Fullscreen HTML5Med Res (??MB)HTML4

COMPUTER BITS BY GREG SWAIN Buying a PC isn’t always hassle-free There’s lots of room for misunderstandings in the com­puter industry. Often, it’s the customer’s lack of expertise that causes the problems but sometimes it’s the dealer who is at fault. If anyone asks me for advice on buying a PC, I always tell them to choose the retailer carefully. Why? –because my own experiences with PCs from several retailers haven’t been all that good. Admittedly, some of the problems are fairly trivial and easily fixed if you have any technical ability. But why should you have to fix problems on a computer that’s fresh out of the box? And if you don’t have any technical ability, taking the machine back under warranty can be downright inconvenient. Falling into the trivial (but inconvenient) category is the recent experience of a near neighbour. He was keen to buy a new PC and, having done his homework, ordered a Pentium II 333MHz machine with 32MB of RAM, a 17-inch monitor and Windows 98. Unfortunately, things didn’t go exactly according to plan. Oh, he got the machine OK within a few days of ordering it but when I asked him a week later if he was happy with his new toy, I was told that it didn’t work. Pressed further, he explained that when ever he attempted to boot If you have two IDE drives on the same port, one must be configured as a master and the other as a slave (unless you are using cable select). This is done using jumpers which are usually located at one end of the drive. A label on the top of the drive or near the jumpers shows the various options. 86  Silicon Chip the machine, it would come up with a dialog box which stated that the mouse couldn’t be found. The machine would then hang and refuse to complete the boot sequence. But that wasn’t the end of the story. Apparently, he had ordered a Micro­ soft Intellimouse with the machine but the vendor didn’t have one in stock and had substituted a Logitech mouse instead. And when he had rung to complain that the computer couldn’t “find” the mouse, he was told that a Microsoft mouse driver had probably been installed and that this was causing the problems. All he had to do was install the correct Logitech mouse driver and all would be OK. Unfortunately, he didn’t know how to go about this, particularly as the machine wouldn’t boot up in the first place. As far as he was concerned, he would have to make a special trip back to the vendor to get the situation resolved. I must confess that I was rather puzzled by the symptoms. Simply changing the mouse shouldn’t bother the operating system. Normally, if new hardware is connected, Windows 98 detects it during boot-up and prompts you to insert the Windows 98 CD-ROM (or a disc supplied with the device) so that the appro­priate driver can be installed. After that, it should complete the boot sequence. Curious about the symptoms, I volunteered to have a look at the machine for him. Maybe we could get it working with my Micro­soft mouse and sort things out from there. As it turned out, the problem was fairly straight­forward. After following a couple of false trails, I eventually discovered that the keyboard wasn’t working either. Further inspection revealed that both the mouse and the keyboard were fitted with PS/2 con­nectors but the sockets they were plugged into were un­ l abelled. A quick check in the mother­board manual revealed that my neighbour had transposed the connections, plugging the mouse into the keyboard port and vice versa. Swapping the connections over solved all our problems, the machine now booting normally into Windows 98. So the dealer hadn’t really done anything wrong and the substituted mouse had only served to confuse the issue. But how was my neighbour supposed to know which PS/2 port was which? Clearly, a couple of 5-cent labels would have saved him a great deal of time and frustration. The NT machine My next story concerns a machine that was bought by a friend from a local retailer for use in a small business that he owns. This was quite a well-specified machine that came with Windows NT Workstation, a 300MHz Pentium II processor and 128MB of RAM. And there were lots of other goodies as well, including a 6.4GB IDE hard disc drive, a 32-speed CD-ROM, an internal IDE ZIP drive, a Matrox Millennium II video card, a network card, an internal 56K modem and a 17-inch monitor. The new machine was bought to play a central role in his office network but there was just one problem – it didn’t work properly. In fact, it came with a number of faults, as follows: (1) the hard drive indicator LED was permanently lit; (2) the machine often hung at the Windows NT splash screen during boot up; (3) it was very slow to log on to other computers on the network and often missed some connections altogether; and (4) the modem didn’t work. Unfortunately, he didn’t have time to take it back to the retailer to sort the problems out. He really needed to have the machine up and running that weekend for an important project that they were working on, which was how I got involved. To begin, I decided to find out why the hard drive indica­ tor LED was staying on, even when there was no drive activity. This turned out to be straightforward – the hard disc drive and the CD-ROM drive shared the pri- mary IDE port but they were not correctly configured as master and slave. Instead, the hard disc drive was configured as a “single drive, no slave present”, while the CD-ROM jumper was in the “master” position. To make matters worse, the ZIP drive was on its own on the secondary IDE port but was configured as a slave. Who ever set this machine up obviously didn’t have a clue about correctly configuring IDE drives in master/slave relation­ships. To overcome the problem, the hard drive was left on the primary IDE port and the CD-ROM moved to the secondary IDE port, along with the ZIP drive. The jumper on the back of Fig.1: if you don’t have NT Server and are using the ZIP drive was then set NetBEUI for your local area networking to the “master” position, protocol, disable the TCP/IP bindings for the while the CD-ROM was network card. If you don’t, you will get an error configured as the slave. message each time you boot up (ie, “The DHCP After that, the indicator client could not obtain an IP address”). LED only lit when there was decided to remove the network card drive activity. for a closer inspection but it proved Of course, we could have left the surprisingly difficult to remove from CD-ROM on the primary IDE port had its slot on the motherboard. In fact, it we wished. The hard drive would was jammed in so tightly that it took then have been config­ ured as the a fair amount of force to free it. master and the CD-ROM drive as the The reason for this wasn’t hard to slave. However, it’s best not to do this find. The backplane blanks on this as having the CD-ROM drive on the machine are normally secured by same port as the hard disc drive can small metal “bridges” at either end sometimes slow things down. and are removed by “knocking” them Problem number 2 – hanging at the out. Unfortunately, this had badly Windows splash screen during boot- distorted the backplane metalwork up – was tackled next. We figured that so that it pushed hard against the this could be a video driver problem, backplane bracket when the network so we logged onto the Matrox web card was inserted. site, downloaded the latest driver for When the metalwork was straightthe Matrox Millennium II card and ened, we found that the network installed it as per the instructions. card now slid easily into its slot on And that was it – the machine now the motherboard. What’s more, the booted into Windows NT every time, machine now quickly found all the although it was still very slow to log network con­nections and logged on onto the network. in the normal manner. Apparently, The network problem was the next the network card had been forced so in line. The symptoms here were far sideways that it was only making rather puzzling – why did it recognise intermittent contact with some of the some network connec­tions on some slot contacts. occasions but not on others? And why How the ham-fisted clots ever got was it always so slow to log onto the the network card into its slot in the network? first place without breaking anything, When the software settings all I’ll never know. Anyway, that was checked out, we tried reinstalling the problem number three out of the way. driver for the network card but this Just one further point here. If you made no difference. In the end, we aren’t using NT Server and are using December 1998  87 This 120MHz Pentium processor caused all sorts of problems in a machine that had been upgraded from Windows 3.11 to Windows 95. There was nothing really wrong with the processor though; it just didn’t like being overclocked at 133MHz. NetBEUI as your local area networking protocol, be sure to disable the TCP/IP bindings for the network card, otherwise you’ll get an error message each time you boot – see Fig.1. Problem number four was the non-functioning modem. The problem here was that this was configured as a Plug and Play (PnP) device but Windows NT4 is not really a PnP operating system (unless you install the PNP drivers). We changed the jumpers to turn the PnP feature off, then checked that the other jumpers set the modem to COM2 IRQ3 (as set out in the manual). This done, we reinstalled the modem and disabled the exter­nal COM2 port in the system BIOS (note: if you don’t do this, you can get conflicts with an internal modem card that’s set to COM2, even if there’s nothing connected to the external port). Finally, we installed the modem driver software as instructed in the manual, after which the modem worked perfectly. After that, it was mainly a matter of tidying things up. For starters, the internal cabling to the disc drives was quite messy and I spent some time rerouting the cables to tidy things up and to eliminate any strain on the connectors. I also noticed that the vendor had only installed Service Pack 1 for the NT Workstation operating system, despite the fact that Service Pack 3 has been out for ages. We installed Service Pack 3 with88  Silicon Chip out any problems, the only glitch being that this wipes out the ZIP drive installation. That problem was overcome by reinstalling the drivers for the ZIP drive, following the step-bystep procedure listed in the manual. We then finished off by changing the screen resolution from the vendor’s 640 x 480 setup to 1152 x 864. This is something the vendor should have done as a matter of course, given the hardware involved (8MB Matrox Millennium II graphics card and 17inch monitor). All in all, the whole exercise involved several hours of work that should not have been necessary. The problems that this machine had were all too obvious. Either the vendor neg­ lected to test the machine properly or if they did, they lacked sufficient technical expertise to recognise the problems and fix them. Or maybe the person who set it up just didn’t care. In the end though, it’s the retailer who will lose out. Guess where my friend won’t be buying his next computer! The overclocked Pentium My last story concerns a Pentium 133MHz machine that was pur­chased several years ago, again by a friend who runs a small business. It came complete with 32MB of RAM, a sound card, a network card, a CD-ROM drive, Windows 3.11 – and the Sepultura virus! Fortunately, the virus was discovered immediately and cleaned off before it had done any harm. The machine worked fine under Windows 3.11 but all sorts of problems arose when the operating system was eventually upgraded to Windows 95. System crashes were a common problem and the machine also often refused to close down properly. On other occasions, it would even refuse to boot correctly. Eventually, the machine landed on my desk and many hours were wasted trying to solve the problem. Initially, we suspected a software problem so we stripped the machine down to its bare essentials and removed all unnecessary drivers. When this didn’t help, we tried upgrading the video driver but again drew a blank. Next, we backed up the data, re­ formatted the drive and reinstalled the operating system. That didn’t help either but one thing was becoming apparent – instability problems only arose after the machine had been running for some time with the lid on. Based on this observation, we decided that the fault must be heat sensitive. We tried swapping the RAM and the video card over from an identical machine without result and then noticed that the processor in the crook machine ran much hotter than the processor in its twin. The full story was revealed when we removed the processor from its ZIF socket and inspected the markings on the underside. It wasn’t a 133MHz CPU as ordered but a 120MHz CPU that was being overclocked! We changed the bus speed from 66MHz to 60MHz (so that the processor now ran at 120MHz instead of 133MHz) and that solved all our problems. Understandably, my friend was furious but his subse­quent complaint to the retailer about the processor mix up was badly handled. There was no apology nor any offer of compensation, although they did eventually agree to exchange the CPU. But it was all too little too late. In fact, my friend was so annoyed at what had happened and by their attitude that he seriously consid­ered taking legal action to recover his costs. In the end, he simply decided to get even. His company has purchased several new machines over the last few years and will buy lots more in the future. Guess who doesn’t get the orders? SC