• Home
  • blog
  • Shuttle SD11G5: Pentium-M SFF PC

Shuttle SD11G5: Pentium-M SFF PC

It’s late to the table, but Shuttle’s long awaited SD11G5 is a landmark SFF. Just the basics are enough to make a winning combination: A Pentium M SFF platform around a proven single fan heatpipe cooling system with a PCIe video card slot and an external fanless power box. Add Shuttle’s usual attention to details, styling and packaging, extra features like the ambitious built-in Sound Blaster 7.1 sound card and dual monitor onboard display capability, and it looks just about perfect.

October 17, 2005 by Mike
Chin

Product
Shuttle
XPC SD11G5
Manufacturer
Shuttle
MSRP
US$450

The Shuttle XPC SD11G5 was first shown early this summer at Computex in Taipei. It
is of particular interest to the SPCR community because it is a socket 479 system designed around the Pentium M processor. It is Shuttle’s first Pentium-M system, fitted into the G5 chassis, which is among the smaller ones in the current Shuttle Small Form Factor (SFF) lineup.

The combination of a fanless external brick power supply and the cool Pentium M platform means that the SD11G5 has instant panache with those who seek quiet PCs. It is reminiscent of the Zen ST62K, which remains Shuttle’s smallest SFF model and the quietest thus far; that model is also fitted with a fanless external brick power supply but has no VGA-specific expansion slot.

Through much of this year, Shuttle appears to have been going after the LAN party gaming market with larger SFF models that allow big, hot, high-end VGA cards to be used. One example is the socket-775 SB81P reviewed in March, which allows the use of a double-width PCIe VGA card. The SD11G5 is the first model since the A64-939 based SN95G5 (also reviewed in March) to appeal to the quiet PC market.

Shuttle is not the first to release a socket 479 SFF system. AOpen beat them to it by half a year with the EY855-II, which we reviewed last spring. This was based on the older 855 chipset. AOpen has been active with socket 479 SFF systems in the interim; the somewhat controversial and tiny Mini PC (Mac mini look-alike) was unveiled at Computex, and recently released the almost as small MZ915-M, based on the newer 915M chipset, the same as in the SD11G5.

The closest rival for the Shuttle SD11G5 is the AOpen EY855-II in terms of size and features. The big visible difference is that the AOpen has an integrated fan-cooled PSU, which limits just how quiet that system can run. Acoustically, the SD11G5 has an obvious advantage with a fanless PSU, and it also has some advantages in its feature set as well. The AOpen MZ915-M may also be a good competitor, but we have not seen samples of either of these new models yet.

BASICS


The usual oh-so-slick box.


What’s inside: External Power Supply, User Guide, XPC Driver CD, Extras CD, IDE Cable, PEG power cable, Composite to S-Video cable, Power Cord, SATA cable, Heatsink compound, Front Feet /Twin Adhesive/Screws


The main event: Pearly white
SD11G5.
NOTE: Shuttle says the US version will be all-silver.

SPECIFICATIONS & FEATURES

SD11G5
Comment
Size WxDxH (mm)
310(W) x 200(D) x 185(H)
11.5 liters

G5 chassis is the same used in SN95G5. Only slightly larger than miniscule 9 liter Zen.
 

Chipset

Intel 915GM + ICH6M chipset

Memory
(2) 400 / 533 DDR2 DIMM slots
1GB per DIMM max 2GB
CPU
Socket 479

Intel Pentium M 533/400 MHz FSB
Intel Pentium M LV 400MHz FSB
Celeron M


Graphics
Intel GMA 900 – DVI, VGA

Integrated Graphics w/ support for Dual Independent Display!
Serial ATA
ICH6-M integrated SATA
Dual channel UDMA 150MB/s SATA
 

Audio

Creative Sound Blaster Live! 24-bit
7.1 channel EAX Advanced HD audio
Digital SPDIF I/O

Impressive!

LAN
Broadcom 5789
100Base-T specifications compliant
10/100/1000 Mb/s Ethernet
Wake-On-LAN function


Gigabit LAN
Drive bays

1 x 5.25″
1 x 3.5″ external
1 x 3.5″ internal



Onboard Headers & Connectors

(1) x ATA100 bus master IDE connector
(2) x Serial ATA connector
(2) x 4-pin fan connectors
(2) sets of 1×5 pin USB 2.0 header
(1) set of 2×5 pin USB 2.0 header
Front Panel FPC header
GPIO header
Parallel port header
AUX in header
Power On header
Mini PCI slot


Front Panel

(2) USB 2.0 ports
(1) Mini IEEE1394 port
(1) Mic in
(1) Speaker out
(1) Power on button
(1) Reset button
(1) HDD LED
(1) Power LED

Back Panel

(1) x 16 PCI E slot
(1) x 1 PCI E slot
(1) Set PS/2 Keyboard/Mouse ports
(2) USB 2.0 ports
(1) RJ45 Gigabit LAN port
(1) Optical SPDIF out
(1) Optical SPDIF in
(1) Set 7.1 audio ports
(1) Line in
(1) IEEE 1394
(1) Clear CMOS
(1) VGA connector
(1) DVI connector
(1) Composite video out
(1) Printer port perforation
(1) WLAN perforation


Power Supply

220W external adapter
Input: 100/240V AC
EMI Certified: FCC, CE, BSMI
Safety Certified: UL, TUV, CB, BSMI


No fan, no noise!
Two very notable features:

  • Onboard VGA supports dual-monitor setup with DVI and VGA outputs. This is unusual!
  • A built-in 24-bit 7.1 channel Sound Blaster sound card is pretty impressive.

EXTERNAL OVERVIEW

The SD11G5 is virtually identical externally to the SN95G5, except that the latter is black and has a built-in power supply. The drive bays and connectivity panel are stealthed; only the power and reset buttons are exposed. The 3.5″
bay and connectivity panel have flip-down covers, while the
5.25″ optical drive bay has a spring loaded cover that automatically
opens when the CD/DVD tray is ejected. Although the button mechanism is adjustable, there can be some compatibility issues with some optical drives.

The chassis is made of aluminum. The cover is painted glossy white, while the front bezel is mostly white plastic (not a perfect color match, but close) with chrome accents. The only vaguely “loud” feature is a pair of stamped “Shuttle”
logos on the sides. The intake vents are on both sides and run most of the length of the case.


Center external floppy bay and connectivity panel covers open.

The hidden front panel connectors include 1/8″ headphone/microphone jacks,
two USB 2.0, and a mini IEEE1394 port. The space between the USB ports allows multiple USB
devices, such as USB flash drives, to be used at the same time. All other ports are located on the rear panel.


The rear panel: The Sound Blaster sound card ports fill the spot usually used by the PSU exhaust in the G5 chassis.

The rear panel includes a VGA port, a DVI port, an IEEE1394 port, a PS/2
keyboard and mouse socket, TV/S-video port, two USB 2.0 ports, a RJ-45 gigabit LAN socket, 7.1-channel
audio in/out jacks, and SPDIF in/out ports. Also the CMOS reset button is accessible via a pinhole,
which saves the user from having to open the case up in the event of an overclocking-gone-wrong
scenario. On the right side are the slot covers for one PCIe-X16 and one PCIe-X1 card. Note the 6-pin power inlet in the orange block. The exhaust grill is decently open, the same as in other G5 chassis.


The underside.

The four rectangular of tapped holes on the underside near the front are for HS mounting; the screws go through the motherboard into the chassis base. The vent holes are too small to be useful, really. There is a thin slot at the bottom of the front bezel, but it’s not clear that there’s really an intake airflow path to the interior.

Soft rubber feet are used to minimize the transfer of vibration.
Near the front two feet are two threaded holes which will accept the included
taller metal feet. The metal feet may have two functions: To
enable more airflow under the case, and to make the front panel more easily accessible when it is mounted lower than the user’s desktop. Unfortunately, the metal
feet are not soft, which means it will enable vibration transfer from the PC to the desk (or whatever it’s placed upon). It also makes the thing look odd, a bit like a puppy dog sitting up to beg.

INTERIOR

The interior is well thought out,
with all the cables routed neatly in order to maintain good airflow and to simplify
installation.


Well organized and uncluttered, especially because there is no power supply inside.

Like 775 socket SFF systems from Shuttle, the CPU is mounted at the very front, BTX style. Because the cooling airflow design of the G5 chassis is intake through the side vents, then exhaust out the back, a heatsink with very long heatpipes is used, as the photo below shows. The specifications don’t mention BTX, so one has to wonder why the CPU socket wasn’t positioned closer to the back, as is the case for Shuttle’s socket 478 and AMD K8 systems. It would have made for a simpler cooling solution.


Side view
shows the long heatpipes, cable management, 6-pin PCIe plug, and the passive Northbridge heatsink in the center.
Note the clip for IDE cable routing on the underside of the drive tray.



Drives are best mounted in the
one-piece drive cage prior to the installation of the drive cage into the chassis.


Drive cage removed: An open, easy-access design.

I.C.E. HEATSINK / FAN MODULE

A variant of Shuttle’s standard I.C.E. (Integrated
Cooling Engine) is used. One end of two long heatpipes are clamped between a copper base and a small aluminum heatsink with four spring loaded bolts. At the opposite end is an array of fins. A 92mm fan attached
to a shroud exhausts the CPU heat through the fins and also serves as a general
exhaust case fan. It is a well-engineered design.


The I.C.E. (Integrated Cooling Engine) Module.

The copper base of the I.C.E. module is just about as flat and smooth as one could want. The wire fan grill on the inside restricts airflow a bit, and the hex-pattern grill on the outside restricts airflow a bit more. Truly noise-conscious users will want to remove these grills.


Intel 760 (Pentium M, 2.13 GHz) locked in 479 socket. Note 4 mounting holes for heatsink.



Heatsink bolted in. Simply turn the screws until you can’t turn any more. The springs are preloaded to the correct tension. Mounting the heatsink does require some care due to the cantilever effect of the cooling fins at the far end, and the exposed die of the Pentium M processor (which is more easily damaged than processors with integrated heat spreaders). Some support on the side of the fins is needed while the mounting screws are tightened.


Fins on the other end.


The 92mm fan has a connector that goes into a 4-pin PWM header on the motherboard.
It is the same fan used in the SN95G5: 0.56A at 12V. It can move a lot of air — and make a lot of noise.

EXTERNAL AC/DC ADAPTER

The AC/DC adapter accepts any AC from 100V to 240V and has a 6-pin connector like the one that comes with the Shuttle Zen. It’s a bit smaller except for length, and is rated for slightly higher power: 18A at 12V (216W) instead of 16A (192W). This is more than enough to power any Pentium M plus any combinations of components (VGA card, memory and drives) than can be plugged into the SD11G5. To keep things simple for international distribution, they’ve made the power a universal input device that runs on any AC from 100V to 240V.


The AC/DC adapter is a bit smaller than the one that came with the Zen.


It’s made by Delta, the world biggest power supply maker.

Because the power adapter only provides 12V, the motherboard itself must further convert the 12V into 5V and 3.3V (in addition to the lower voltages required for the CPU, which are normally performed on the motherboard). The back corner close to the input connector for power appears to be where this drop down conversion occurs.

Power conversion components on left? PCIe X16 slot in foreground; the long heatpipes above in the center.


The other back corner, taken by the PSU in other G5 systems, is occupied by the Sound Blaster sound card.
NOTE: The sound card was not tested beyond plugging a pair of high quality headphones in to play some video clips and music.
It worked and sounded fine, as did the Creative SB software drivers and tools (all 170 MB’s worth).

SYSTEM SETUP

Assembling a system in the SD11G5 is a cinch. A phillips head screw driver is just about the only tool you need, and the step by step manual is very well illustrated. There are some steps that may need a little care or finess, but if you are capable of plugging in connectors, lining up things mechanically before screwing them in place, and double-checking your work before turning on power, you can do this.

The following components were installed in the Shuttle SD11G5:

  • Intel 770 (Pentium M, 2.13 GHz + 2 MB cache) processor
  • 2 x 512 mb Corsair DDR2
  • Seagate 5400.2 SATA 120GB 2.5″ notebook HDD – suspended with clothing elastic in 3.5″ HDD bays. No floppy drive was used.
  • Microsoft Windows XP Pro SP2 fully updated
  • On-board video as well as AOpen Aeolus 6800GT-DV256 PCIe video card

Even with the extra work of stringing up an elastic suspension for the hard drive, the assembly took less than an hour.

The SD11G5 is designed to accept a standard 3.5″ HDD, but in our experience, even the quietest 3.5″ desktop HDD makes too much noise — particularly vibration-induced noise — to be recommended for a quiet SFF system. A SFF system is most often placed on top of the desk, within reach of the user for easy access to the optical drive and front panel ports. In such proximity (typically close to ear level 2~4′ away), it is almost impossible to obtain a level of noise SPCR considers quiet (<30 dBA/1m) without resorting to the very quietest HDD in an elastic suspension. Our last few SFF systems have all been reviewed with quiet notebook drives for this reason.


Seagate SATA notebook HDD suspended in the two 3.5″ bays.

BIOS

There are many user-configurable options in the Phoenix BIOS, as befits an enthusiast PC. The Advanced Chipsets menu shows memory timing configurations and various other settings, including those for VGA.


Advanced Chipset menu

The Frequency Control menu only allows overclocking, up to about 24% above the standard clock speed for a 533 MHz Pentium M. This menu is not exactly a tweaker’s delight.


CPU Frequency Control menu.

PC Health

PC Health is the menu that shows all the monitor temperatures, voltages, and fan speeds. The only user adjustable setting here is for the CPU fan. There are five settings in all.

  • Smart Fan (default – 820 rpm)
  • Ultra
    Low
    (820 rpm)
  • Low (1,000 rpm)
  • Mid (1,800 rpm)
  • Full (3,000 rpm)

It’s interesting to note that even though this is the same fan used in the SN95G5, the maximum speed is 600 RPM slower, which implies that the fan is deliberately limited by the BIOS fan controller. This decision reflects on the cool operation of socket 479 processors for which the SD11G5 is designed.


Fan settings in PC Health Status menu.

Smart Fan and U(ltra) Low are the only settings of real interest to us. Either of these settings resulted in a fan speed of 800 RPM (with the stock fan). In a hotter environment than the SPCR lab’s current 20~21°C, it is possible that the Smart Fan setting would result in a higher speed than Ultra Low. The latter is a fixed speed while Smart Fan is thermally controlled, likely tied to the CPU thermal diode output.

Unlike the fan controls in the SN95G5, there is no option for the user to choose the temperature at which the HS fan will start to ramp up in speed. Nor is there any option to choose which fan header is controlled by the fan controller. This features were probably left out deliberately because there is only one fan in the SD11G5 and all the socket 479 processors are so cool-running.

Power Management

The Power Management menu is where Enhanced Intel SpeedStep Technology (EIST) support can be enabled for dynamic, automatic control of CPU clock speed and voltage in order to minimize power consumption — and heat.The enabled setting is Max Saving. (Note: EIST has the same basic functionality as AMD’s Cool ‘n’ Quiet.)


Power Management menu.

The operating system must also support EIST. In Windows XP, this control is under the Power Schemes tab in the Power Options, accessible in the Control Panel. “Max Battery” is the setting to engage EIST.


Windows XP Power Schemes must be set to Max Battery for EIST to work.

EIST brought the processor speed down to 800 MHz and the core voltage down to 0.992V in light load. The system works seamlessly, switching instantly between mininum and maximum speed/voltage as required by the load on the CPU. The details of the effect of EIST on power consumption, temperatures and noise will be dicussed in the testing section.


CPU-Z v1.3 shows CPU clock speed of 801.8 MHz and Vcore of 0.992V in low load on left.
Compare that to high load on right: 2.1382 GHz and 1.328V.

TESTING

No performance benchmarks were run on this system. The results are highly unlikely to vary more than 2~3% from any other Intel 915GM based system using the same CPU, hard drive, memory and video card. This is insignificant in any practical application. The system was as fast and responsive as just about any we’ve seen in the lab.

Tools

  • Seasonic Power
    Angel
    power meter, used to measure AC power consumption during various system activity states.
  • CPUBurn processor stress testing software to load system to full.
  • 3DMark05 was used at default test settings to engage the VGA card and increase power consumption.
  • SpeedFan system monitoring sofftware to monitor temperatures and fan speed. (It would have been nice to get a compatible version of Shuttle’s XPC Tools utility, but the SD11G5 is still not supported. I’m told it’s coming.)
  • B&K 2203 sound level meter (capable of measuring well below 20 dBA)
  • SPCR’s Digital Audio Recording System

The ambient temperature was 20°C during testing. The ambient noise was 18 dBA.

Configuration 1: With on-board VGA

  • Intel 770 (Pentium M Dothan core, 2.13 GHz + 2 MB cache) processor. 27W is the rated TDP; maximum “junction temperature” is 100°C — whatever that means. It’s worth noting that in his review of the AOpen i915GMm-HFS socket 479 motherboard, Ralf Hutter recorded a temperature of 85°C for a Pentium M 755 – 2.0GHz, and that processor is still doing fine.
  • 2 x 512 mb Corsair DDR2 memory
  • Seagate 5400.2 SATA 120GB 2.5″ notebook HDD – suspended with clothing elastic in 3.5″ HDD bays. No floppy drive was used.
  • Microsoft Windows XP Pro SP2 fully updated
  • On-board Intel GMA900 video

SD11G5 w/ on-board VGA
System Activity
AC Power

CPU Temp

Fan RPM
SPL*
Board Temp?†

EIST Idle

32W

24°C 

820

21~22

40°C

Idle

41W

30°C

820 

21-22

41°C

Full Load

56W

45°C

960

23

46°C

3DMark05

58W

46°C 

980

23

47°C

* Sound Pressure Level measured in dBA at one meter from the front bezel with the fan control set to Smart Fan or Ultra Low.
† It’s not clear what several temp sensors picked up by SpeedFan actually monitor. These readings are from the sensor that always gave the highest temps — just for the record.

The testing and data collection was anticlimactic because there was little change between low and high loads. The system barely changed in any perceptible way:

  • The noise went up just enough to let you know that the fan had sped up a touch.
  • The air blowing out the back of the system felt slightly warmer after max load and 3DMark05 for >30 minutes than it did at idle.
  • The power draw was the biggest change, but the maximum peak with 3DMark05 was just 56W.
  • The hard drive temperature stayed at a constant 31°C throughout the tests.
  • Idle temperature with EIST enabled was just 4°C above ambient room temperature.

There were several other temperature sensors detected by SpeedFan, but it’s not clear what they were really reading. The column marked “Chip Temp” provides readings from the sensor that always gave the highest temperatures.

As expected, video performance in 3DMark05 was unacceptably slow, but this is what we can expect from any Intel GMA 900 onboard video. In 2D applications, video performance was perfectly good.

The Dual Display option was also tried briefly, using two 19″ LCD monitors. Having worked with Matrox Dual Head video cards since this technology first came to market, I am used to a high level of dual-display performance. Although extensive use testing was not done, I was surprised by the dual display functionality of the onboard video. It was decent and worth exploring further


The Intel display options menu.

Noise Quality: The primary noise from our sample system (as configured for System One) was from the I.C.E. cooling fan. There is virtually no wind turbulence noise to speak of at either the lowest speed or the maximum speed reached during testing at the Smart Fan setting. Most of the noise is a buzzy hum that gets a bit louder when the fan speeds up. It’s not until ~1,000 RPM or higher that any of the wind turbulence noise becomes audible. Much beyond that speed, the hum begins to get masked by the wind turbulence whooshing noise. Despite the low measured level at slow speed, this fan hum is annoying and intrusive, at least here in the quiet lab.

Shuttle provides an acoustic report for many of their SFF models. The acoustic report for the SD11G5 was not available on the Shuttle web site at time of writing, but a copy was provided along with the review sample. It is reproduced in its entirety below.

Shuttle’s SPL measurements were made with the microphone placed ~0.5 meter away from the front panel of the PC in accordance with the Seated Operator Position as defined in ISO 7779. The distance between the microphone and the PC is ~0.5 meters. When our measuring microphone was positioned half a meter away from the SD11G5 system, we obtained SPL results that were 3 dBA higher compare to the one meter readings. They jibe almost perfectly with Shuttle’s acoustic report. We did not take any SPL measurements of the fan at higher speeds; there does not appear any reason for anyone to use the higher speeds, as cooling is excellent with the fan on the lowest speeds.

Odd PSU Noise on Standby

It’s a noise that became evident when the power to the PC was turned off. It would be normal to simply keep the AC/DC adapter plugged into the AC outlet. The AC/DC adapter emits an odd squealing noise at a very low level, continuously, when the SD11G5 is turned off. Its stops when the computer is turned or when the adapter is unplugged from the AC outlet.

The noise is quiet enough that it might not be noticed if the adapter was under the desk. For our testing, the device was atop the desk, so it was clearly audible, although it semed to be at around the room ambient level of 18 dBA when measured from a meter away. This happened with any system configuration. We made a recording; you can listen to it along with the other recordings further on. It’s impossible to say whether this is a sample anomaly, but a review of the SD11G5 at the French web site www.matbe.com also noted the same phenomenon.

TESTING continued

Configuration 2: With nVidia 6800GT PCIe video card

  • AOpen Aeolus 6800GT-DV256 PCIe video card was used in place of the onboard video. All other components remained the same.The switch to PCIe-X16 required a change in the J1 jumpers of the motherboard, right next to the PCIe video card slot. The switch from 2-3 to 1-2 is best done before installing the video card. It consists of repositioning two strips of 6 jumpers each.


J1 jumpers (one pair) set up for onboard DVI output on left, for PCIe X16 on right.

Getting the big 6800GT card in place called for a bit of finess and care, but it’s the same in any other SFF system where space is always a bit tight. The 6-pin 12V connector for the video card is well placed for easy insertion on the video card. The stock fan on the Aeolus was undervolted to 5V using a Zalman Fanmate 2 as shown below.


High performance AOpen Aeolus 6800GT video card installed; stock fan undervolted to 5V w/ Zalman fanmate2.


SD11G5 w/ AOpen Aeolus 6800GT VGA
System Activity
AC Power

CPU Temp

Fan RPM
SPL
Board Temp?†

EIST Idle

74W

26°C 

830

23 dBA

47°C 

Idle

83W

32°C

830

23 dBA

47°C

Full Load

94W

47°C

1120

26 dBA

48°C

3DMark05

133W*

47°C 

1220

26 dBA

54°C 

* Peak value. Sustained maximum was 125~130W.
† It’s not clear what several temp sensors picked up by SpeedFan are actually monitoring. These readings are from the sensor that always gave the highest temps — just for the record.

With the addition of the 6800GT card, idle power jumped 42 watts, or slightly more than double compared to Configuration 1. This difference stayed fairly consistent through all the various system activities, except for 3DMark05. The highest peak for AC power of 133W was seen during 3DMark05, but it was only for a second or two. That was a whopping 75W higher than with the onboard video, much more than double the power. The longest sustained maximum (more than a few seconds) was about 130W.

The >130W AC power draw seen during 3DMark05 suggests that the 6800GT was adding a minimum of ~25W additional heat into the small volume of the case (compared to CPUBurn). This assumes a minimum conversion efficiency in the AC/DC adapter of ~70%; such devices can often approach 90% efficiency. Yet, the CPU temperature was unaffected by the additional heat, partly as a result of the Smart Fan setting, which increased fan speed to compensate for the additional heat in the system. Also, despite the much reduced speed of the VGA cooling heatsink/fan (HSF), there was no evidence of video misbehavior through several loops of 3DMark05 benchmarking. All this suggests that the side vent of the SD11G5 allows enough cooler outside air intake for a hot video card. The 6800GT is not the hottest card around, but it’s close, and newer top-end video cards actually appear to be improving in power/heat efficiency. Overall, the airflow and thermal management design of the SD11G5 appears very good.

The noise level went up mostly because of the fan in the 6800GT, even though it was reduced down to 5V. At high load, the Smart Fan speeding up a bit more also added to the noise, but overall, the increase was a very modest 2~3 dBA@1m, which is excellent given the more than doubling of total heat in the system (from 56W in Config 1 to 130W in Config 2). If anything, the character of the noise was slightly more palatable because the annoying buzzing of the stock fan was masked a bit by more broadband wind turbulence noise.

MAKING IT EVEN QUIETER

The overall noise of the SD11G5 was very low, especially the measured SPLs. Yet, the subjective perception was not as positive. It has to do with the annoyance quality of the stock cooling fan. It’s simply too buzzy to sit next to for any length of time. If your’re using the SFF as a desktop PC, its close proximity will mean that even in rooms with higher ambient noise than our quiet lab, you may still be bothered by the buzz. I would not choose live with it.

Even though our standard review policy is to assess products in stock form, in this case, a fan swap mod was too simple and too tempting not to try. The CPU cooling system in the SD11G5 has enough headroom to accommodate a quieter, slower fan like the Nexus 92. Even if the fan controller ends up pushing it to a higher speed than with the stock fan, the smoothness of a Nexus would lower the overall noise. Couple this with cutting out the back grill and removing the inside wire grill, and you might end up with a very quiet system, one that’s subjectively as quiet as the low SPL measurements would suggest, without the annoying buzz of the stock fan.


A Nexus 92 fan swap.

It wasn’t quite as simple as just swapping the fan, however. When a 3-pin fan is plugged into the 4-pin fan header for the CPU fan, all the functionality of the fan controller in the BIOS is lost. Regardless of the BIOS fan setting, the fan runs at full speed. So in order to accomplish the mission, another Zalman Fanmate fan controller was plugged in between the motherboard fan header and the fan. With the monitoring feature in the BIOS, the fan speed was adjusted to ~920 RPM with the Zalman Fanmate. The outside grill was left untouched, although it is far more obstructive than the inside wire fan grill, which was left off. A full set of tests was then run again.

Configuration 3: SD11G5 w/ Nexus 92 fan swap + on-board VGA

The measured SPL was only one or two dBA lower, but what a huge subjective difference this fan swap made! It was like night and day from a typical operator position at the desk. With the buzzing eliminated, all that’s left is the softest of whooshing, a gentle broadband sound that’s easily tuned out. Without the slight speeding up of fan speed that the BIOS Smart Fan setting provided under maximum load, the CPU temperature went up a couple degrees higher, but it was more than compensated by the absence of any change or increase in system noise.


SD11G5 w/ Nexus 92 fan swap + on-board VGA
System Activity
AC Power

CPU Temp

Fan RPM
SPL*
Board Temp?†

EIST Idle

32W

27°C 

920

20

40°C

Idle

41W

30°C

920

20

41°C

Full Load

56W

48°C

920

20

46°C

3DMark05

58W

48°C 

920

20

47°C

* Sound Pressure Level measured in dBA at one meter from the front bezel with the fan control set to Smart Fan or Ultra Low.
† It’s not clear what several temp sensors picked up by SpeedFan actually monitor. These readings are from the sensor that always gave the highest temps — just for the record.

Configuration 4: SD11G5 w/ Nexus 92 fan swap + AOpen Aeolus 6800GT VGA

CPU temperatures climbed a bit higher still when the AOpen Aeolus 6800GT VGA
card was reinstalled. The “mystery” temperature also climbed further, and SPL went up by just 2 dBA. But the subjective increase was much greater — the VGA cooling fan has as bit of a buzzy quality too, although it is not as bad as the stock Shuttle 92mm fan. Again, despite the low overall level, this acoustic performance is not one I would tolerate at my desk because of the higher annoyance factor.


SD11G5 w/ Nexus 92 fan swap + AOpen Aeolus 6800GT VGA
System Activity
AC Power

CPU Temp

Fan RPM
SPL
Board Temp?†

EIST Idle

74W

29°C 

920

22 dBA

48°C 

Idle

83W

32°C

920

22 dBA

48°C

Full Load

94W

51°C

920

22 dBA

49°C

3DMark05

133W*

51°C 

920

22 dBA

56°C 

* Peak value. Sustained maximum was 125~130W.
† It’s not clear what several temp sensors picked up by SpeedFan actually monitor. These readings are from the sensor that always gave the highest temps — just for the record.

SOUND RECORDINGS IN MP3 FORMAT

Sound Recordings of the Shuttle SD11G5, various test configs

Config 1: SD11G5 Onboard VGA, Smart Fan, idle: 21 dBA@1m

Config 1: SD11G5 Onboard VGA, Smart Fan, max load: 23 dBA@1m

Config 2: SD11G5 + 6800GT (5V fan), Smart Fan, idle: 24 dBA@1m

Config 2: SD11G5 + 6800GT (5V fan), Smart Fan, 3DMark05: 26 dBA@1m

Strange AC/DC adaptor sqealing noise when SD11G5 was turned off

Config 3: SD11G5 + Nexus 92 fan, 920 rpm (any load): 20 dBA@1m

*

Sound Recordings of System Comparatives

AOpen EY855-II w/ P-M 1.6 GHz: 25 dBA@1m
First 10 seconds in normal config; last 10 seconds with PSU fan forcibly stopped.

Shuttle XPC SB86i with Samsung Notebook Drive, Idle: 29 dBA@/1m

Shuttle
XPC SB86i after 92mm Nexus Fan Swap, Idle: 26 dBA@1m

(No point checking the SB86i at full load; it jumps to 33 dBA@1m)

Shuttle XPC SN95G5 system in idle: 27 dBA/1m
(The first 7 seconds of this recording are in idle, the next 7 seconds are with the case pressed tightly between my hands. The last portion reverts to the case sitting free.)

Shuttle XPC SN95G5 system at maximum load: 30 dBA/1m

Arctic Cooling Silentium T2 with test system (3.5″ HDD suspended) at idle: 23 dBA@/1m
(No point checking the T2i at full load; it jumps to 34 dBA@1m)

HOW TO LISTEN & COMPARE

These recordings were made with a high
resolution studio quality digital recording system. The microphone was 3″ from
the top edge of the front bezel. The ambient noise during all recordings was 18 dBA.

To set the volume to a realistic level (similar to the original), try playing this Nexus
92mm case fan @ 5V (17 dBA/1m) Reference
and setting the volume so that it is barely audible. Then don’t reset the volume and play the other sound files. Of course, tone controls or other effects should all be turned off or set to neutral. Don’t touch the volume setting afterwards, and use the same one speaker when you listen to any of the other files. The end result should be reasonably close to the actual recorded sound levels; the fidelity will be highly dependent on the quality of your audio playback system.

For more details on how to calibrate your sound system to get the most
valid listening comparison, please see the yellow text box entitled Listen to
the Fans
on page four of the article
SPCR’s Test / Sound Lab: A Short Tour.

CONCLUSIONS

The Shuttle XPC SD11G5 is a landmark product, much like the earlier Shuttle ST62K
Zen
.
The latter Shuttle is about the quietest SFF barebones PC even now, and a modified version of the Zen is still used in the SPCR lab as a silent audio recording PC. The SD11G5 is a more versatile choice for both low noise and performance. Pentium M is the stepping stone to Intel’s high efficiency computing future, with its high performance and miserly power consumption. Just the basics alone are enough to make a winning combination: A Pentium M SFF platform around a proven single fan heatpipe cooling system with PCIe video card capability and an external fanless power box. Add Shuttle usual attention to details, styling and packaging, extra features like the ambitious onboard Sound Blaster 7.1 sound card and dual monitor display capability, and it looks just about perfect.

The AOpen EY855-II was first the Pentium M SFF barebones system out of the gate, but it cannot stay even with this latecomer. Its power supply makes a bit too much noise and it is not easily made quieter. Its cooling system is adequate, but again, cannot be made quieter easily. Its feature set is good but outshone by the SD11G5’s. AOpen’s newer socket 479 SFF systems, the EZ915-M and minute half-as-big MZ915-M might be worthy competitors, but we’ve yet to obtain samples.

Shuttle’s earlier socket-775 XPC SB86i doesn’t even come close to being a quiet SFF, despite the early PR hoopla about its quiet features. It’s flawed in too many ways to repeat here, but the big flaw is the socket choice itself: Prescott core Intel processors are not suitable for a quiet SFF, and this applies to every socket 775 SFF we’ve examined. Shuttle’s AMD socket 939-based XPC SN95G5 comes much closer, but ultimately, its built in power supply with too-small fans is an acoustic hurdle that cannot be easily overcome.

The suggested price of US$450 seems to be on the high side, but the market will quickly tell everyone whether the price is right and adjust accordingly. Aside from that possible wrinkle, the choice of a mediocre fan is the only serious spoiler in the SD11G5. With such low maximum heat in the box and tremendous cooling headroom, a 0.55A high speed fan is simply out of place. It suggests sloppy or hasty work, but Shuttle had many months to finess this design. The SD11G5 was first shown back at the start of June at Computex Taipei. Shuttle designers could have spent a little extra time and effort to pick a better matched fan for their creation. It’s not just about measured SPL, the quality of the noise matters! As our little Nexus fan swap amply demonstrates, it isn’t hard to greatly improve the acoustics of the SD11G5, just a single fan swap. Why didn’t Shuttle do this? Antec’s new NeoHE 430 PSU was fitted with a specially sourced high quality fan which really makes that product. The same could certainly have been done with this Shuttle.

Some readers will say Mike is nitpicking again; others will agree with me. Admittedly, sound levels comparable to SPCR’s
silent modified Shuttle Zen
are very possible with the SD11G5; the very same modications can be applied, and the only significant tecnical limitation is that the video card can use up an extra slot only on the trace side of its board. With over 200W DC output capacity in the AC/DC adapter, just about any current video card can be handled.

The SD11G5 will make many friends. It will appeal to:

Entertainment PC enthusiasts. The built in vidcard and sound might be perfectly adequate, as could the stock acoustic performance when placed across the room alongside a big TV screen while the speakers pour out music and video soundtracks.

Small / Home Office users. Low energy consumption, very low noise, excellent performance all around in stock form, great feature set (especially dual monitor support) and super easy to assemble: It’s hard to beat.

Schools. The aggregate noise of multiple computers in a computer workshop has become a major learning impediment for many teachers and students. A class full of these Shuttles would certainly be quieter than most computer classrooms.

Gaming / Performance PC enthusiasts. The Pentium M is already respected as a high-performance gaming processor. This SFF can handle a pretty hot PCIe X16 video card without flinching, even with ultra slow fan settings. Gamers don’t mind a bit more noise for a bit more cooling airflow. With its small size and clean looks, it could be the perfect “sleeper rig” for LAN parties.

Shuttle is firmly back on the quiet track with the SD11G5. (Now, if they could only use a 4-pin fan that’s as quiet and smooth as a Nexus…)

Much thanks to Shuttle
for providing the early XPC
SD11G5
sample.

* * *

Discuss this article in the Silent PC Review Forums.

Leave a Comment

Your email address will not be published. Required fields are marked *