Dare to Upgrade
John Saxon installs a new motherboard and hard disk.
This saga recounts some recent (fun?) replacing my motherboard and my "small" 450Mb HDD with a new 1.6Gb unit. Hopefully the real hardware guru's will not laugh too hard! One recurring theme is that the Net is a wonderful source of information - the answers to most computer problems are definitely out there - somewhere.
First a description of my "old" system.
It was a generic 75Mhz Pentium with 32Mb Ram (of various types) with a 850Mb C: drive and 425Mb D: drive (generally used for backing up C:).Both drives were "doubled" - seems hard to believe one can run out of disk space with 2.5Gb on board! The P75 had been clocked at 100Mhz for a while with no ill effects. Graphics were via an ATI "Win turbo 64" PCI card with 2Mb DRAM on board. The system (originally purchased in the USA) also sported a 14.4KB internal Maestro modem, a Logitech hand held colour scanner using a bus card, and a "generic NEC" 6X CDROM drive. A fabulous sound card (Ensoniq Soundscape Elite) completed the ensemble, which ran Windows-95 (W95) pretty well - very well actually.
So Why Upgrade?
On a recent trip to USA I wanted to get a new BIOS, as my old one was not PnP and did not recognise HDDs larger than 528Mb - so messy driver software had to be loaded both for DOS and W95 use. A few inquiries via the net revealed that a new BIOS EPROM cost almost as much as a cheap motherboard, and so the plot was hatched. Also the old board used a PCI EIDE board with a "legacy" connection to use IRQs, plus extra boards for the floppy drives, slow comm ports, etc. W95 never would totally recognise my HDDs (the device manager said there was a conflict whatever I did - but they worked O.K.), and I never could configure my CDROM drive to the secondary EIDE channel, so I had to use it via the slower sound card I/F. All this coupled with a ready user for the old mother board and hard drive (not to mention the normal urge to play with the latest stuff), got me looking around the net for the best buys.
What did I get?
Once the decision was made to buy - the web search was started to find the most recommended motherboards and hard drives, I even checked an auction site at On sale, as well as hardware manufacturers and retailers. The beauty of the net is that it allows you to not only check specifications and prices, but also options such as warranty, shipping and payment methods, and it also gives a very good indication of how much support you can expect - do they have a 24 hr 800 number tech support? Can you work at long range via Email? Do they maintain a site to download the latest drivers? Etc. Of course the news groups are also an endless source of opinion (some probably uninformed) about the relative merits of different products. Quote:
"Usenet is like a herd of performing elephants with diarrhea--massive, difficult to redirect, awe-inspiring, enter- taining, and a source of mind-boggling amounts of excrement when you least expect it."
-- Gene Spafford (spaf@cs.purdue.edu), 1992
Unquote:
In the end I selected a Tyan Tomcat I motherboard - which (at time of writing!) is pretty much "state of the art" with the Intel HX chipset, eight SIMM sockets, 512Kb Cache, onboard EIDE, USB, Infra red, high speed Com ports (16550s), support for Cyrix CPUs, etc. Good info on this board can be found at www.tyan.com including a "clickable" motherboard photo! For the HDD I selected the Western Digital "Caviar" 1.6Gb as within my price range and offering good (9-10ms) access time - see WD's site. Both the manufacturers have good Web sites and offer loads of information and advice. I also got more advice and a check on some of the Aussie prices at Information Symphony
I went to two "swap meets" in Los Angeles to check the lowest possible local offerings and prices. One of the meets is huge (a bit like the Albert hall times 20), between 300-400 dealers (about 95% selling new equipment), all keeping an eye on each other's prices - but many do not even have stores or warehouses, and some of the items are so cheap, one wonders which truck might have had an accident nearby. My ground rule is to establish the base prices at the swap meet, but for expensive items like these, to buy from reputable (out of state) mail order houses - if possible. That way you avoid the California sales tax of 8.25% (mini GST?), and you can also check their support policies on the Web.
So in the end I got both items via mail order for a total of just over US$400. The motherboard from Motherboard express and the hard disk from Insight After getting back home, the real fun began!
Swapping the motherboard:
The motherboard came with a set of cables, some mounting hardware and two pages of (reasonable) instructions. I also got some magazine articles from the Web at PCWorld is fairly typical. Actually I anticipated more problems with the HDD size increase than the motherboard changeover (a well founded fear). But there was the possibility of mechanical fit problems as I had not taken full details of my case size with me to the USA. In the event the changeover was very smooth the majority of problems that were encountered were related to getting CMOS, DOS, and Windows 95 reconfigured - more on that later.
The first job was to note everything possible about the existing set-up. This info included copying all the CMOS info (specially the HDD parameters) all the various peripheral addresses, IRQs, DMA settings, etc. And labeling all the various cables as they were removed - turned out that most of the small cables that came with the computer case were already labeled (E.G. System reset, HDD activity LED, etc.). Then all the cables and cards were removed, the single screw holding the old board was removed, and the plastic stand-offs were slid out of the chassis/case, and the old board removed. Precautions against static build up were maintained throughout, and the old CPU and SIMMs were transferred to the new board - after the new board had been trial (mechanically) fitted. Actually I considered myself fairly lucky - mostly due to P.C. designer's having the foresight to standardise on the "baby-AT" form factor and mounting hole positioning. Once the keyboard connector was aligned with the hole for it in the chassis, the majority of the screw holes and plastic stand-offs lined up perfectly. Actually I ended up with no less than 4 screw hold downs (each with it's mandatory insulating washer) and 3 plastic stand-offs. A very solid installation that does not allow too much flexing when inserting cables or expansion cards. Due to SIMM sockets etc., one corner of the new card only had a single screw-down hole for about 1/4 of the card area. Unfortunately this hole was located over a large gap in the chassis - but a small Aluminum plate was made up to span the gap and secured to the chassis with large (ground short - to avoid shorts) self tapping screws. This plate provided a firm platform to mount the final metal screw-down standoff. I was rather pleased that I had the foresight to buy a $2 bag of miscellaneous computer screws, washers, and stand-offs at one of the swap meets - they certainly came in handy.
Once the board was securely fitted (and the CPU and SIMMs swapped), the cables were connected appropriately and a minimal card set fitted - actually only the video card was needed to get a minimal system up and running, all the other "essentials" are built into the new mother board.
The big "turn-on", any smoke?
Somewhat to my surprise, there was no smoke! But there were a number of problems, none of which (in retrospect!) were really big ones. The actual time from disassembly to power up was about 4 hours, but the problems (listed below) took a further 12 hours or so, and much gnashing of teeth and snapping at my long suffering "better half", to return the system to fully working order!
Problem 1: The POST check only returned 24Mb RAM instead of the previous 32Mb. My SIMMs are 2 * 8Mb (70ns), 4 * 4Mb of various types brought at various times. No amount of changing SIMM positions resulted in more than 24Mb. Eventually when checking some of the advanced CMOS settings I found that the SIMM speed settings could be set to either 60 or 70ns - but one of my 4Mb SIMMs is 80ns. So I was able to establish that this was the problem - the perils of getting the latest hi-tech stuff. 32Mb will now have to wait for a new 4Mb SIMM and I will still have two spare SIMM sockets.
Problem 2: CMOS would not recognise my old 850Mb HDD or the CDROM on the secondary EIDE channel. This was a real surprise as I had left both HDDs configured to the old cable and merely plugged the cable onto the motherboard connector. This one took much swapping of the old and new drives and jumper reconfiguration to find. I had documentation for all drives except the 850Mb Connor (old C: drive). Eventually I removed this drive and found there was some info printed on it's label. Apart from the normal Cyl, Hds, and Sect information there was info about the single/master or slave jumper, and some rather cryptic info about an "R/C" jumper - "IN=ATA CAM Master/slave" and "OUT=Conner Master/slave". The jumper was "IN" on the previous setup, but a change to "OUT" finally enabled CMOS to recognise it, and the CDROM drive! However W95 was still reporting that it needed DOS mode LBA drivers (no 32 bit access) so the decision was made to re-install W95 from the CDROM after adding the remaining peripheral cards (sound, scanner and modem). This was done using my emergency start disk which I had previously modified to include the necessary DOS CDROM drivers. In retrospect it was probably not a good plan to re-install W95 at that point (caused other problems!), but at least full HDD and CDROM drive capability was achieved once the re-installation was completed.
Problem 3: This was probably potentially the most serious one. There was smoke involved! I almost didn't include this one to avoid embarrassment - but eventually decided to tell it as a "cautionary tale". While trying to extract the Connor 3.5 inch HDD I found that the computer had been originally incorrectly assembled. The main power switch was mounted in such a way that the plastic front panel could not be easily removed and the switch needed to be re-oriented. After carefully noting the switch contacts and mains/power supply wiring colours the wires were detached, the switch orientation changed, and the wires carefully re-attached on the wrong contacts! I can only blame the lateness of the hour or something! Anyhow when the mains connector was plugged in again there was a large flash, some smoke, and a complete lack of lights and fan noise! Immediately expecting the worst, I removed the power supply unit (PSU) and took the cover off. The fuse was O.K but there was some suspiciously brown looking components on the PC board. Plugging in the PSU outside the computer appeared to confirm my fears - the 5V supply measured correctly but the 12V only measured 7.9 volts. I then panicked and "borrowed" a PSU from my son's 40Mhz 386 machine - but that only measured 6.9 volts instead of 12 volts - again the 5 volts seemed O.K. After a while, the penny dropped, the computer switched mode PSUs only produce the advertised voltages under load - something learned every day. But why no lights or fan noise? Eventually I established that my nice $60 Tandy "protected" mains distribution strip was not supplying 230 volts even though the red light was still glowing in the power switch! Re-setting the circuit breaker in the strip had no effect, and disassembly established that a track had blown off a small PC board between the surge limiting MOV devices and a large inductor. When the track was rejoined - all was well again. Including the original PSU despite the apparently burnt components. Phew! Not quite sure why they fitted a nice looking circuit breaker to the power strip.
Problem 4: Finally the first attempt to cure the Internet withdrawal symptoms. Unfortunately this was greeted with the W95 error message "the modem is being used by another program" - how does one establish what programs and processes are running at any time in W95? Probably I'm showing my ignorance but I could not find the offending program. Actually it turned out that W95 was speaking with forked tongue. The real problem was a conflict between the internal modem using Com2 and the on-board Com2 port. That was (eventually) fixed by changing the CMOS to identify the second on-board comm port as Com4.
Problem 5: Now the computer was causing the modem to dial and connect with TIP but then another error message " Unable to negotiate a compatible protocol" or something. Actually translated into real terms, this really means "I'm not running your script so your login id and password have not been sent" - there is no doubt a good reason why these error messages should be so far from the point - but I've yet to discover it. So eventually I was able to log on "manually", but despite slavishly following Mike Gellard's June & July 96 Sixteen Bits articles - I am still baffled by the failure of the script to run - possibly something to do with reloading W95.
Problem 6: After the W95 reload the system output another error message each boot time. The message was "The Netware compatible shell is not available". I still don't know how I got rid of this one! But I thought I had either not installed (or got rid of) anything to do with W95 networking, Mail, FAX, etc.
By now you may have gathered I am no fan of W95, but tolerate it in order to run "the latest and greatest"!
Installation of the new HDD.
Now the 450Mb slave HDD was removed and the 1.6Gb drive substituted. The CMOS was reconfigured (first automatically but later manually with "user" values to avoid the small delay each boot time for the software to identify the drive parameters). Does anyone know how to do this for the secondary channel CDROM drive? Note the new drive has to be first installed as a slave drive to allow partitioning and formatting. I originally decided to partition into 4 logical drives to keep the size below 528Mb and to minimise wasted space due to 16Kb cluster size for drives over 511Mb. But then we hit -
Problem 7: It seems hard to believe that the Fdisk software provided with the W95 DOS still does not immediately recognise disks larger than 528Mb (but more on this later). Despite the CMOS reporting 1.6Gb, Fdisk reported only the first 503Mb. So when the first 500Mb partition was created, only 3Mb was left for the remaining partitions! In the end I thought I had no alternative but to use the Western Digital provided EZDrive software to partition and format the new drive. I ended up with two 500Mb partition (8Kb cluster size), one 250Mb (4Kb clusters) and the remaining 374Mb (8Kb clusters) as the 4th partition.
But I did not like the EZDrive software at all. It added a substantial 10-20 seconds to the boot process, wrote stuff into the master boot record of my C: drive, required special recovery disks, and is generally "non-standard". It also did not seem to allow me to name the various new disk partitions. So eventually I removed it - or I thought I had, but despite using the EZDrive un-install some vestiges still remained in my C: drive master boot record (MBR).
Now the time had come to reverse the two drives (make the larger and faster new drive as the master, and the old C: drive the slave). I already knew that Bill Gates in his wisdom does not include some critical files if you copy W95 from one drive to another. The official explanation is that W95 should be re-installed on the new drive to allow proper configuration in the new environment (but there is also a measure of copy protection in this policy). Microsoft expects you to re-install the operating system and all your applications. Actually you can save a lot of time by copying everything to the new drive, and then installing W95 again "on top of" the old system. This saves re-installation of most applications, but leaves you with the same old mess of your old set-up. Mine is pretty messy, but for reasons explained later, I have decided to live with it for a while.
Problem 8: The drives were swapped and the CMOS reconfigured. But the system would not boot W95 giving various "Disk I/O errors". The reasons for this are too complicated to explain in less than 2-3 pages and this article is long enough already. Then began a series of swaps between the 850Mb and 1.6Gb drives in primary and secondary configurations - I got to know the jumpers, cable positions and CMOS procedures really well - It's amazing that I didn't loose all my data - but I believed I had a mirror image of my original C: drive on my new 1.6Gb drive as I had used "xcopy C:\*.* D: /s /e /h /c /k" to copy the data. But the wretched system was still try ing to use EZDrive software now on both HDDs MBRs! By now I presume the hardware experts will be really laughing! I was able to finally get back to the original C:=850Mb (doubled) and D:=1.6Gb eventually - but to finally get rid of EZDrive I had to "Fdisk /MBR" one piece of invaluable information found in Holland. This is a superb FAQ that tells all about EIDE and ATAPI interfaces and their interaction with BIOS, etc.
Problem 9: After getting rid of EZDrive - how to get Fdisk to recognise all 1.6Gb of the new drive? The answer was found in the FAQ mentioned above. I had used the "NORMAL" drive type CMOS setting for the new drive, assuming that if CMOS recognised it as 1.6Gb, everything else would - wrong! The FAQ suggested that either "LARGE" or "LBA" should be used. "LARGE" got Fdisk to declare that the drive was just over 1Gb only (at least more than 503mb! But CMOS was still declaring the drive to be 1.6Gb). "LBA" finally allowed Fdisk to "see" the full disk capacity and it was set up as one large partition. After formatting, the contents of C: drive was again copied, and attempt was made to use Drivespace-3 with no compression to reduce cluster size. However drivespace states "the disk is larger than 800Mb" so it will "compress 890.93Mb so it contains 1.0Gb of free space, and create another host drive F: with 656.25Mb of uncompressed free space" sigh......
I guess this is problem 10 but I gave up at this point! I don't think I will try to reconfigure further, but will wait for the new (Currently OEM) W95 "FAT32" drivers, to eventually let the system recognise large drives. But I'm open to suggestions of course!
CONCLUSIONS:
I might not have attempted some of this without the Internet information - in particular the Western Digital hard drive installation book did not mention the various CMOS settings needed to get DOS and W95 to recognise their drives - but the Net eventually provided the explanations.
So for a moderate financial outlay I now have a system capable of considerable extra expansion. But there was an investment of some 20 hours in fault finding time. In return I learned a great deal and would feel more confident trying this again - not too soon!
The main lesson learned is that hardware configuration is relatively easy these days, but software re-configuration (particularly W95) is nowhere near as simple!