Quite some time ago, my home network went partially wireless. Since that time I've lived with moderately lame performance from the Linksys components, which were the only components that worked even remotely close to as advertised for sake of reference. My wife's machine recently started having troubles, however, and they refused to go away despite my completely reinstalling Windows 98. I didn't know why they was occurring, but I didn't really care. She was now using a four-year-old machine, and I was getting tired of fussing with it. I decided it was time for her to graduate to Windows XP.
The problem with this, of course, was that her machine was a Dell Dimension system with a 400 MHz. Pentium II processor and only 128 MB of RAM. (Did I really just say only 128 MB? Sheesh.) That was a pretty beefy system when it was purchased in 1998, but today it's the proverbial 98-pound weakling of the computing world. Needless to say, I didn't think XP was going to run very well on it, so I started scrounging around the house for other parts. Fortunately, I had a Fong-Kai 320 case sitting around with an ASUS A7V133 motherboard and 512 MB of RAM intact. It lacked only a CPU to be a complete system.
So I headed off to Fry's Electronics, home of all things technological. I managed to snag an Athlon 2400+ CPU quite inexpensively (less than $200) and a ThermalTake Volcano 9 fan/heat-sink combo to keep it cool. Since the CD-ROM and floppy drives in the old system were also ancient by computing standards, I picked up a Samsung DVD drive and a Sony floppy drive on the cheap as well. The old system had what seemed like a perfectly good 20 GB drive in place—more on that later—the Linksys wireless networking components installed over a year ago, and I had a spare SoundBlaster SB Live! card lying around elsewhere after my recent upgrade to an Audigy 2.
Getting the hardware assembled went very quickly. I cleaned out the Fong Kai case—it is positively astonishing how quickly dust and other crap builds up out here in California—and removed the hard drive from the old system. I got the CPU and cooler mounted and ready to go within a couple of minutes. Installing the hard drive, the DVD drive, and the floppy drive were also pretty trivial. The hardest part of the whole affair was making sure that all the important leads (e.g., case fans, power switch, etc.) were still connected to the motherboard. Less than thirty minutes later, I was ready to give it a full-up test.
When I fired up the new system I was pleased to see that everything seemed to work. All the fans were turning, I got to the BIOS setup screen and configured the date/time and the new hard drive, and so forth without any trouble. After saving those changes, the computer booted into Windows 98 and started going through the drudgery of detecting new hardware. Once it was done doing its thing, and Windows 98 verified that all the components were working, I shut the system down and decided it was time to install Windows XP, a process which I expected to be equally simple in light of past experience with my laptop.
Unfortunately, things didn't go at all as planned. I powered up the system, configured the BIOS to boot from a CD-ROM, and put my Windows XP installation CD in the drive. Setup started just fine, and I soon had it trundling away at converting the existing hard drive to the NTFS file system. Windows XP completed its installation procedure, and I was somewhat surprised to see that the Linksys networking components were not recognized properly. XP "knew" it had a network card, but it didn't know what it was.
For the record, the machine had a Linksys WDT11 PCI card adapter, which supplies a PC-Card slot into which a Linksys WPC11 wireless network interface card (NIC) may be inserted. These days Linksys has a one-piece solution, namely, the WMP11 wireless PCI card—more on that in a moment—but no such thing existed when my wife's system first went wireless, so I had to use those two separate components at the time. That combination had worked well enough for the last year and a half or so, giving a pretty consistent 20 - 30% signal strength and 1 - 2 Mbps performance over the 25+ feet that separated the card from the wireless access point (WAP). That's a far cry from the advertised 11 Mbps up to 90 feet, but I've vented my frustrations about that already.
At any rate, these Linksys components were on the market before Windows XP was released, so I was surprised to see that they weren't detected automatically. Still, I had downloaded the latest drivers before I started this process just to be safe. I've learned to expect such failings over the years. I installed the latest drivers for the WDT11/WPC11 combo, restarted the machine, and... nothing. The card was working properly, or at least that's what device manager told me, but I wasn't getting the wireless-networks-detected message with which I was by now familiar. Whatever happened to the zero-configuration wireless features of XP? I was now doubly puzzled.
Worse, no matter what I did I couldn't get a connection to the WAP. It wasn't being detected at all. I set the NIC parameters properly (e.g., infrastructure mode, channel three, the right SSID, etc.), but none of that made any difference. I tried enabling and disabling wireless encryption at the WAP, but that made no difference. I fussed around with antennas, drivers, and virtually everything else, but nothing I did seemed to work. I simply could not get that system to find a connection.
I decided to move my wife's new machine into my own office, which is where the WAP is located. To be clear, I didn't think it would make any difference in the network situation. Her machine had been connecting just fine from her office for over a year, after all, so it surely couldn't be any kind of weak-signal or range-related issue. I just wanted to move it closer to my own computers, so that I would have all my tools easily at hand in working on it further.
No doubt you can imagine my surprise when it connected immediately to the network on startup the very first time. The machine that previously couldn't find the WAP at all from roughly 25 feet away could now find it reliably from roughly 5 feet away. This struck me as odd to say the least, but I assumed that I must have missed something before. After all, it couldn't be a range issue. The zero-configuration wireless stuff still wasn't working, but I didn't much care; it was connecting, and that was what I really cared about. I offered quick thanks to God for the rare stroke of good luck and returned the machine to my wife's office to continue configuring it.
Yes, you guessed it: as soon as it was back in my wife's office, it again failed to locate any wireless network. Further experimentation proved conclusively what simply could not be the case; i.e., the same hardware that previously connected from my wife's office could no longer detect a wireless network at all outside of my office. Since the range obviously couldn't be the real problem, again because the machine worked just fine for over a year from that same location under Windows 98, I decided it was time to call Linksys technical support. I didn't have much success with them before, but I hoped they might have improved by now.
I called their toll-free number, and I was connected immediately with a technical support representative. That in itself was shocking. Every other time I had called, I had been forced to wait on hold for a non-trivial length of time. I (foolishly) took this as a good omen, and I began answering her questions, giving her my phone number, my name, and so forth. As soon as I started to describe the problem, however, she pressed the wrong button on her phone and disconnected me. I should have known it was too good to be true.
I waited. And I waited. The whole reason they ask for your phone number, you see, is so that they can call you back should the line get dropped. Or at least, that's the ostensible purpose for which they ask for your phone number. I don't know what they really use it for, but I'm sure of this much: they're not using it to call you back. It became obvious after five or ten minutes that the woman who disconnected me obviously wasn't calling me back, so I figured I would have to try again.
The second time I called was more typical. That is, I had to wait on hold for about ten minutes before someone picked up the line. I wish I could tell you who that was. Heck, I wish I could tell you what sex that person was. Honestly, I have no idea. The only things I can say with certainty are that (1) the person was Asian, and (2) English clearly didn't qualify as even a third language in light of the speaker's "proficiency". I hesitate at using that term insofar as it implies at least some competency, but I cannot think of a better word. I really tried to make that call work, but after about five minutes of nonsensical sentences, repeated questions, and complete non sequiturs, I gave up. I thanked whatever it was I was talking to and hung up.
The third call to technical support left me on hold even longer—fifteen minutes this time—before I got transferred eventually to some Indian (dot, not feather) chap. He was very helpful. He explained that this was a known problem with the latest drivers under Windows XP. He said that he would send me a set of drivers that weren't publicly available on the web site. Once these arrived by email, I needed only to update the drivers on my wife's machine and all would be well. I made sure he had the correct email address, thanked him, and hung up. In the spirit of J.R.R. Tolkien, I now offer you a musical interlude, sung to the tune of "Rawhide":
Waiting, waiting, waiting.
For drivers we are waiting.
Waiting for some email, rawhide!
Well it doesn't matter whether
I wait roughly forever,
while technical support laughs at my plight.
Ok, that wasn't much of a musical interlude, but it was the best I could do on short notice. The point remains the same: he never sent me any email. I worked on other things for almost an hour while waiting for his message, but it never arrived. In retrospect, I have to admit that he is a very effective support engineer. You see, companies frequently grade their technical support personnel on how quickly they can resolve individual calls. He had me off the phone in mere minutes, so I imagine he'll get a great score for the way he handled my call. Of course he didn't do anything to solve my problem, but that's not what support is about anyway as near as I can tell.
So, I called Linksys technical support yet again, and yes, I waited on hold even longer—twenty minutes this time—before I got to speak with a black fellow who had an admirable ability to pronounce words in the English language. I was practically jumping for joy when I could understand his speech. It's such a rare and beautiful thing to call technical support and get someone who speaks with any semblance of proper diction that this man's voice alone was like a breath of fresh air.
I explained the problem to him, and I described the previous three calls. He confirmed with me that support technician number three didn't have a clue what he was talking about; i.e., this wasn't a known issue, and there were no non-public drivers to try. His next step was to take me through some basic and uselessly redundant procedures—why is it that support folk have you repeat the very same steps you've just told them you already tried?—before he rendered his verdict. He knew what the problem was. I awaited his pronouncement from the mount with giddy anticipation. Ok, I wasn't actually giddy, but I don't use that word enough. What was my problem? Yes, dear reader, you guessed it: the computer was too far from the WAP.
It didn't matter to this fellow that the computer had maintained a wireless connection from that exact same location for over a year using precisely the same hardware under Windows 98. No, that's not relevant; the computer was too far from the WAP. What I needed to do, you see, was move the computer closer to the WAP, or perhaps move things around so that the signal passed through fewer walls. Yup, sure enough. That makes sense, doesn't it? (sigh)
The technician simply would not listen to reason, and he wasn't interested in escalating the call up the chain to his supervisor. I eventually hung up in disgust and decided I was on my own. Frankly, I'm better off wasting my time trying to fix the problem, rather than wasting my time trying to communicate with the Linksys technical support department. They're not going to admit that their drivers are at fault, and if they're not bright enough to at least come up with a creative lie, then they're not worth the effort.
What I didn't expect was that I would need to reinstall Windows XP. The system made this decision for me, however, and I wasn't consulted. During the process of fussing about with the drivers, I started to get errors invoking the management console. I didn't know what to make of them, really, so I simply kept working. Eventually, right in the middle of installing a driver of course, the system gave me the blue screen of death (BSOD), stating something to the effect that an awful thing had gone terribly wrong with the paging file. Oh goody.
Upon restarting the computer chkdsk ran for an interval just shy of forever, declared my drive fixed, and let me back into Windows XP. I breathed a sigh of relief and thought fondly of my decision to let it convert my drive to NTFS. After all, NTFS is supposed to be a safe file system. If the machine doesn't get powered down properly, or if other mistakes are made in writing to the drive, the operating system is supposed to be able to fix them thanks to the record-keeping structures of NTFS.
Unfortunately, some files on the disk remained pure, unadulterated crap. Windows XP had apparently repaired the damage to the logical structures of the file system, but it didn't repair the damage done to the files themselves. I was still having problems with the management console, and now my drive was running constantly, even when the machine was idle. The CPU utilization was bouncing all over the place. And still worse, I had no networking components installed. Seriously, if I examined the properties of the network connection, it showed no card, no protocols, no services, no nothing. Still worse, any attempt to change the networking stuff resulted in a host of cascading errors.
So I ran chkdsk again in its most thorough mode, hoping against hope that it would find some heretofore unnoticed error, but it didn't. The drive checked out. The operating system had screwed itself completely. I tried using the repair console, but nothing I did seemed to fix the problem. Windows XP couldn't repair itself, I couldn't repair it manually, and even re-installing it over itself failed to resolve the issue. Not knowing what else to do, I put the installation CD back in the drive and installed a fresh copy of Windows XP in a different directory.
I was secretly hoping that the installation would detect my hardware properly this time around, but it didn't. I was also hoping that I might see different behavior with the proper drivers in place this time around, but I didn't. Still, the operating system was at least working, and the drive wasn't constantly running. Thus, I set about investigating other possibilities, trying to come up with some explanation for the wireless network signal problem.
Eventually I made a discovery: I could just barely get a connection if I used the WPC11 card from my laptop instead of the WPC11 card that was originally in my wife's machine. What's the difference, you ask? Well, the WPC11 card in my laptop was a third-revision part, whereas the other WPC11 card was a first-edition part. Everything that I had seen to date led me to two conclusions: (1) the Windows XP drivers for the WDT11/WPC11 combo are broken compared to the same drivers for Windows 98, and (2) the WPC11v3 card is noticeably more capable than the WPC11v1 card.
I decided to see if I could come up with a workaround for my current woes. I went back to my office, and I relocated the WAP across the room, moving it a total of roughly ten feet closer to my wife's office. With the WAP in the new location, her machine could maintain a connection to it at about 10 - 15% signal strength. That's a spotty connection at best, and it's incredibly lame given that the two were less than 20 feet apart at that point, but I figured it was better than nothing and might serve as a useful workaround.
Further testing showed that I had found something good enough. As long as I kept the WAP at the most awkward end of my office, and as long as the computer was using the third-revision NIC, it could stay connected. All I needed now was a second WPC11v3 card. Sure, I could have simply swapped cards with my wife, but I teach at the University of Southern California, where a wireless network has recently been deployed on campus. It took non-trivial effort to get signed up for that service, and they authenticate on the basis of your NIC's internal ID. As such, I didn't really want to go to the trouble of getting a second card registered and working—if I could get the first-edition card working with the USC network at all.
Thus, I headed back to Fry's on a mission for two components. I needed a second NIC and a long run of ethernet cable for the WAP (I really had to stretch things uncomfortably in my prior testing). When I got to the store, however, I discovered that Linksys currently offers the WMP11 card instead. Until then, I didn't know that Linksys made such a thing. It's a wireless NIC that plugs directly into a PCI card slot, but that's not its best feature. Its best feature is that it has its own dedicated antenna on the back of the card. I figured that such an antenna would likely do much to solve my signal problems, and since it was on sale for $79 (with a $10 mail-in rebate to boot) I liked the price as well.
I got home, uninstalled the old equipment from the Windows XP device manager, powered down the system, removed the old hardware, installed the new card, and fired the system up again. Not surprisingly by now, Windows XP failed to identify the new card properly. I inserted the driver CD, however, and it was able to install the drivers therefrom without any further fuss. Once the card was in place, I edited its properties and configured it for my network. This time, for whatever reason, the zero-configuration wireless service seemed to be working. I had to fight with the system for a while, but after discovering how to enter a wireless encryption protocol (WEP) key directly, I was able to uninstall the Linksys configuration tools and use the built-in features of Windows XP to get connected.
My patience was ultimately rewarded. With the WMP11 card in place, I was getting a solid 11 Mbps connection at roughly 45 - 50% signal strength. My assumptions about the relative merits of an antenna were clearly justified. I still couldn't get better than 1 - 2 Mbps throughput from her machine, but I was accustomed by now to that kind of lame performance from wireless networking generally. I was nearly ecstatic to see that the system could get connected to the wireless network at all after roughly eight hours of working toward that goal.
Unfortunately, my ordeal was not yet over. Now that the network was working, I decided to get some of the other drivers installed. I needed the files for the SB Live! card, the Microsoft Intellimouse Explorer, and the video card as well. I was able to copy them all across the network, but when I tried to install the drivers for the sound card, I got some bizarre gee-I-can't-read-from-the-disk error that I hadn't seen before. I would have written the error message down, but while I was busy reading it the system decided that it was high time for another BSOD.
Upon rebooting the hard drive featured all sorts of errors, all of which chkdsk said it fixed. When Windows XP finally started, however, I was greeted with an entirely different set of bizarre errors. I could no longer access the control panel, and I was getting nasty messages about how kernel32.dll was failing to load. For those with little understanding or imagination, suffice it to say that's not good. Again, the system quickly rendered itself completely unusable. I was unable to repair it using the recovery console or by re-installing Windows XP over itself. Believe me, I was about ready to drop kick the whole bloody machine through the wall at that point.
What didn't make much sense to me then, and still doesn't to be honest, is that the hard drive had been working fine in the old system. Or at least, it seemed to have been working fine. Every diagnostic procedure that I could run reported that it was fine. Running chkdsk in its most thorough mode reported no bad clusters or anything like that. Yes, the drive was roughly four and a half years old, but it had been working just fine during that entire time. Still, all of the errors I had been seeing seemed to be related to problems with file corruption, file system corruption, and the paging file. That made me pretty suspicious of that old hard drive.
Thus, I decided that I wasn't going to reinstall Windows XP again until I had a different hard drive in its place. Since I had already made two trips to Fry's (and didn't want to spend any more money), I pulled the second hard drive from our house server after copying its contents to the first and erasing its partition table. It didn't have much on it, and it was mostly being used as a tertiary backup system anyway—yes, I really am that paranoid. A few minutes later I had the drive installed in my wife's computer, and I installed Windows XP for the third time that day.
It took quite a while for the setup routine to examine and partition the much newer 40 GB drive, laying down the NTFS file system before continuing with the installation. Nevertheless, the setup routine eventually copied all the requisite files and completed successfully. I subsequently installed the network drivers for the new WMP11 card and immediately obtained a connection to the WAP. I copied the audio, video, and other drivers to the machine across the network and installed those. I even installed Microsoft Office 2000 without incident. The third time seemed indeed to be the charm.
All told, I spent no less than ten hours working on her system that day, the majority of which was spent fighting the demons of wireless networking. I still don't have a clue why the old IBM hard drive from her original system was a problem, but something clearly didn't like it. Whether that something was the motherboard, Windows XP, or some other component remains unknown, and I can't say I care. I'm going to throw the old hard drive out in light of what I've seen already; I just don't need hassles. I did draw some conclusions along the way, which I'll summarize here for future reference:
Wireless networking still sucks as far as I'm concerned, but the newer hardware is definitely better than the old. At least now my wife's machine can get a solid 11 Mbps connection across the roughly 25 feet that separate her computer from the WAP. The throughput never rises much above 2 Mbps, but at least it's able to connect. The older hardware can't accomplish even that much. Perhaps the forthcoming 802.11g standard will do something about that. Only time will tell. For the moment, I'm hoping my wireless woes are over.
12/21/2002