There is a noise comparison test for 2.5" drives here:

http://techreport.com/articles.x/9378/14

Difference is about 3 decibels between 5,400 rpm and 7,200 rpm. Is that audible in a very quiet desktop system? Can anyone actually hear the difference?

The dB scale for sound is such that 10dB = 2X loudness. This means that 3dB = 23% increase in loudness. I think a 23% increase in something tends to be noticeable. However, it really depends on the situation. The louder drive may still be quieter than the ambient environmental noise -- this would certainly be true in an office environment or if you lived on a high traffic road (without super good windows). It also depends on the quality of the noise. Hearing is quite selective and you can detect sounds that are well bellow the noise floor by concentrating (this concentration can happen subconsciously if the sound is one you find particularly unpleasant).

Not taking any weighting filters into account, remember that 3dB is the magical number. In Fletcher-Munson type compensation, 10dB is translated to twice the original loudness in the human brain. However, doubling of intensity always causes an increase of approximately 3dB in decibel's logarithmic scale.

This however is only information on paper and should not be taken too seriously. Remember that ambient noise varies greatly from place to place just like the human perception of sound varies from person to person. Also, it should be taken into account that a noise level measured one metre away from the noise source is not always accurate in relation to the way you are going to place the said noise source (for example many people have their PC on their desk next to their display, making all noise sources in the reach of less than 30cm to the human ear and so overall noise can sound twice as loud than having the said PC placed one metre away from the human ear).

So to say it straight, don't believe in all numbers you see. Always take numbers, which are the result of specific measurements, with a grain of salt. The same applies to SPCR's results. No matter how much you invest in special equipment for the measurement of noise level, there's always little inaccuracy. Just like everything else in physics, there are actually no absolute answers, but everything is the result of approximations that are very close to the real figures. Having that said, SPCR's results are done (or I believe that they are) with utmost precision and inaccuracy is kept at a very low level.

Bottom line, the best what you could do is to buy a 5400RPM and 7200RPM drive and measure the noise level subjectively by your own ears. I am sure though that the said procedure is not applicable so the best what you can do now is to select a drive you see that suits you best and hope it is as quiet as you expect it to be. Some reviews list 7200RPM drives as being quieter than certain 5400RPM drives, so the mechanics inside the HDDs play a very big role in this. It doesn't come always down to the rotational speed.

Perfect rotating hard drives should be noiseless (aside from some tiny wind noise.) At the same level of imperfection the slower drive will generate less noise at low frequencies. However, slower drivers can also operate with more imperfection (i.e., 5400 rpm drives can be made with looser tolerances than 7200 rpm drives than 10,000 rpm drives than 15,000 rpm drives.)

Small samples (i.e., a single reviewed drive) are almost meaningless in determining what class or even model of drive is quieter.

I think the greenpower 640GB Western Digital is much quieter than the blue version.

discs dont just spin.
what about the heads?
on faster drives they must move faster too, thus increasing total noise.

You mean seeking?
With most drives you can slow down the seek-speed, thus lowering the seeking noise by the acoustic management (fast vs silent).

xan_user: "what about the heads?
on faster drives they must move faster too, thus increasing total noise."

They don't absolutely HAVE to seek faster if platters spin faster. Access time = latency + seek time. Increasing rpm will make latency smaller. Keeping the same seek time, access time will be reduced by the same amount as latency, so without increasing seek speed, performance is improved.

It depends on manufacturer's decision whether seek speeds are "improved" for 7200rpm model. They may want to cut an extra 1ms of access time (since 5400 vs 7200 is only about a 1ms difference)

7200rpm => average latency 4.17 ms.
5400rpm => average latency 5.56 ms.

OK. 1.3 milleseconds so it's a bit more than my quick guess/approximation.

Increasing rpm has more effect on performance than aggressive seek speeds if seeks are localized (for example: on a small partition at beginning of the disk). 1.3 ms less latency is 1.3 ms off each non-sequential access, but 1.3 ms off so-called "average seek time"(*) is equal in significance only then used in high-duty server environment, i.e where seeks are completely random. Or if seeks are de-localized to be longer than random seeks.

(*) Real-world average seek time is not same as "average seek time" because seeks are not random in reality.

So-called "random seek time" is an equally false name since random seek can take a variety of times, and they're all random seeks. If random seek time was just one number, it wouldn't be random any longer. What people call "random seek time" or "average seek time" is always "average random seek time".

Reviewers on Newegg are reporting more "vibration" on 7200 rpm drives compared to 5400 rpm, which is not necessarily measured in dB. The aluminum body Mac laptop seems more vulnerable according to users upgrading from 5400. They report that they can "feel the vibrations through the keyboard". There are even some 10,000 and 15,000 rpm by Fujitsu.

Hi guys.

Sorry for what may seem like a bump far of the edge.
However, I'm in need of some clarification here.

I'm going to buy a new HDD. It's going to be the one and only disk in my barebone system. The computer may be used a bit for video editing and suchlike, but mostly acts like a HTPC. Silence is therefor worth a lot. So, I'm thinking about either "Samsung SpinPoint F3 HD103SJ 32MB 1TB" (7200 RPM) or the "Samsung EcoGreen F2 HD103SI 32MB 1TB" (5400 RPM). Both of these drives cost just about what would be equal to 60 USD brand new here in my country, which is really not too much money to put on a HDD with the standard terabyte of storage space. However, it's the fuzzy question about the noise levels? Both drives have official specs on having a noise level on 25~28dB. But what's it like in reality? Will I hear any noticeable difference? I'm really curious, and would love your input guys.

Cheers!

Start giving a look to some SPCR article, or relevant SPCR review.

However, if you're in the wild for a quiet DAW, I think a unique, simple 5400rpm drive could not suffice: if in case, you may do better opening a brand new thread on the System Advice forum.

The rotational speed doesn't matter much so 5400 rpm would not necessarily a bad choice. It does matter but if a fairly new 5400rpm isn't enough, chances are no single consumer drive would be enough. That said, 7200 rpm have an extra advantage right now: they're available in high densities. These drives are 500G/platter and therefore not the fastest (you could get faster 5400 rpm drives)but they're not slow either.
To make a good choice, you'd need to know exactly what performance you need. That's an application-specific issue and the system advice forum is better at spreading prejudice than determining requirements unfortunately.

Do you mean your thoughts are quite devoid of any prejudice, HFat? Don't take offense, it's almost a joke.

I think any opinion is biased (by previous experiences, or lacks of), but despite of this the System Advice forum may - more probably that not - collect some more opinions (and discussion about them).

If I may take the liberty to put you some requests, I would like to see a thread by you about what do you think (according to your experience) matters or not in usual disk drive benchmarks (I mean where and how they may matter). If you mind, of course.