Larry Osterman's WebLog

Confessions of an Old Fogey
Blog - Title

A snapshot of a developers morning...

A snapshot of a developers morning...

  • Comments 23

Wednesday evening, I went to a local computer store and bought a new 250G SATA drive for my dev machine.  Yesterday morning, I tried to install it.

I was a little bit apprehensive - my dev machine already has 2 IDE hard disks in it, and although the motherboard has two SATA connectors on it, I was quite concerned about what was going to happen when I finally put a disk into the SATA connector.  I've heard stories that many systems only let you use one or the other.

 

Fortunately I put the new disk in, cabled it up and powered the machine back up.  And it worked!  Yipee!!!

 

Now it was time to put things back together.  Because I was concerned about being able to use the disk, I had tested it out with the disk just cabled up and not installed, so it was time for me to find a home for the drive inside the chassis.

And that's where the problems started.

My machine only has 3 3 1/2 inch drive bays, and I was already using two of them.  Fortunately there was a floppy drive bay available so I decided to put the new drive in the floppy drive bay (this machine has no floppy drive).

Unfortunately the drive was a smidge too large to fit into the bay - the drive would physically fit, but the CPU's heat sink got in the way when I tried to line the drive up with the drive bay.

And now I made my big mistake.  I thought "Hey, no problem - I'll just take the CPU out, put the drive in and put the CPU back".

 

At this point anyone who's ever built a computer should see what's about to happen.

 

The first two steps worked great - the CPU came out just fine (although the rather weird clips that held the heat sink took a couple of seconds to figure out) and the drive fit in the bay just fine.

Then I tried to put the CPU back, and I realized what a horrible mistake I'd made.

You see, the socket for the CPU was a ZIF socket (in other words it was one of the sockets with a little lever that you pick up before inserting the CPU - that way you can't bend pins while inserting the chip into the socket).  But the heat sink extended out beyond the edge of the CPU by easily 3/4ths of an inch in each direction.  So there was no way I could insert the CPU back into the socket without first removing the CPU from the heat sink.

 

For those of you that don't know, the connection between the CPU and the heat sink is critical to the performance of the machine - if there isn't a good thermal connection between the two, your CPU will melt.

Unfortunately I didn't have any thermal paste on hand (for some reason it's not one of the bazillion things in my office and I didn't pick any up at the computer shop because I didn't know I'd need it).

 

So here I am, my development machine is literally in pieces on the floor of my office, the CPU is literally sitting on my desk (undoubtedly picking up a passive static charge that was going to nuke it the instant I touched it).  And I'm not getting any work done. Oh, and it's only 8:00AM - there's almost nobody around that can help me.

 

Fortunately we have a wonderful support group here at Microsoft, I opened a ticket with them and "Beckey the wonder-tech" was in my office within 45 minutes with a syringe filled with thermal paste.  10 minutes later my machine was reassembled with brand spanking new thermal paste between the CPU and heat sink and everything was wonderful.

Vista recognized my new drive, I expanded the existing dynamic drive that holds my source enlistments and I now have 250G more disk space to fill up with stuff.

 

Hmm, I wonder how long it would take to do a timebuild[1] on my dev machine?

 

 

[1] A timebuild is Windows-speak for a complete build of windows that results in a bootable OS installation (as opposed to taking individual files and copying them onto an existing installation).

  • You are lucky that you are *allowed* to buy a harddrive.  We have to put a ticket in to request for a new drive.  The tech ops team will then issue a issue a purchase order to an official vendor and wait for it to delivered, etc. *sigh*

  • Phew!

    I just wondered why a "timebuild" should be so-called. Any idea how the name came about?

  • I build my own systems and do it frequently.  The biggest problem with upgrading a retail computer is that the cases are designed to be difficult to allow upgrades beyond what the manufacturer decided should be allowed.  Some OEMs design special cases that are designed to take only their parts, but that more true with low-end systems.

    The Antec P180 case has some nice features to permit expansion, but it is not a typical case for OEMs.

  • Your dev machine has IDE disks ?? My dev machine has 2 15000 RPM SCSI disks in a RAID0 setup and those feel slow to me !

  • vajrang, my machine's actually quite reasonable - 3GHz P3 with hyperthreading enabled, 1.7G RAM, NVidia 6800U video, lotsa disk space.

    It does everything I need it to do.

  • That's a scarily overclocked P3, and it must be some special kit from Intel to have HT ;-)

  • My Bad, P4.

  • The last time I installed another harddrive in my machine I decided to put it in a 5.25" drive bay instead of the typical harddrive bay. I have it suspended on bungy cords so it's decoupled from the metal of the case.

    Because I'm into building quiet computers (I go to silentpcreview.com allot) doing this helps in reducing the noise that drives create when they're solidly attached to the case. It's also gives them allot of room to prevent excessive heat.

  • > I thought "Hey, no problem - I'll just take the CPU out, put

    > the drive in and put the CPU back".

    > At this point anyone who's ever built a computer should see

    > what's about to happen.

    I was about to correct that to "anyone who's ever unbuilt a computer" but then your story diverged.  A more common sad story involves an unbuilder expecting to remove the heat sink from the CPU.

    Now I'm trying to figure out how you were able to remove the heat sink + CPU combination from a ZIF socket without the same lever getting in exactly the same way.

    Friday, March 16, 2007 2:42 PM by Dominic Cronin

    > I just wondered why a "timebuild" should be so-called.

    It came from someone who rebuilt the Windows API without all its design bugs -- a stunt which requires using a time machine ^_^

  • Norman, I just pulled up on the heatsink and they came out.  Then disconnecting the CPU from the heatsink was a matter of twisting it.

    I DO NOT RECOMMEND ANYONE DOING THIS IF THEY DON'T KNOW WHAT THEY'RE DOING>

  • Are you sure you yank the cpu+heat sink out without first releasing the ZIF lever?  From what I read it didn't sound like you have release the ZIF lever first.  How on earth did you get it out w/out first releasing the ZIF lever.  Also did tech guy scrub the old thermal paste clean before apply the new coat of thermal paste.  If not you might have some problems.

  • cK - I just pulled the chip out and it came out.  If you think about how a Zif socket works, it makes sense (a ZIF socket is essentially "big holes for the pins, with a spring that keeps the pins in place (and a lever that controls the spring)" - I just tugged against the resistance of the spring and the CPU came out.

    And yes, the tech lady cleaned off both the CPU and the heatsink before re-

    installing it.

  • Pulling out the CPU along with the heatsink is quite "common" in my experience. Good thermalpaste glues the two parts together pretty hard at times, so it can be hard not to do at times.

    (My heart skipped a few beats the first time I did it though...)

  • Hey, Larry, long time no talk.  Remember me back from Exchange 4.0?  DS guy.

    Anyway, why'd you buy the drive.  Couldn't you just ask for one?

  • Hey Tim!

    I bought the drive because I needed it immediately.  I didn't want to wait for 2-3 weeks.  So I went out and bought a new one and I'll be expensing it.

Page 1 of 2 (23 items) 12