HDD Diet: Power Consumption and Heat Dissipation

The problem of power consumption and heat dissipation in modern computer components does not need any special substantiations or introductions. It exists and should be somehow dealt with. It's especially critical with the present-day processors and video cards. But the object of this article is another computer element, critical to overheating — hard disk drives (HDD). Manufacturers measure off quite a modest range of operating temperatures — from +5 to +55°C as a rule (occasionally from 0 to +60°C), which is obviously less than in case of processors, video cards, or chipsets. Moreover, reliability and durability of these drives depends much on their operating temperatures. According to our research, increasing HDD temperature by 5°C has the same effect on reliability as switching from 10% to 100% HDD workload! Each one-degree drop of HDD temperature is equivalent to a 10% increase of HDD service life.

It goes without saying that servers and professional data storage systems pay special attention to cooling hard drives — drives are installed into special metal cages and cooled by fans. In such cages the HDD temperature stays within 30-40°C even under heavy load (sometimes it's even close to the environment temperature), which drives away all overheating concerns.

However, much less attention is paid to the problem of HDD cooling in more consumer-like cases, including personal computers (from hardware integrators or self-assembled), workstations, and even entry-level servers, to say nothing of growing increasingly popular "computerized" consumer electronics with hard drives inside (play stations, personal digital video recorders, etc). That's partially due to lower requirements to data storage reliability, partially due to economic reasons, and also because any additional fan makes a device noisier, which is very undesirable. The following two components grow especially important under these conditions:

1. Construction of HDD mounting in a case (relative to other active cooling systems, main airflows inside the case, and passive surfaces that channel the heat away relatively well — metal chassis); but still, our article does not deal with that issue, to be more exact it deals with a slightly different thing.
2. Heat dissipation of drives in various operating modes. That's what our article is about.

I hope there is no need to explain why the heat dissipation of a drive matches its power consumption from a PSU almost perfectly: if we dismiss minute mechanical work, performed by some ill-balanced storage devices by vibrating themselves and the neighbourhood (where they are installed), as well as the power of acoustic and electromagnetic (radio-frequency range) vibrations generated by the operating disk, there are no other ways the drive can transmit energy outside, except for the thermal form. And the only power source of a drive is electricity (we shall reasonably ignore heating from external sources so far ;)). That is we face a classic "electric oven" of a hard drive (the same also applies to a processor — CPU or GPU), it will interest us in this article only in this respect. :)

HDD Temperature Reading Mumbo Jumbo

Some users are too naive to think that all they need to understand everything about heat dissipation of a drive is to measure temperature of the drive during its operations or tests. They think that if they compare several hard disks by this temperature measured under domestic conditions, they could draw profound conclusions that one disk is cooler than the other, that is better and dissipate less heat. Several authors of HDD reviews even base their statistics on it, mistaking its validity and relation to the realities of life. Their readers buy a reviewed hard drive and expect it to keep below 42°C or, say, 47°C — that's because "computer gurus" reviewed it…

Why is it a delusion? Because taking correct readings, that is trying to judge about hard drive's heat consumption by its temperature, all the more to determine the real operating temperature of a given hard drive compared to other drives, requires at least knowing the ropes or helluva onions. :)

That is to ensure accuracy and validity of temperature readings with the measurement error within at least 1-2°C, you must put hard drives into a heat chamber, provide similar heat dissipation conditions (chassis mounting, air circulation), and read the temperature by an external sensor (that is not by the built-in one) at least at several surface points (temperatures inside hard drives can be interesting to only manufacturers, so we shall not analyze them). You must agree that organizing such measurements on a system basis even under conditions of a regular computer testlab is problematic at best — it requires special expensive equipment, rare labs can afford. Otherwise, the measurement error of all makeshift readings under improvised conditions or "system units" will be minimum 10°C, which reminds the notorious "average temperature in a hospital". Furthermore, under these conditions you shouldn't try to compare temperatures of various drives, differing by 2-5°C. It's utterly useless and even harmful, because it misleads credulous readers!

Moreover, if you possess a good heat chamber and other accessories to take correct temperature readings, the results obtained will also be useless to some extent for those who want to know what real temperature their hard drives will reach! That's because real systems channel heat absolutely differently and it cannot be calculated in detail. Conclusion: you'll have to put a given system unit into a large heat chamber (with specified airflow conditions) and take readings. If you risk taking these readings outside a heat chamber in a regular room, a large measurement error due to the drift of room temperature and local airflows will bring to nought the idea of such experiments. However, even if you manage to take these readings, you will not be able to tell for sure that the operating temperature of this drive will be the same in a different chassis, because systems may have quite significant differences in HDD cooling conditions.

Another question — what can be used to measure HDD temperature (if you still want to measure it ;)). It goes without saying that readings of the built-in sensor are absolutely unreliable! Yep, this thermal sensor may roughly guide you in everyday "consumer" practice (e.g., in order to be sure that your drive is not overheated above the safe limit), but these readings cannot be used to compare different storage devices! The fact is that different models have built-in thermal sensors in different locations so that they measure temperatures of totally different parts, which may have different operating temperatures — even in the same drive in different operating modes! Unfortunately, there is no common industry standard on this issue so far. So, if you are still keen on being informed on the real temperature of the hard drive case (specifications usually limit this very characteristic) and, all the more, comparing various drives by their case operating temperatures, you should use an external thermometer of the proper accuracy class.

Power consumption is the "correct" unit of heat dissipation

But enough of measuring temperatures — we are absolutely not going to do it in this review. :) That's because we shall take power consumption for their heat dissipation measure (see above). Moreover, power consumption turns out a much more flexible characteristic in this respect, because it allows to quickly obtain precise data on heat dissipation of a drive operating in various modes (from idle to seek, read, and write), which would have been problematic by reading temperatures. Moreover, you cannot use temperatures to measure for example start-up power consumption. Besides, it's much easier to measure power consumption than to read temperatures with a given degree of accuracy.

Thus, the most correct measure of hard drive heating is the electric power it consumes. But power consumption of hard drives is also important for us because power saving in modern computers is becoming an issue of primary concern. Power consumption of processors and video cards is growing, a couple of dozens of Watts in a hard drive against these nearly-hundred-Watt ovens does not seem so critical. But it depends: in case of a low end PSU (250-300 W), an additional hard drive (or even the simplest RAID) may result in the necessity to upgrade a power supply unit to a more powerful one. Besides, no one abolished the problem of high start-up power consumption – for example, the plain Barracuda 7200.8 may draw up to 2.5A from the +12 V line at start-up. Add 3 W drawn from +5 V to get the peak start-up power consumption of 33 W! What if there are two or three such drives in a system? In this case you should play safe and take a PSU at least by 100-150 W more powerful than processor+video+motherboard require. Food for thought.

So, the object of today's review is to compare power consumption and heat dissipation of modern 3.5-inch hard drives in various operating modes. We shall mostly review desktop models with Serial ATA and UltraATA interfaces, as the most interesting to the majority of our readers. But we'll also include some recent SCSI models as a reference.

Hard Drives' Specifications

As a reference, Table 1 contains power consumption data for the main HDD series, provided in their specifications. We shall start from the beginning. :)

Irregardless of specifications, you should be well aware that they are not a panacea and cannot provide complete facts of life: sometimes manufacturers specify only upper limits, sometimes – typical values, sometimes these figures have nothing to do with reality, if you compare them with the readings taken from these drives. Nevertheless, specifications exist and we should face them.

Another funny delusion - users often consult cases of a hard drive and fondly believe that the power consumption data printed there is true for a given sample of a hard drive ("this data is printed there for a reason!" ;)). Having compared these parameters with real figures, you will see that it's often not the case. Moreover, these parameters often mismatch even the specifications on these drives. It's often not so easy to understand the principles, which manufacturers follow to mark technical parameters of drives on their cases.

Contenders and Test Methods

We have tested 35 models of modern 3.5-inch hard drives from all major manufacturers. The drives are listed in the table with test results below. We used the following testbed configuration to measure power consumption of hard disks:

1. CPU: Intel Pentium 4 3.0C
2. Gigabyte GA-8KNXP Ultra-64 motherboard based on Intel E7210 chipset (the i875P with Hance Rapids 6300ESB southbridge and PCI-X bus)
3. RAM: 2x256 MB DDR400 (2.5-3-3-6 timings)
4. Ultra320 SCSI Adaptec AIC-7902B controller on PCI64 bus
5. The main hard drive: Maxtor 6E040L0
6. Power supply: Zalman ZM400A-APF, 400W
7. Chassis: Arbyte YY-W201BK-A

We measured the power consumption of hard drives in various modes: Idle, ATA or SCSI Bus Transfer, Read, Write, Seek, Quiet Seek (additionally, if supported), as well as Start. A package of these parameters renders the situation with HDD heating (a product of current and voltage gives the heat rate dissipated by a drive) as well as with its economy in the most complete way. Operating modes of a hard drive were controlled by the corresponding tests in AIDA 32 Disk Benchmark, read and write modes were measured "in the beginning" of a disk (on the most frequently used outer tracks; power consumption on inner tracks is usually lower). The tests were carried out under MS Windows XP Professional SP2. The hard drives were tested non-partitioned. Before the tests, we warmed the hard disks for 20 minutes using a utility with active random access.

We measured the +5 V and +12 V draw (accurate voltages at the output of the above mentioned unit were +5.08 V and +11.82 V) simultaneously with two digital ammeters of the 1.5 accuracy class with the resistance below 0.15 ohm (including the leads' resistance). The refresh rate of readings was approximately 0.3-0.4 sec. The table provides average values for several seconds (current fluctuations usually didn't exceed 30 mA), except for the Start-Up current (the table contains maximum values).

Test results

Our readings are published in Table 2. The last column contains the data specified on a case of a hard drive.

The table holds a lot of numbers and there seems no point in commenting them all — they are self explanatory. However, we have a comment to the table with results - the Samsung SP2004C hard disk supporting SATA II interface (its transfer rate is doubled to 3 Gbit/s) was also tested connected to Silicon Image SiI3124-2 controller that supports this new interface. The results are quite expectable — its power consumption remained the same from +12 V line and grew by 20-40 mA from +5 V line (compared to its connection to ICH5 SATA 1.5 Gbit/s) in data transfer modes (+40 mA in Read mode, +30 mA in Bus transfer mode, +20 mA in Seek mode). Thus, a faster interface (SATA II) will hardly provide real performance gain to your data storage system so far, but it will contribute to its heating (by 0.1-0.2 W).

But if you connect a SATA 1.0 hard disk with NCQ support to the SiI3124 controller (we carried out this experiment with Maxtor MaXLine III 7B250S0), in order to see whether NCQ support has an effect on power consumption of hard disks, you will see that the current remains the same in all the modes mentioned (we haven't evaluated possible average power savings due to a faster execution of same tasks). The only exception is Idle mode, when the current was much higher than in case of the ICH5 controll