point in the future).

Just quickly, in terms of:

Quietness
At idle it seems just as quiet as the WD10EADS, by which I mean I cannot hear them with my ear next to the open door of my MiniP180 over air whoosh noise.
During activity.. sorry I just added it to the storage pool of the WHS so haven't heard it do anything just yet (need my other drives to fill up).

Vibration
This is the biggie - it has noticeably more vibration than the WD10EADS, more than the WD10EACS & more than the WD640AAKS. Definitely should be soft mounted or suspended. Luckily the MiniP180 soft mounts & loose coupling keep it from transferring vibration to the HD cage / case.

Temperature
The drive runs quite a bit hotter - about 3-4 degrees C - than the WD10EADS. I've also wedged it in between 2 WD10EADS which has increased the idle temps of all 3 drives. They 2 WD10EADS spaced out were spectacularly low previously - 21-22 deg C, this has gone up to 27 deg. So the WD20EADS appears to idle at 31 deg C in my setup. Might need to update this later as I observe their behaviour longer.

So in the future my plans of splitting out the file server and slowly adding an array of 10+ 2Tb drives still stands however I'm concerned about the vibration that the combined drives will produce & I'm not sure of any case that will accomodate 16 drives with soft/suspended mounting.

Note: the idea is for WHS to spin down the drives when idle but I've noticed that temperature monitoring stops the spin down. Going to make a separate post for this, maybe someone can help.

It's good to know the idle noise (or the lack of it) haven't changed.

Vibration is hit or miss attribute, so it's highly unlikely that your drive would be a typical WD20EADS. It could be badly handled individual, or there may be different models of spindle used in that model, or just manufacturing variation between spindle motors.

"The drive runs quite a bit hotter - about 3-4 degrees C - than the WD10EADS. I've also wedged it in between 2 WD10EADS which has increased the idle temps of all 3 drives. They 2 WD10EADS spaced out were spectacularly low previously - 21-22 deg C, this has gone up to 27 deg. So the WD20EADS appears to idle at 31 deg C in my setup."

You just revealed a likely cause to higher temperature there. As you already noticed, adding a third drive between the two old ones increase the temperature of old drive by 7 degrees, this means that airflow is quite low compared to proximity of the drives.

If the airflow in your case is so low that temperature rise of drives next to the new added drive is a noticeable 7 degrees, it would be natural to assume that the drive in the middle will naturally be the hottest as it's cooling will be affected by TWO other drives in proximity instead of just one. This alone would explain additional 3 degrees.

Other explanations would be that SMART temperature monitoring is not accurate. It may vary between individual HDDs (sensor accuracy variation, cheap components as they are not mission critical), and between models (potentially different sensor types or positioning).

___


This is just my guess, but I believe building a server with 2TB is better than using twice the number of 1TB drives. The fact that you got a bad unit once doesn't mean it's likely you get more of them. Vibration is really a hit-or-miss.
For example: my first 1TB 7200rpm Samsung F1 has very low vibration, on par or even less than my 1TB 5400rpm Greenpowers (both 1st and 2nd generation).
My second 1TB F1 is absolutely horrible. And my F2 Ecogreens are on par or better than my better F1 sample, thus way lower vibration than my Greenpowers.

So what doesn't add up? My Samsungs tend to have lower vibration than my WDs (which aren't bad either, just "not that special").. but that one sample just is absolutely horrible. There's no performance or reliability issue with it so far, just the vibration. It also happens to have a different colour spindle motor cap than all the other Samsungs in my posession, which probably means a different subcontractor for spindle motor, and a possible explanation. Samsung has had similar variation for ages. Back in P80 days, there was both NIDEC and JVC spindle motors. Nidec was quiter acoustically but may have vibrated more. JVC had a bad acoustic noise quality.

It may be "hit-or-miss" more so than idle whine due to the tolerances and speeds associated with a spinning platter, but in this case the increased vibration is most likely due to the extra weight associated with the 4th platter in the 2TB version. The previous EADS drives were all 3 platters: 1TB was 3x333GB and the 1.5TB was 3x500GB.

Extra weight = more weight in motion = more vibrational energy.

The extra weight also puts more load on the motor which would likely make it run hotter as well.

OP, I wouldn't worry about suspending 16 drives, you'll have a hard enough time fitting them all in any standard case, let alone keeping them cool. If it came down to it, you can get a case that has an all 5.25" front (like the Antec 1200), but you'd still have a hard time fitting 16 drives unless you were able to fit 4 - 4x3 drive enclosures. More than likely you'd have to go the rackmount server chasis route, or go with a 9U cube design (Lian-Li has one).

Beyond that, you'd be getting into some serious territory with storage, parity/redundancy, backups and etc, so I hope you know what you're doing.

"but in this case the increased vibration is most likely due to the extra weight associated with the 4th platter in the 2TB version"

But 4-platter 1TB (1st gen GP) don't typically vibrate that much either. Why would 4-platter 2TB (3rd gen GP) be doomed to vibrate due to same number of platters?

"Extra weight = more weight in motion = more vibrational energy."

Nope. If center of mass resides on spindle axis (platters are balanced) even with thousand platters = ZERO vibration.

This is why vibration is hit or miss. Of course 4 platter drive has more theoretical potential for vibration as there's more to be misbalanced, but misbalances can also cancel each other out.

"The extra weight also puts more load on the motor which would likely make it run hotter as well."

Nope. Weight has nothing to do with power consumption. Air resistance does, and air resistance is dependent on number of platters and heads. But given the same number of platters, power consumption would be pretty much the same whether they were made of aluminum or uranium.

You say yourself it's "hit or miss", we call this sample variation. I'm trying to explain why it might be different from other drives of this class



This is contradictory. While I agree with the perfect balance=zero vibration point, I think we can both agree that there is no such thing as a perfectly balanced rotating mass. Moreover, any variation will be represented in the vibrational energy of the spinning mass. The more mass in motion, the more kinetic energy which is translated into vibrational energy. It's simple physics.



I won't disagree about air resistance, though it's effect is necessary (it's what keeps the heads off the platter). However, that logic would say that putting more people in an electric vehicle wouldn't put more strain on the motor? No change in air resistance, just mass. Of course the engine will draw more power, and since no engine is perfectly efficient, the waste energy is radiated as heat.

I see no contradiction here, and I don't think it's a given that more platters will always mean more vibration. Yes, given an equal imbalance, more mass means more vibration. But the slight imbalances among each platter on a multi-platter drive are not likely to be lined up to create the maximum vibration. In fact, the more platters you have, the more likely it is that the imbalances will offset. With average luck, this mitigates the effect of the extra mass. But it should also lead to increased variability, as some drives will have nearly perfectly offsetting imbalances, others will have a worst-case alignment, and most will be somewhere in between.


The added mass would be significant for a car because it must accelerate and go up hills. But in a hard drive, acceleration should occur only when the drive spins up. After that, it's just compensating for friction, hence mass doesn't matter.

The vibration is all transferred to the spindle making the whole array vibrate, along with the casing. The other consideration is that the extra platter requires an extra head. The heads move very rapidly, which are also another source of vibration (maybe even larger, given the counterweights).

As for the car analogy, try riding a bike over a flat road for 20 miles. Then try doing it with a 300-pound weight on your back. Aside from the discomfort of the weight, tell me which one is harder to do. Inertia only takes you so far; gravity increases tire friction and provide resistance to motion.

However, I fear we've strayed too far from the original topic. To the OP, is the vibration from the drive itself audible?

Obviously.


Agreed. Thus programming for gentle seeks (AAM) would pay larger dividends when there are more platters.


Here we also agree. I considered mentioning the effect of tire/bearing friction myself, but decided not to because the car analogy was introduced with "just mass". I would expect the added air friction and the drag to fly the extra head would be a larger factor than the increased friction in the bearing, but I could be wrong.


[/tangent]

jhhoffma: "However, that logic would say that putting more people in an electric vehicle wouldn't put more strain on the motor? No change in air resistance, just mass."

Power consumption of an automobile is a factor of many thing. Driving a car at constant speed, energy is lost to heat by means of air resistance and rolling friction of tires. Air resistance does not change with mass, rolling friction does. However, rolling friction is meaningful only because tires are soft and produce heat as they deform under weight of the car. Add more weight, add more deformation. If you think of trains, their rolling friction is minimal and air resistance totally dominated loss of kinetic energy. HDDs have metallic bearings so I think rolling friction doesn't have that big of an effect.

Cars are typically driven in non-constant speed. This means that during each braking, energy is converted to heat in brakes. Extra power is required to return the kinetic energy that is lost and regain lost speed. Kinetic energy = speed^2 * mass. If you keep speed constant, adding mass doesn't waste energy after the initial spin-up.

jhhoffma: "extra platter requires an extra head. The heads move very rapidly, which are also another source of vibration (maybe even larger, given the counterweights)."

Heads are extremely light. It's the actuator arm that weighs a thousand times more. And when a head is removed, it is usually replaced with ballast.

The only way to remove moving weight of actuator is to replace the whole actuator with one that has less arms. This usually done only for the entry-level HDDs which typically have otherwise different construction from flagship capacity drives. Downscaling from 4 platters to 3 or 2 while remaining in the same design of flagship usually doesn't reduce mass of actuator even a little bit... only the mass of spindle. (And naturally spindle assembly mass is only reduced a little bit as most mass is in the motor. For vibration, the motor itself resides close to rotational axis and isn't as big source for imbalance as the light-weight platters are.)

jhhoffma: "As for the car analogy, try riding a bike over a flat road for 20 miles. Then try doing it with a 300-pound weight on your back."

For energy consumption, air resistance didn't change now, or did it? You're talking about a bike now, no longer a car. And of course the rolling resistance wastes energy. Replace the tires with solid steel rims and it would probably require quite a bit less force to keep it moving. (It would be pain in the arse with the bike bouncing around all the time, and you would have trouble not falling down all the time.)

Also, 300 pounds of weight add weight high above the ground, making it difficult to keep your balance. This is not analogous to adding weight to a HDD as HDD is not really tipping over in any direction.

When you ride the bike, gyroscopic effect of rotating wheels help you stay in upright position... or more accurately, it slow down any tilting, giving you more time to react, making riding bike at high speed easier than at low speed. If you add weight to the cargo tray over rear wheel, you make the bike unstable as you can't control this weight by shifting your own center of gravity. Also, you add overall mass of the bike without adding weight to gyroscopes (=wheels). Should you add 300 pounds of weight to wheels of the bike (by installing uranium rims or whatever) the bike would be heavy to accelerate but once in speed, it would be very easy to drive straight at any speed and more stable to handle even at very low speeds.

jhhoffma: "is the vibration from the drive itself audible?"

It depends whether you are able to hear 90 Hz acoustic noise. Because that's exactly what spindle imbalance causes on 5400rpm HDDs. Every single 5400rpm HDD. Only amplitude varies.

The question is rather useless as there's very different background noises and we would need to know the distance at which it become audible. Also, higher acoustic frequencies coming from the drive (even if ambient was totally silent which it isn't) might mask 90 Hz because human ear isn't quite that sensible for subwoofer-like frequencies (we mostly sense those frequencies in out bones, after all, not ears).

My bet is that with any HDD 5400rpm or 7200rpm you tend to hear acoustic frequencies more clearly than vibrational ones. (Assuming HDD is suspended and vibrational frequencies are thus not amplified by external surfaces.) The only way to hear the airborne low frequencies is to listen at a very close distance where barrier effect becomes main defining factor. The closer you listen, the more the higher pitched noises start to interfere and cancel each other out, leaving only low frequency noise non-distorted. (This is why measuring HDD noise at a few millimeters it ridiculously inaccurate. They basically measure HDD vibration only. Also, close-distance measurement tends to measure emitted noise per surface area, not the total noise emitted. Measuring too close will make 2.5" HDDs appear about as noisy as 3.5" HDDs.)