May 8, 2003 — by Mike Chin
This article began originally as a review on the Samsung 160G SP1604N hard drive. It turned instead, into a roundup of 4 hard drives, the others being the 120G & 60G Hitachi 180GXP and the now venerable 40G Seagate Barracuda IV. The idea of a single drive review in the absence of any comparatives in our database made no sense at all. Our focus is on noise and vibration, but effort was made to get a reasonable assessment of performance as well.
Dru’s Seagate Barracuda 7200.7 HDD article in March this year was the first hard drive review at SPCR. Regular readers and anyone who studies just a few of the articles, sections and forums on hard drives quickly learn that the Seagate Barracuda IV remains the quiet hard drive reference, at least to date. Introduced by Seagate nearly 2 years ago, it was the first 7200 rpm drive to be essentially free of the high pitch whine that plagues 7200 rpm drives, and quieter than even the quietest 5400 rpm drives.
Initially, the challenge of expanding acoustics-focused reviews into the storage area seemed too much of a challenge to tackle. So many other interesting products remain to be examined! However, the absence of detailed acoustic analysis of hard drives by hardware reviewers became ever more glaring by their absence. This article grew out of that realization.
Unfortunately, there was no time to get the drive samples to the UBC anechoic chamber for serious acoustic testing. Given time constraints, the best that could have been obtained were only idle readings, which are not that useful by themselves. Like Dru’s effort, this review is also a first shot done with just benchmarks and careful listening.
A major caveat: Please note that ALL hard drives are subject to great changes in noise output depending on the way they are installed in a case, and in the mechanical / acoustical characteristics of the case.
|Samsung SP1604N 160G, 7.2K rpm|
|Sample supplied by Samsung Canada|
|Price To Be Announced|
Samsung Electronics of Korea have developed a reputation in the computer marketplace for high quality monitors and quiet hard drives. Both of their 5400 rpm and 7200 rpm HDD have been noted by many observers and users as exhibiting very low noise.
The model submitted by Samsung Canada is a brand new Ultra/133 compliant 7200rpm high capacity drive. At 80GB/platter, it is on par with the higest density models available today.
The sample came equipped with 2 MB of buffer memory, but 8 MB buffer versions are available as well. The 8 MB buffer versions are denoted by different model numbering:
SP0421N for 40 GB capacity
SP0812N for 80 GB capacity
SP1213N for 120 GB capacity
SP1614N for 160 GB capacity
- 4 Heads, 2 Disks, 7200rpm,160.0GB, 2MB Buffer
- Data Transfer Rate: 741Mbits/sec
- Ultra DMA/133 Compatible
- 8.9 / 10 ms Average Seek (Rd/Wt) Time;
- Average latency 4.17 ms
- Max operational power: 9W
- Support ATA Host Protected Area Feature
- S.M.A.R.T Compliant
- Idle: 2.7 Bels typical
- Seek: 2.9 Bels typical
Two noise specific features deserve some special attention here.
NoiseGuard is described as a set of mechanical noise reduction features implemented after thorough acoustical analysis. Material choices, mechanical design and damping geared to reduce noise and vibration are all part of the mix. It is not clear whether SilentSeek includes a Fluid Dynamic Motor, which is mentioned as one of the technologies Samsung is current working on.
SilentSeek is designed to minimize the acoustic noise generated by the actuator movement, rather than suppressing the emitted acoustic noise. Samsung claims an average 4 dBA of noise reduction through all operational states, during idle as well as write/seek. The very small 0.2 Bel difference claimed between idle and seek is notable.
|Hitachi 180GXP 120G, 8 MB buffer; 60G, 2 MB buffer|
|Supplied by Hitachi Global Storage Technologies|
IBM sold its hard drive division to Hitachi last year. Once a hard drive technology leader, and the performance leader in the industry for many years running, its reliability issues with the 75GXP series drives seemed to bring on too many problems for IBM to handle in recent years. Never mind that the 75GXP was 3 generations ago; many PC component buyers still paint the IBM HDD line with that brush of old, so perhaps it is just as well that the line is rebranded with Hitachi’s name.
The 180GXP flagship Deskstar series is available in 5 capacities: 30, 60, 90, 120 and 180G. As their OEM specifications sheet indicates, the drive is available with 2 MB or 8 MB cache data buffers.
The samples provided were:
1) IC35L120AVV207-1, a 2-platter 4-head drive with an 8-MB buffer. Geared for very high performance.
2) IC35L060AVV207-0, a 1-platter 2-head drive with a 2-MB buffer.
- Rotational speed 7200 rpm
- Latency average 4.17 ms
- Average 8.5 / 8.8 ms seek
- Data Transfer Rate: 699 Mbits/sec
- Sustained data rate 56 to 29 MB/sec
- Typical power 9.4W total max.
- Idle: 2.8 Bels typical / 3.2 Bels max
- Performance seek: 3.4 Bels typical / 3.7 Bels max
- Quiet seek: 2.9 Bels typical / 3.3 Bels max
- The above acoustics data is for the 2-platter models. The data for the 1-platter model is 0.2 Bel lower for idle and 0.1 Bel lower for quiet seek.
A key feature of the 180GXP is the use of a Fluid Dynamic Bearing motor to reduce motor noise. Interestingly, Hitachi differentiates between the 8 MB and 2 MB buffer Deskstar drives by providing a 3 year warranty with the former and only 1 year with the latter.
|Seagate Barracuda IV 40G, single-platter|
|Sample purchased; 3 months old|
The Barracuda IV became an instant hit with noise-conscious PC users when it was first introduced in 2001. With a claimed noise level of <2.5 Bels idle and 2.5 quiet seek, the original Barracuda IV single-platter 20G and 40G models were dramatically quieter than other mainstream drives, even the 5400 rpm models. They retain the reputation of being the quietest even today.
Their performance was competitive when they first appeared, but there’s little question that others have spun smartly past them since. They appear to be in the process of being phased completely out of production, displaced by the newer Barracuda V and faster Barracuda 7200.7, both with Serial ATA options.
Regardless, the single-platter Barracuda IV may be at least marginally quieter than its newer relatives, and remain widely used in the SPCR test labs.They have been reliable, quiet workhorses on a wide variety of platforms. They are our reference low-noise hard drives.
The Barracuda IV’s defenses against noise include:
- Highly proven Fluid Dynamic Bearing motor
- SeaShield protects circuit board against accidental handling damage — it even has a layer of blue-colored foam damping that is visible beneath this metal shield.
- Acoustically damped seek adjustable via software (default setting is quiet seek)
- Ultra ATA/100, 7200 rpm
- Maximum Internal Transfer Rate (Mbits/sec) 555
- Average Seek 8.9 ms
- Average Latency 4.16 ms
- Idle: <2.5 Bels typical
- Performance seek: 3.0 Bels typical
- Quiet seek: 2.5 Bels typical
It was established early on that all of these hard drives idle at a noise level well below 30 dBA @ 1 meter, where my old Heath SLM has no resolving power. Rather than resort to close-field microphone techniques and come up with yet another set of non-standard acoustic measurements that cannot be easily compared, I opted to go with careful extending listening analysis. These tests were repeated many times with rest periods in between over the course of many days to minimize the role of human perception error. It was tedious, but necessary.
Another caution: Please note that the noise analysis results are relevant only for the specific samples tested. While one would expect quality assurance from hard drive makers to be of a high level, I have personally heard audible variances in specific models from a variety of HDD brands. Because all of the hard drives reviewed here are relatively quiet and grouped fairly tightly together, even small differences in noise could impact the ranking results.
The system used as the test platform for these hard drives consisted of:
- Intel P4-1.8A CPU
- Intel D845PEBT2 motherboard
- Thermalright SLK900 heatsink + Panaflo FBA08A12L1A 80mm fan + Zalman fanmate1 — The fan was often turned off during listening.
- Matrox G550 VGA card (AGP)
- 512 MB OCZ DDR RAM – PC2700
- Seagate Barracudua IV 20G (in Smart Drive from Silicon Acoustics)
- Seasonic SS-300FS power supply (modifed w/ Panaflo “M” and grills cut away)
- Two-level plywood platform with foam damping feet.
- Windows XP Pro (SP1 w/ all updates)
The following software was used during testing. All are linked from SPCR’s Software page under Useful Web Links on the main menu:
- DTemp – simple, highly useful utility to monitor hard drive temperature available from the S.M.A.R.T. features in these drives
- SiSoftware Sandra v. 2003.3.9.44 — to run benchmarks and put the drives into seek mode consistently
- HD Tach 2.61 — to run HDD benchmark analysis
The test environment, as usual, was the converted kitchen that is SPCR’s test lab: hard reflective surfaces and a very low noise floor allows all kinds of noises to be heard easily, especially in the evening.
Acoustics and Vibration
Each drive was installed in turn as the second drive in the test platform system. It was physically placed on a massive cabinet counter, and the system turned on to get the drive running. While in operation, the drive was allowed to lie loosely on the countertop. Varying amounts of pressure was also applied with the fingers to the top of the drive, and acoustic effects listened for. Each drive was also picked up and held in the hands, as well as placed on soft foam damping to consider only acoustic effects (apart from vibrational ones). The listening was done with the drive in idle as well as in seek.
The noise was divided into four distinct components:
- Vibration: Identified by Seagate as being the parameter that most dominates HDD-caused noise in most PC systems, it can be manifested in many different kinds of sounds. Typically it causes low level buzzing that is often not noticed until eliminated — and then subjectively, is as if an acoustic fog has been lifted.
- Idle noise: This is the minimum noise of the drive in operation, composed largely of broadband noise (shhhhhhhh) and some low frequency elements associated with the drive spindle rotation. Somehow, this is the least intrusive or annoying part of the overall drive noise.
- High frequency noise: Probably considered the most annoying of all hard drive noises, most drives of 7200 rpm spindle speed or faster are characterized by this whine, typically at 8~16 kHz. Even trace amounts can be very annoying because of its psychoacoustic impact and its high directionality: A simple movement of the head can cause the highly directional sound to come in and out of audibility, causing a dynamic stuttering effect that drives some people nuts.
- Seek noise: It usually sounds like grinding or chattering. The quality of the noise varies considerably, but not everyone hears it as a serious annoyance. This is especially true for IT “old timers” who are used to listening for the seek noise as an assurance that the computer is doing work. The main issue here is how much louder than the idle noise it is: The bigger the contrast between idle and seek noise, the more annoying and noticeable it usually is.
The table below summaries the results of this arduous listening analysis. Rather than trying to give a quantitative value to the noise level, which is an incredibly difficult thing to judge, I ranked each drive in the order of noise performance, 1 being the quietest (best) and 4 being the loudest (worst). It is not that hard to judge, “This is louder than that.” But to say “This is 10% louder than that” (with any accuracy) is almost impossible. There were some ties on individual parameters, but the overall SCORE (sum of the individual parameter rankings) are consistent with my subjective impressions for the relative noise levels of the drives. It should be noted that the real differences in noise levels among the 4 drives were small.
|Barracuda IV 40G|| |
|Samsung SP1604N|| |
|Hitachi 180GXP 120G|| |
|Hitachi 180GXP 60G|| |
Seagate Barracuda IV: That the Seagate should come out on top is no surprise. The single-platter model is truly quiet. It is only in the seek category that the Barracuda IV was bettered. Even with the acoustic management of the Barracuda IV set for maximum damping, I’ve always found its seek noise to be easily heard, perhaps in ontrast to the very low idle noise.
Samsung SP1604N: The Samsung turned in a very credible performance as the noise runner-up. While exhibiting slightly higher noise overall, its high frequency whine noise seemed even lower than that of the Barracuda IV, although this could not be clearly confirmed. It was when the Barracuda IV was turned off, and then the Samsung turned on that one sometimes got the sense of a slight absence of high frequency noise. But when listened to it by itself, the whine could only be heard at trace levels from the ‘cuda at distances under 6 inches.
The seek noise of the Samsung was vanishingly low, and their claim of a mere 0.2 Bel rise from idle seems justified. This lack of dynamic noise activity means that while the Barracuda IV has slightly lower overall noise, it may be easiest for the Samsung drive to become psychoacoustically transparent, for its noise to fade into the background for most people.
Htachi 180GXP: Both models were marred by high frequency noise. There was simply no getting around that noise. It was not at a very high level in the 120G model, but plainly audible. The whine was more marked in the 60G model, which was a surprise. Perhaps it was the lower broadband idle noise of the 60G model that made its whine more audible? The vibration level of the 120G model was excellent, clearly the best of all the drives. The comparatively higher vibration exhibited by the single-platter 60G model makes me suspect that perhaps this unit was not quite representative of the norm. Single-platter units usually have less vibration than multi-platter ones.
Still, these Hitachi drives are definitely quieter than other high performance drives that have passed through the SPCR labs in recent weeks. That includes models by Maxtor and Western Digital, which were far noisier. Those units are not in this roundup because there is no point — I have yet to hear any models from either Maxtor or Western Digital that were quiet enough for me to consider using in a quiet PC. (Having said that, I must also add that I certainly have not heard all the available models.)
The issue of the “head reset” or calibration in the Hitachi and the “chirping” noise that it makes is real. It has been discussed in the SPCR forums, as well as at Storage Review. I did hear this noise a few times, but it did not happen often. I doubt I heard it more than 3-4 times in total, which is much less often than reported by those who have discussed it. It is an odd noise, significantly louder than any other noise the drive makes. How annoying it is will depend greatly on the user.
With the exception of the 60G Hitachi, the drives were all benchmarked several times with SiSoftware Sandra (v. 2003.3.9.44) and HD Tach (2.61). The 60G Hitachi was left off for lack of time; besides the 120G sample provides a good picture of performance for the series anyway. The median result were screen captured and are presented here. No detailed commentary is offered; you will find the best analysis of hard drive performance at Storage Review. These benchmarks do tell a story of their own, however.
SiSoftware Sandra Benchmarks
HD Tach (2.61) Benchmarks
Average Hard Drive Temperatures as measured via DTemp
The ambient temperature was 20C in the test lab, and the drives were out in the open. These were the highest temperatures seen with each of the drives after several runs of the above benchmarks.
|Barracuda IV 40G|| |
|Samsung SP1604N|| |
|Hitachi 180GXP 120G|| |
|Hitachi 180GXP 60G|| |
There are no surprises in performance. The Hitachi 180GXP wins decisively in all categories, which is expected with its 8 MB of buffer cache memory. Its’ random access time as measured by HD Tach, which is likely the most noticeable hard drive performance parameter in desktop computing, is right in line with its claimed specs at 12.7 ms. Take away the 4.17 ms or so of latency and we have right around 8.5 ms, as claimed. In actual use, it also feels like the fastest of these drives.
The Samsung SP1604N is a very good performer and in real use, feels significantly faster than the Barracuda IV. The random access time is a bit of a puzzle, as it neither feels that slow or is specified to be that slow. Even after accounting for the latency, this drive measured 2~3.3 ms slower than specified.
The Barracuda IV brings up the rear, but remains a perfectly good performer in the various test platforms for SPCR. It is also the drive being used to create this page.
It is interesting to note that the Barracuda IV consistently ran at least 2 degrees C hotter than the other drives. Perhaps the SeaShield and foam damping over the PCB circuit has this effect.
The Seagate Barracuda IV remains the hard drive to beat for low noise. The acoustic standard established nearly two years ago remains intact today; that is notable in this fast moving industry. The performance of the Barracuda IV is no longer enough to keep up against the competition, and it is no surprise that the IV is being replaced by the V. Still, at their current prices, the IVs must be some of the best bargains quiet computing enthusiasts have ever seen.
The Hitachi 180GXP continues the tradition inherited from IBM of very speedy performance with low noise but a bit of whine and that questionable chirp. For a performance machines, it is a great choice, and there will be some application where its performance bias will make it the ideal choice. The Hitachi has the reputation of being only a hair slower than the much louder Western Digitals of similar specification. A different sample of the single-platter version may well be capable of better noise performance than my sample exhibited as well.
The Samsung SP1604N, whose arrival instigated this roundup review, turns out to be a most pleasant surprise, coming within a hair of the Barracuda IV for noise and falling in between the Hitachi and Seagate for performance. The slow seek times seemed anomalous but were consistent. In actual use, I often found the Samsung to be slightly less obtrusive than the Barracuda IV because its seek noise was so low.
In the decouple-mounting I normally use to install hard drives (DIY suspensions or NoVibes), the seeks became completely inaudible, making the noise of the Samsung utterly stable and consistent. There was no appreciable overall noise difference between the Samsung and Barracuda IV in a noise-optimized case/system, but the seeks of the latter could sometimes be heard at low levels. This is despite the 2-platters of the Samsung.
The lack of dynamic changes in noise is a key ingredient to good low-annoyance noise performance, and in this regard the Samsung is the best I have yet to hear. Its performance comes as a bonus. We can anticipate that the single-platter versions of this drive would be even quieter, and the 8 MB buffer versions would be faster; I look forward to examining both variants in future.
All of the drives reviewed here are suitable for low noise PC systems. The Barracuda IV and Samsung models are nearly ideal, and the Hitachi models more challenging to silence — but not impossible. All have their place in the silent PC enthusiast’s repertoire of components.
* * * * *
Discuss this review in our Forums.