Once a year or so, I get the crazy idea that I want to build a small form factor PC.
It’s not the actual small form factor system that’s crazy. It’s what I want it to do. What I typically want is to build a high performance gaming system using a micro ATX motherboard, building it into a micro ATX chassis. I’m typically lured in by some cool looking micro ATX case. That happened to me with a couple of Silverstone SUGO models (SG01 and SG03). Most recently, it was the NZXT Vulcan, which I reviewed for Maximum PC. The Vulcan is actually a good looking case with an attractive feature set that makes it sound like a badass, high end case.
As I discovered, the Vulcan is really a normal micro-tower in the skin of a high end, badass case. I don’t want to talk about the Vulcan itself much, though. Go read the review if you want more detail on the Vulcan. I want to talk about the process of building a true micro ATX system.
The idea of building a small, powerful rig that can take on most modern PC games is an attractive idea, and when you see a good looking case like the Vulcan, it’s easy to be seduced.
What you can’t do is build a small form factor system the same way you build a typical mid or full tower PC. For one thing, the order of installation is critical. In some SFF cases, for example, you can’t install the power supply until everything else is in. In addition, the interior of these cases is often very, very cramped, so you’ll need to double-and triple-check that some stray wire or cable hasn’t crept underneath the motherboard. Routing wiring is also important, as its easier for wiring to drop onto a fan, resulting in a horrible sound and possible CPU overheating if it’s the CPU cooling fan.
At any rate, I had an early Gigabyte GA-H57M-UD3 motherboard. Given its layout and use of USB 3.0, I thought it would be an ideal motherboard for building a gaming rig.
Gigabyte's GA-H57M-UD3
The CPU would be a Core i7 870 I had leftover from an old test system, a Seagate 1TB 7200.12 hard drive would handle storage and a Radeon HD 5850 would take care of GPU chores. Installation was the usual painfully slow process (it always is with these tiny systems), but the system booted up just fine. Installing Windows 7 went swimmingly.
Then the stability problems began cropping up. Initially, warm reboots would always hang — I’d have to disconnect power before the system would boot. After installing all the drivers, the system would periodically crash with ugly blue screens. I tried different DDR3 modules and memory settings, but the crashes still occurred.
Even BIOS updates didn’t fix the problems,. What I suspect is that this board is actually an engineering sample, not a retail release, since Gigabyte had shipped it to me within days of its announcement. Note that this board had never been used, and is curiously lacking in markings, though it does have a serial number.
So I thought I’d try another board, an Asus P7H55D-M EVO.
Asus P7H55D-M EVO
I actually had this board working in a another small form factor system that I’d partially disassembled, so I thought it would be pretty simple to rip it out of the other case and slap it into the Vulcan.
By this time, I was in what may be charitably described as a “system building fugue.” You know the drill — you’re so into solving problems with a new build, so into getting it completed, that you start taking shortcut, and start getting careless.
The process of removing the P7H55D-M EVO was less gentle than it should have been. I also encountered several issues when trying to get it into the NZXT case. First, several of the mounting nuts had slipped out, and I didn’t notice them until I’d wiggled the motherboard into the case. So I wiggled the Asus board back out, and reinstalled the nuts.
In my haste, I’d replaced one of the nuts in the wrong hole. Out came the motherboard, then back it went. Oops, a piece of flashing on the ATX I/O plate had been pushed back down, blocking the Ethernet port. Out came the board again, and back in it went.
Somewhere in the process, I’m sure I killed the board. Note that the P7H55D-M EVO had been working perfectly when I removed it from its original case. When I got everything installed, the board wouldn’t POST. Fans would spin up, but no video would appear. I attached an external beeper to the speaker connector, and no beeps could be heard. The memory diagnostic light would stay bright red, no matter what DDR3 modules were installed. The board seemed completely hosed.
Out came the board one more time.
I had one more micro ATX board, an Intel DP55WB. This is by no means a high end board, but it was what I had available.
Intel DP55WB
This time, I realized I had just one more try at this. So I walked away from the whole affair until the following day. When I returned, I exercised extreme care during the installation — something I should have done with the Asus board. I made sure all the cables were cleared out of the way, all the mounting hardware properly installed and that the ATX I/O shield was pristine. The board went in as easily as it could — the interior is still pretty cramped, after all — but in it went.
The system POSTed just fine. Driver installation went smoothly, as did the BIOS update. The system burn-in tests completed without a hiccup.
In the end, what I had was still a pretty powerful system for gaming, but it won’t be particularly overclockable. But then, a small case with restricted airflow precludes any serious overclocking anyway.
There are lessons to be learned. One is to be sure to use actual retail gear, not engineering samples. Most users won’t run into this anyway, but it’s still worth stating. The second lesson is more important: if you find yourself frustrated and tempted to rush the building process, STOP. Walk away. Come back later, after the fever has abated. You should always approach the actual act of building as dispassionately as possible. Celebrate after everything is up and running, and get excited about how you plan on using the system. If you get too wound up during the building process, you’ll likely do something dumb, or overlook something obvious. It’s a lesson I need to re-learn once a year or so. I just wish the price this time wasn’t two motherboards.
Loading tweets...
9 comments
1 ping
Daniel J says:
July 20, 2010 at 7:25 pm (UTC -7 )
Great read Loyd! I am 14 and have built two systems very similar to this. The only real difference is that I used an EVGA p55 FTW and an Intel core i5-750.
I always wondered what building a Micro ATX machine would entail. I thought it would be far more tedious.
Graham says:
July 21, 2010 at 6:19 am (UTC -7 )
Lloyd this is a wonderful article. It’s one I shall re-read a couple of times, as it’s a topic I have been considering for months. A small, quiet yet powerful PC is a dream I have had for a long time.
As a father of three teenagers with limited room in my study (strange how the children’s unwanted tech ends up in my space), the self built, full tower, workhorse is looking decidedly 1980′s. I have recently bought shuttles and laptops for the wife and children but poor old Dad is last on the list for new kit. This article has re-fired my interest in building another PC and this time it might even fit on the desk!
It’s strange that all of the other tech I use (iPod, iPad, Netbook, Router) has shrunk in size over the last 10 years and yet I have continued to place new motherboards and components in my trusty old Lian Li tower case.
Keep up the great work.
Graham
Loyd Case says:
July 21, 2010 at 5:14 pm (UTC -7 )
It’s pretty tedious — more so than building a mid-tower. And with some cases (NZXT Rogue, Silverstone SG01/02), the order of what you install is critical — the PSU has to go in last, for example.
Sam says:
July 22, 2010 at 3:34 pm (UTC -7 )
I found myself laughing while reading your article, since I’ve been there myself (just like you, I love small quiet PCs). My last attempt has been my mATX gaming rig built in a Silverston SG-03, with a MSI X58M, Core i7 920 D0 cooled by a Scythe Shuriken Rev.B, 6 Gigs of 1600Mhz Patriot RAM, two 3.5″ 7.200rpm Seagate HDs (1TB and 500Gb) + Sony DVD-RW, a lovely (and surprisingly quiet!) Sapphire HD4890 Vapor-X and a Antec Quattro 850W PSU. I love this machine and wouldn’t trade it for anything else in the world, but boy was it hard to assemble!!!
One thing I learned the hard way in this build is that YOU MUST PLAN THE THERMAL IMPLICATIONS OF SFF BUILDS IN ADVANCE!!! You, of all people, probably know how daring a proposition it is to try to fit (thermal wise) an i7 920 into a SFF case, right?
I ran rings around the house to solve the constantly overheated IOH and CPU when gaming. The solution was to undervolt by 0.1V my overclocked CPU –3.2Ghz– (which shows you how incredibly powerful and resilient these 920 D0s are) and to install a slightly moded Thermaltake Extreme Chipcooler Spirit II with Arctic Silver 5 thermal grease on the northbridge (which brought the IOH idling temp from 85C to 55C and load temp from 105C to 75C). Not to mention that I had to mod a small Silverstone FP33B fan controller to be able to tone down my two 120mm Enermax Magma TWISTER 1500 RPM fans when I’m not gaming.
It was a lot of work!!! But the end result is a beautiful, small (albeit slightly heavy for its size), cool, silent (when I’m not gaming that is
and incredibly powerful system that I can lug around without having to rent a Uhaul. The best part is that I have been using this little monster for over a year without a glitch! So, yes, it is a lot of work (and sometimes “re-work”), but the end result can be incredibly rewarding if done right.
50crckt51 says:
August 3, 2010 at 5:09 pm (UTC -7 )
Then the audacity of some standoffs to be course threaded! I mean, why couldn’t they stay with a common standard! It’s about as annoying as having to mod an IO shield
Nicole Ellis says:
August 12, 2010 at 10:16 am (UTC -7 )
for CPU cooling, i always use heatsinks with heat pipes because they are more efficient in cooling CPUs”:”
Lyle says:
January 24, 2011 at 2:50 pm (UTC -7 )
I’d like to add a couple of other caveats to the list of no-no’s
Never have small children or playful animals in the workspace.
Never believe for a second that just one more coctail makes it easier.
There are many many more, but these two each got me once in the past.
Been building personal computers since my first Imsai 8080 in 1976, literally hundreds of generic, obscure and some really nasty upgraded machines for family, friends and research.
Difficult or easy, wouldn’t have it any other way.
regis says:
January 27, 2011 at 11:02 am (UTC -7 )
Hello Loyd:
I’ve been a long time follower. You may remember my emails from your days at extremetech.
What is the command to turn off all background processes in Windows 7 before you start testing hardware?
Thanks for your help with this.
Regis
admin says:
February 8, 2011 at 2:09 pm (UTC -7 )
I no longer do this for Win7. It’s pretty stable after about ten minutes, and does a good job of minimizing or shutting down background processes on its own. You can obtain very repeatable benchmark results now without this.
Tweets that mention Third Motherboard’s the Charm « All Topics « Improbable Insights -- Topsy.com says:
July 21, 2010 at 4:07 pm (UTC -7 )
[...] This post was mentioned on Twitter by loydcase, Tom Ratas. Tom Ratas said: RT @loydcase: Latest blog post, in which I kill two motherboards to get a working system: http://bit.ly/dfMLYO [...]