|
|
|
|
|
|||||
|
Systems |
Home | About | Contact | |||||
|
|
|
|
|
|
|
|
||
| Overview | Specs | Bench | Gallery | Record | YouTube | DXZeff | ||
| 640x480 Recommended - or zoom-in. |
Website Copyright © Paul Monteray |
|||||||
|
|
|
|
Overview Tawnia as of September 2022 There's no glamour here: Tawnia is a big white box with no remarkable qualities on the surface. Performance seems lacking and the 80286 running at single figure clock speeds crawls compared to those later 12 and 16MHz models, as well as running at higher temperatures. Indeed, the whole machine is full of NMOS parts, which use significantly more power than the later CMOS technology and, as such, dissipate far more heat. This is because Tawnia would have been an early attempt to clone a relatively high performance system. At that time, CMOS was less mature, prohibitively expensive and thought less resilient due to much smaller margins for error. The motherboard has had many parts replaced over its lifetime, but it seems that it has its origins in 1985, probably less than one year after the IBM 5170 was introduced. The Intel 386 would be launched later that year, but would not be used in a PC until September of 1986, meaning that for a time, our 286 here would have been the pinnacle of performance. Something that is often overlooked is the sheer speed that the 286 was capable of over the original 8088. At the same clock speed, the 80286 will perform more than twice as fast as the 8088 in many applications. It also had a 24-Bit address bus, so could theoretically support upto 16MiB of RAM, though this would not be very useful until later in the CPU's lifetime, if only due to the cost of memory. Our board tops out at 1MiB without using expansion cards and actually incurs a limit that was in place on the original 5170 - while you can configure the board for 640KiB base memory, it would be split into banks of 256K+256K+64K+64K. If you populate the last two banks with another 256K each and configure the dipswitches, you'll get only 512KiB of base memory, with the remaining 512K mapped as XMS. The early IBM AT machines supported only 512KiB of RAM in the motherboard. Today this sounds like quite the limitation, as we're used to seeing even XT class machines show up with the full 640KiB, but at the time, many if not the majority of machines wouldn't have been fully populated and you could buy a PC with as little as 64KiB of RAM. One thing our clone does allow you to do is disable wait states, but you're pushing your luck doing this with the slower NMOS parts. It will make the machine faster, but some components won't really like it very much and instability creeps in. Adding to this is the sheer fact that the motherboard isn't very well made, as it was almost certainly designed to cut costs wherever possible. There's a fairly good argument to be made that they simply took their existing XT clone board, shoved a microcontroller on it, added a second 8237 and 8259, then called it a day after sloppily replacing glue logic with the C&T 8220 chipset. That chipset is itself interesting as it was the first 'real' LSI chipset, but it doesn't do much yet. IRQ, DMA and most other functions are still discrete parts on the motherboard and the chipset serves mainly to replace the large number of 74 series logic chips. It is made up of four PLCCs using NMOS, though later versions of the 82100 were switched to CMOS and its successor, the 82200 would be entirely CMOS and move even more board logic into just four LSI ICs. All in all, the 286 would have a long life. It performed well and with most software still targeting the 8088 for several more years, it would be a viable platform leading into the turn of the decade. Our machine marks a time when clone makers were starting to branch off on their own more and more, attempting to produce ever faster machines and still trying to nail down the compatibility - Tawnia's BIOS was updated at some point, but still isn't quite 100% IBM Compatible - and for the end user, this could only be a good thing, in the end. With the vastly increased performance, 16-Bit expansion options, tighter integration and later, an ever lowering price, the 286 really was a winner whose merits people often overlook today. Also, from what I hear, it was and still is used in aircraft. |
|
| Insert Disk 2... |
Smashed Windows |
|
|
|||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||
| Bad Command of File Name |
Hosted by AOL Hometown |
||||||||||||||||||||||||||||||||||||||
|
|
|
|
|
|
|
System Performance There really isn't much to say here. The results say it all and the performance gap is really self evident. The 286 would eventually scale to upwards of 20MHz and at that speed, could often even marginally outstrip the later 386. Tawnia even incurs a marginal hit from having the 5MHz 287 in the board, but even still, the poor old 8088 has no chance of catching us. |
|
| Cannot divide by Zero |
MSCDEX not loaded |
|
|
|
|
|
|
|
Pictures |
|
![]() |
Tawnia's cover removed. Our motherboard is small enough to be considered 'Full Length Baby AT', thanks to the Chips & Technologies 82100 chipset. Still, being a mid-80s machine, it's quite busy inside the chassis. Worthy of note is how the steel of this chassis is thinner than that of the XT clone. |
|
360K Floppy Drives were the standard of the time and you'd be hard pressed to find a machine that didn't have one. Of course they exist, but most software still came on floppies, so you'd need at least one drive in your office in order to load it. |
![]() |
|
A second shot of the floppy drive, purely because I like the casting and machining. Whilst I generally dislike floppy disks, I can at least appreciate the class and sophistication of the 5.25" versions, versus the peasant 3.5" drives with their ugly stamped and folded sheet metal. I swear 5.25s are faster, too. |
|
|
The Seagate ST-251 has a few variants, but was probably the most common hard drive at one time. This would have been an expensive luxury, though at 21MiB, the capacity would be more than workable in a world that was still largely dependent on floppies. It doesn't hurt to have more capacity, but 21M was 'enough'. |
![]() |
![]() |
That hard drive requires a host adapter. This provides the MFM interface for the hard drive, which isn't immensely dissimilar to the floppy interface, which it also provides. ESDI had arrived by now and SCSI was on the horizon, but the original MFM interface would still be far more widespread for some time. |
|
Serial, Parallel and a Game Port are provided by this I/O card, with options to add a second Serial port later. This card adheres to the old LPT numbering standard where 'LPT3' would be 378h. The assumption is that LPT1 is on the video card at 3BCh - watch out for this in old machines and software. |
|
|
Everex used C&T's first VGA clone from 1987, though the parts on this card are largely unchanged from their EGA clone. This is pure VGA with slight incompatibilities and a mere 256K of DRAM. There are no SVGA modes or anything fancy going on here. Supposedly it can emulate a Hercules on the TTL port at the same time as doing VGA. |
|
|
AdLib cards were swiftly cloned and this is one of the earlier examples. Our card was made later, as evidenced by the hastily added '386' and '486' markings. Even the FM ICs are clones. This card has a nice added feature, in that you can input PC Speaker signals and also output the card back to the case speaker. |
|
|
Twisted Pair would have been very costly in this time, but this 3Com EtherLink II TP has it (denoted by the TP). It's a 10BaseT Half-Duplex card, so not speedy, but it's faster than the hard drive, which it could otherwise fill in under a minute. Like everything else in this machine, this thing uses a ton of NMOS and barfs heat. |
|
|
A 1990s modem with an unusual plastic slot cover. It is designed such that once your RJ-11 cable is inserted, it vanishes so far down the hole that you need tools to get it back. It also has a reset button on it (a sign of reliability, surely) and can achieve 28.8K maximum speeds. Still, it has dialled a fellow enthusiast in Canada. |
|
|
Motherboards at this time still varied quite a lot. Ours is decidedly cheaper looking than many and isn't all that well made. It has needed many repairs in its lifetime, but that's the price you pay for buying a cheap, early clone. It is also stretching, so the lower mounting holes no longer line up with the chassis. |
|
|
Further signs of quality appear in the form of bodge wires inside IC sockets. Also note the old style 'lumpy' solder mask. Overall, they don't quite seem to have gotten the hang of cloning AT class machines just yet, but it's mostly there, albeit a little ugly. All that said, I wouldn't have it any other way and love this thing. |
|
|
Tawnia's old appearance. This case always seemed too modern. We shall inevitably see this chassis again some day, as there's nothing wrong with it and certainly no reason it couldn't house something else, say a 486. The larger hard drive developed faults long ago and will be used in my shed, in another machine, until it fails completely. |
|
|
You can click the above pictures to see full size versions in a new window |
|
| Stack Overflow |
Loading World Domination... |
|
|
|
|
System Achievements Tawnia holds a few records:
|
|
| Keyboard error, press F1 to continue |
Ooh-laa!!! |
|
|
|
|
Videos Available
Tawnia appeared in a video with Lisa, though both machines were not in their current configuration yet. It is still pretty amazing just how rapidly things changed in only a few short years. |
|
| Requires QuickTime for Windows 95... |
Video for Windows v1.1 |