Networking Headache

June 4, 2007 1:42 PM

Internal wireless vs. USB. What's better for desktop networking?

I think I have finally identified why I keep getting a "machine_check exception" error and BSOD on my desktop. I unplugged the Linksys wireless USB network adapter this morning and so far so good -- this after getting BSOD within minutes of powering on with the wireless USB connected. But of course I need Internet and an ethernet or direct connection is not practical. My question is, if I ditch the USB connection and install an internal wireless adapter in one of the expansion slots do I have a better shot of avoiding the error? I suspect it was related to overheating or some internal conflict caused by the USB connection but I'm not sure. Do these internal adapters produce less heat, and are there other factors I may be overlooking? I can go ahead and try it, but I wanted to ask before making the trip to Radio Shack.

posted by terrier319 to technology (12 comments total)

I can't say for sure if the Linksys adapter was causing the BSOD, but it wouldn't be necessarily from overheating, but rather sapping more electricity from the motherboard than it could handle. Did you have other USB devices installed on that machine? A PCI wireless card will not draw its current from the same circuitry on the motherboard as the USB controller, so if that was indeed the issue then it shouldn't be repeated.
posted by Burhanistan at 1:52 PM on June 4, 2007

Sounds like a hardware fault to me. Can you still return it? If not I've had a lot of luck with this linksys. You'll also get better reception as it has a full size (no idea what -db) antenna.
posted by damn dirty ape at 1:53 PM on June 4, 2007

First off - you'll probably be overpaying for anything at Radio Shack - so you might want to look online for an internal wireless card. I'd suggest trying newegg first.

Secondly - yeah - you probably have a good shot at fixing the problem by eliminating the USB aspect. Do you have a more detailed error message you could post, though? That might give you a better clue as to what exactly is causing it.
posted by odinsdream at 1:53 PM on June 4, 2007

I meant this card, which doesnt have that speedbooster crap.
posted by damn dirty ape at 1:55 PM on June 4, 2007

USB networking has always been kinda wonky, for whatever reason. Depending on your chipset and network drivers, it can range anywhere from 'tolerable' to 'sucks incredibly badly'. It wouldn't shock me to find out that you have an ATI chipset; early ATI motherboard chipsets had USB implementations that were so badly broken that they were useless for anything but keyboards and mice.

PCI is well-understood, and very, very thoroughly debugged. I'd definitely go that route over a USB connection.

I don't have any personal experience with it, but this ASUS card at Newegg would be what I'd try first if I were in your situation. I know ASUS makes good stuff in general, and tends to have good support... so for $22, I'd probably try that card first. It's gotten a couple of recent bad reviews, which worries me a little, but.... I'd still try it if it were my $22.
posted by Malor at 2:13 PM on June 4, 2007

As my fate would have it, just as soon as I posted this question the BSOD returned. I unplugged the only other USB device connected -- a wireless mouse -- so I will monitor this over the next 24 hours. I have read elsewhere that machine_check_exception errors may be related to overheating or overtaxing the motherboard/power supply. The only other thing I can identify was installing an upgraded USB port, to 2.0, a while back. So my next step would be to remove that and see what happens. The machine is not young -- over five years old -- but it works well otherwise. I would hate to dump it, but this is frustrating as hell. Sorry but I can't post the exact error as I rebooted quickly in frustration and it's gone now.
posted by terrier319 at 2:24 PM on June 4, 2007

First of all, don't just think that this is USB related. There's many other possibilites.

"A machine check exception occurs when Windows XP and your hardware platform cannot recover from a hardware error so that the system can continue to run successfully and reliably. More specific diagnosis of machine check exceptions is difficult, and there is no general solution. Contact your hardware manufacturer or a computer hardware technician for help with troubleshooting this issue.

Machine check exceptions are frequently caused by one of the following conditions:
• You are running the processor or mainboard beyond its specifications. For example, you are overclocking the processor or bus. We recommend that you run your hardware at the manufacturer-rated speeds.
• Noisy power, overstressed power strips, outmatched power supplies and failing power supplies can destabilize your computer. Make sure that you have a stable, reliable power supply to your computer.
• Extreme thermal conditions caused by the failure of cooling devices such as fans may damage your computer. Make sure that your cooling devices are all working.
• You have damaged memory or memory that is not the correct type for your computer. If you recently changed the memory configuration, revert to the previous configuration to determine what is wrong. Make sure that you are using the correct memory for your computer."



The first step I would take towards reaching a solution would be to examine your error logs and see if there's any information there. Sometimes there's a hardware code present that can lead to you finding the problematic piece of hardware.
posted by nokry56 at 3:21 PM on June 4, 2007

On a five year old machine with Machine Check exceptions, the first thing I'd be doing is whipping open the case and removing five years' worth of dust bunnies from the cooling fins.

Then I'd check for capacitor plague. If I found it, I'd have a serious think about upgrading; replacing mobo capacitors without excellent soldering/desoldering equipment and skills is not for the faint-hearted (I've done it several times with only an el cheapo soldering iron, but that really is about as much fun as taking out your own appendix using table cutlery).

Assuming the mobo looked OK, I'd run Memtest86+. If it found a fault, I'd run it again with only one RAM card in the machine until I found the one that was failing. Then, before replacing it with a new one, I'd clean up its edge connector with a pencil eraser (not an ink eraser - too abrasive) and test it again. Make sure you blow all the tiny rubber crumbs off before putting the card back in its slot.

If Memtest didn't find anything, I'd remove as many PCI and AGP cards as I could and still have a working machine, then refit them one by one until the machine check errors came back; then use the pencil eraser trick on the problem card's edge connector, or replace the card if cleaning it up had no effect.

If I still got machine checks with no cards fitted, I'd remove the CPU heatsink, clean off all the old heatsink goo, loosen the CPU socket lever a little, wobble the CPU around in its slightly loosened socket for about fifteen seconds to encourage any crap to scrape itself off the pins, then tighten it up, apply a blob of new heatsink goo and refit the heatsink.

If it didn't come good after all that, I'd have reached the limits of my considerable random-breakage experience, and I'd be forced to go Googling for more specific test tools.
posted by flabdablet at 5:11 PM on June 4, 2007 [1 favorite]

Flabdablet, provided an awesome response. I can only add a few suggestions.

One is that a power supply sometimes cleans up noisy power that can destabilize your system. I recommend them for anyone who cares about having their hard drives fail. They last a long time with battery replacements every four or five years.

If you are operating your computer in a hot environment, adding fans can help considerably. Pulling off all cables, blowing on them, and plugging them back in helps with some issues, but I'm not too confident it will fix yours.
posted by BrotherCaine at 11:39 PM on June 4, 2007

BrotherCaine raises an issue I'd forgotten: power supplies can suffer from capacitor plague too, and when they do, it causes random machine misbehaviour. If your present power supply is a cheapie generic, replace it with a nice Antec or Coolermaster or Thermaltake unit - as a bonus, this will likely make your machine a fair bit quieter because those guys use nice fans.
posted by flabdablet at 5:25 AM on June 5, 2007

And when he says "power supply", I think he's probably talking "uninterruptible power supply". I'm talking about the one inside your system box.
posted by flabdablet at 5:26 AM on June 5, 2007

Yes, I accidentally edited out the U in UPS, thanks flabdablet.
posted by BrotherCaine at 6:31 AM on June 5, 2007