Up until necessity drove me to transplant a Windows system, that is to take a functional boot drive from a failed computer and place ithem into a newer case, I hadn't considered it a viable option. Necessity is the
mother of invention, however, and I once had the opportunity to see if this counter-intuitive approach to upgrading would actually work. Given the right constraints, it does - my newer healthier system proves it - and
this article is intended to give the curious reader a step-by-step overview as to the process of making its success more likely.
The Problem
First of all, this particular article does not concern itself with Windows95 specifically - it is primarily about NT4, the precursor to XP and much less amenable to "plug and play" than any of the 9X variants. Because this
particular operation was successful, however, a similar operation with Windows95 can likely be as easy and rewarding as this one was.
The specifics of the old hardware were as follows:
AT case with 250 watt power supply.
FIC 503 Plus motherboard, the last (AFAIK) motherboard to work with an AT case and power supply.
Award BIOS, vintage 2001, hard-wired into the motherboard.
AMD K6 processor running at its rated clock speed of 500 MHz.
256Megs of RAM.
Revolution #9 video card w/8 Megs RAM onboard.
two hard-drives (both formatted NTFS) and a CD-ROM.
For nearly four years, this hardware performed fine with no issues. One morning, on startup, the BIOS emitted an audible error code in the form of five beeps - one long, two short, then a two-tone "high-low" like a fog horn.
Touching the "reset" button, allowing the system to start itself over from scratch, resulted in a trouble-free normal boot-up. Looking up this error code, it seemed to indicate a bad video card though there were a few other
things it could be (including the motherboard itself). I made note of this, acquiring a newer video card for later replacement, and continued using the machine. It worked OK for about eight or nine weeks, occasionally giving me
the strange beep sequence but always starting fine after hitting the reset button.
One morning, thankfully with no pressing deadlines, I started it like normal and no amount of resetting would allow the machine to boot. It sat there with a dead monitor screen making the same beep sequence mentioned above. I
was ready - I took the machine offline and put it on the bench with a test monitor to do some troubleshooting.
Removed all cards from the machine except for the video card. Reseated the original video card, turned it on - same beeps, no video.
Swapped video cards, replacing old suspected one with newer identical one. Same beep code, no video.
Placed my trusty "post board" into last ISA slot, powered up - all four critical voltage LEDs lit up fine, code of "0d" showed when beeps sounded. Manual for that board showed that the "0d" signified the video test
as well as a few other tests which concerned the motherboard.
Tried two other video cards, one AGP and one PCI (one after the other), to see if results changed - same beeps, no video.
For kicks, RAM board is removed and unit is powered up again. Beep code is different, implying that the self-checks are functioning correctly up to the point of "0d" at least.
RAM board is loaded into another computer and checked - it works fine.
Original suspect video card loaded into another computer and checked - it works fine. Headscratching begins in earnest.
BIOS battery is removed for an hour to change all BIOS settings to defaults. No change in behavior - startup with RAM board and new video card in place results in same beeps & no video.
To summarize:
difficulty cannot be the video card, though all error codes seem to indicate it.
None of the other peripheral cards in the system could be the problem, as they were removed before the tests started.
RAM is not the problem.
BIOS settings corruption is not the problem.
LATER ADDENDUM: swapping out the processor also ruled that component out: it performed fine in another board. The remaining possible culprit, the motherboard, has been recycled and no further testing will be done on it.
The Solution
I had built a newer machine some time earlier, a faster machine with more RAM, which was currently sitting in storage waiting to be placed into service. After doing some homework on the issue, and
finding that both the old and new motherboards had been built around the same VIA chipset (a blind miracle), I took the hard-drives out of the old case and put them into the new one. For now, I left the other peripherals out of the
system - the main purpose here was to see if the OS would boot with the old hard-drives attached to the newer motherboard.
Not really knowing what to expect, as I had never done this before, I attached the monitor to the new system and turned it on. I watched in amazement as NT4 recognized the chipset on the new
motherboard and loaded the drivers successfully. I got to the desktop cleanly first time through. Wow!
The peripherals were next, and this was the only difficulty encountered with this setup:
On-board sound was being used on the newer motherboard, which meant that the trusty Soundblaster ISA card (and its accompanying driver) was retired and removed from the system.
The SCSI card, an older ISA-based Adaptec 1510 unit, had to be re-installed and be allowed its own IRQ in the BIOS settings in order to work. (Later, this was replaced with an UltraSCSI LSI card so IRQs were no longer an issue.)
The D-Link network card needed to be reinstalled since its IRQ and addresses had changed from the older system.
That recombinant machine, as of this writing, has been running for two years with no issues.
Conclusion
Conventional wisdom suggests that this should not have worked. NT4 is very unforgiving, particularly where its motherboard drivers are concerned, and if any transplant would stand a chance of failing it would be this one. Because this worked so
well for me, I feel I can safely say that the other versions of Windows will have no more difficulty in similar circumstances. When I say "similar", here's specifically what I mean: The Chipset Must Be The Same - how much the same
I couldn't say, but if the installed drivers will work on the newer hardware you're going to have an easy time of it. Otherwise, results are unpredictable.
If you'd like to migrate your existing Windows95 installation to a newer box someday, something that is possible, you'd be well advised to plan the new box and build it while your current system is running fine. This will include the following
steps:
Keep to the same chipset if you can. If you can't, then keep to the same brand and make sure you have a floppy disk or CD-ROM which has chipset drivers for Windows95.
If you want to keep your peripherals, then make sure you still have the drivers disks for those units you wish to keep.
Odds are that your newer chipset will be running much faster than your old one - please read my page about updating Windows95 to run at faster speeds before attempting such an upgrade. Without this, you could
wind up with a paper-weight instead of a computer.
If your newer motherboard has an on-board AGP video card, and your transplanted operating system is Windows95A, then you cannot use the AGP card - only the last version of Windows95 (95C) comes with the proper support for AGP video,
and 95B can be upgraded toward it by applying the AMDK6UPD.EXE update as mentioned above. You can get around this whole issue by disabling the AGP video in the BIOS and using a 95-compatible PCI video card instead.
I will be adding info to this page as I learn more, but this is the nugget of the whole presentation: if you have a reliable running Windows installation and your hardware packs up, then you've got another option besides a clean install on a newer
box - Transplant!
