Salvage: Techway Endeavour II Computer (Part 1)

Oh wow. A large package, wrapped in “fragile” tape with a load of stamps and a mention of “heavy parcel”. I wonder what that might be? A Christmas present? Indeed.

The Story of the Parcel

I went in to visit some colleagues at the university and was lucky enough to have had a computer “saved” for me. It was an old beige desktop which didn’t seem to have been touched for the better part of at least ten years. I was free to take it away, or it would be scrapped, so I had a thought about it. Do I need yet another vintage computer in my collection?

But then, after taking a good look around and inside, I decided that I had to have it. This unit was too good to let go to the scrap heap. It was about 13kg and would be a handful to lug home with me on public transport.

That’s when I stumbled on the bright idea of wrapping it up in cardboard and packing tape and mailing it to myself. It wasn’t the best way of packaging it, as parcels often do get beat-up in transit, but I didn’t care quite enough to spend an extra half hour begging people for some bubble wrap. In the end, it cost me AU$11.45 to post it to myself, which was about the same price as travelling return via public transport.

Thus, the parcel above. A gift from a friend, sent by myself, to myself, delivered by Australia Post.

A Tour of the Unit

The unit is a beige desktop box with the Techway and Endeavour II badges on the front. Being a case of the era, the case has a keyboard lock switch and a turbo button/LED. A hardware power button is provided (as is normal for AT systems) and a reset switch. There is a single 3.5″ floppy drive and a slot-loading CD-ROM, a rare later addition.

Looking at the rear, the unit proudly boasts it was “Assembled in Australia from local and imported components”. It has one of the smaller AT power supplies, along with the standard pair of serial, single parallel and keyboard DIN ports. VGA video is provided from an expansion card and a sound card is also installed. One of the blanking plates does not seem to match the colour of the system, suggesting that one of the original expansion cards was probably removed at one stage.

Already, the system is interesting purely because of the Techway branding. Techway was a local PC manufacturer which was forced to exit the market in 1996 due to tough conditions resulting in heavy losses due to reducing hardware margins. They continued on as “Tangent” and “Network Services” divisions according to the article above, but a check of the company’s status seems to show that they only made it to 1997 when they were taken over by Intellect Holdings Limited. This latter company went into receivership in 2008 and almost all traces of Techway have since disappeared.

Techway does, however, deserve more recognition as it seems to have been started back in 1979 by A/Prof Peter Jones, a pioneer of computing networks in Australia with a number of associations with universities in Australia, passing away in 2016.

It is definitely a glimpse into our past and if I had realised this, I might have packaged it a little more carefully as it didn’t make it entirely unscathed.

Removing the lid reveals the labels inside the case, common with pre-built machines that list key specifications and configuration details. We can see that the system was built on 23rd May 1995, serial number 2000249 (249 for the year?). It claims to have been built with a Pentium 90Mhz CPU, 24MB of RAM, 1MB Hercules graphics card (MDA/CGA), with a single 1.44MB floppy disk and a 528MB hard drive. Oddly enough, the parallel port base address is listed as 3BC-3BF – normally on single parallel port machines, I’m used to seeing 378h as the base address.

A peek inside the chassis shows that the present configuration doesn’t (necessarily) match the configuration listed on the label, so lets take an inventory of the system while cleaning it out and performing a thorough inspection.

Power Supply

The power supply is a Seventeam ST-230WHF made for 240V. It’s a name that I recognise and one of the few brands that still persist through to today. The supply is a switchmode supply which claims to provide 16A on 5V, 12A on 12V, 0.3A on -5V and 0.5A on -12V. The supply is cooled by an 80mm Sunon fan.

Before even risking a power up, I decided to take the cover off and blow out the oodles of dust inside. On the whole, the heatsinks are small compared to modern power supplies and it almost looks like a generic ATX supply, but that’s because power requirements back then were not as high.

A look at the capacitors sees no bulges. It’s also heartening to see that they have used high quality Rubycon capacitors, although they would still be well past their design lifetime.

I didn’t take as close of a look at the secondary capacitors, which seem to have 94 quite prominently printed on them. It’s good to see that 105 degree C components were already standard at this time. Another feature rarely seen after the AT era is the presence of a pass-through IEC connector allowing a monitor to be plugged into the back of the computer and have the power switched simultaneously.

The cabling in the case was “organised” using plastic clips which used double-sided foam tape to adhere to the chassis. The tape has since started to fail, so I don’t see the point in reinstalling the clips and have instead removed them.

Mainboard

The mainboard is an ASUS PCI/I-P54SP4 Rev 1.5. The board itself has a warped CPU socket lever for some reason, but other than that, looks fairly healthy with the exception of using the dreaded Dallas DS12887 potted RTC module. I wonder if the board will still run with a dead RTC SRAM as some will not.

The board is populated with 512kB of cache memory and uses an SiS chipset along with an Award BIOS. Unfortunately, no Intel chipset here, so I wonder if the performance is any worse. The board uses the CMD0640 controller (apparently buggy) to provide dual channel IDE at up to 13.3MB/s rates. The board also has an SMC controller providing onboard serial and parallel I/O capabilities – by the Pentium era, such integration was becoming more normal, freeing up expansion slots.

The board runs both 16-bit ISA and 32-bit PCI buses, being in a transition period. RAM is provided using 72-pin SIMMs as is standard, installed in pairs, although depending on your modules you may have to move some resistor packs to configure double/single sided.

Despite the board having the capacity to provide two IDE channels, it was surprising to see that the CD-ROM and IDE hard drive were installed sharing the same channel and thus, limiting their maximum throughput as they contended for throughput along the same cable. The secondary port remained unused – the reason for this might become clear as we continue our examination.

CPU

A quick glance at the CPU says it’s an Intel Pentium, so you might be thinking that this is the one that it shipped with.

That might be a good guess, but this system has a surprise up its sleeve. To find out the actual identity, the fan must be removed to reveal its S-Spec code.

This CPU is an SU032 which is actually a Pentium 100Mhz CPU. I wonder if that means the computer had been upgraded or whether they just “ran out” of a given SKU and gave the owner a free upgrade. Whatever the case may be, the first thing I’m doing is getting rid of all the dust!

RAM

Taking a look at the SIMMs reveals the second reason why I decided to salvage the machine. See the green label with the TP logo on it? That stands for Total Peripherals. Who are Total Peripherals? Why, they were the predecessors of TPG Telecom, known as Total Peripherals Group. In fact, here’s an even older example I found online. Unfortunately, it doesn’t seem they had their website up when they were dealing with such old systems, but here’s a page from 2000 where they’re still selling Pentium II and Celeron machines.

This particular stick has been hand-dated as Week 24 of 96, using NPNX chips dated Week 19 of 1996. The TPG warranty seal has 9606199489 implying the same.

Two of the above sticks were in the machine, each of them with 16 NPNX NN514405J-60 chips on board. The NPNX brand is rather unfamiliar to me, from what I can find it is Nippon Steel Semiconductor Corporation. Each module is 8MiB 60ns.

The remaining two modules are NEC Singapore branded, marked 9516YE1656, suggesting they may be 1995 modules. Each of them has 16 NEC 424400-70 chips on them dated Week 12 of 1995.

Each of these modules are also 8MiB 70ns for a total memory of 32MiB in the machine. I suspect that these NEC modules may have been originally included in the machine with a pair of 4MiB modules to make for 24MiB as listed on the case. The latter two may have been swapped out as the machine was upgraded by Total Peripherals Group.

It is also rather surprising to see that this machine has a mixture of gold and tin-edged modules. In the era, the advice was always to stick with the one that matches the slot or the original modules and never to mix, on the claim that dissimilar metal connections can cause bad contacts. I’m not certain how true this is, especially as gold is relatively inert, so a gold-to-tin connection shouldn’t be a problem.

Hard Drive

The hard drive was provided in the system, which would make for good fun in examining. The drive itself was very loosely mounted in adapter brackets to fit a 5.25″ bay owing to a lack of dedicated 3.5″ bays in the case. However, whoever installed it did a very amateur job and didn’t screw the adapter brackets to the case on one side at all – probably out of laziness as they might have avoided taking off the floppy drive bay to do it.

The drive itself is a Western Digital Caviar 22000 (AC22000-00LA). This is a 2000.3MB drive with a translated geometry of 3876/16/63. The drive spins at an odd 5200RPM, providing about 10MB/s transfer rate and having a 256kB cache. Seek times are claimed to be 11ms. This drive was manufactured 6th November 1997, making it over 20 years old, but also probably the second-last component to be upgraded. I wonder if this drive was installed fresh or as a “hand-me-down” pull from another machine. It’s definitely bigger than the 528MB drive it (apparently) was shipped with.

The warranty seal tape extends along the complete circumference of the drive. The standard mounting holes are provided.

The rear of the drive has the power input, IDE interface and configuration jumper block.

The front of the drive is where the sealing tape overlaps itself.

As such old hard drives are often quite “fragile” and tend to be almost certainly problematic, I wasn’t going to just try to start the machine with the drive attached. It calls for some special handling.

Floppy Drive

The floppy drive was a Sony MPF520-1, dated Week 3 of 1995. The drive is Made in Malaysia.

I’ve generally been pretty fond of Sony drives as they tend to work reliably and cause fewer errors.

However, unlike other drives, this one came with a bit of a surprise – there is a mylar and foil shield on the underside of the drive. I’ve not seen this even with older Sony drives from 80386-class machines.

From the side, the drive has the standard mounting holes and is encased in sheet metal.

The front of the drive is traditional Sony design – the shape of the indicator light and size/shape of the eject button easily give it away.

On the rear, just the ordinary interfaces. The drive was quite dusty, so it was good to give it a clean before reinstalling it. The drive cage design of the case requires the floppy drive to be removed before any other drives can be accessed.

CD-ROM Drive

The CD-ROM drive was the most interesting by far. At that time, caddy drives would have just gone out of style with the majority of drives being tray load drives. But this one was different – it was a slot loader which was something more commonly reserved for automotive use. I thought computer-based slot-loaders only began to exist in the DVD days, myself having been given a Pioneer DVD-120S drive in the past.

The drive makes use of an IDE/ATAPI interface, as expected, but surprisingly the drive jumpers uses a “rubber” jumper that is basically a silicone rubber elastomer with conductive rubber gripping the pins. I wonder how reliable this would be in the long run – first time seeing it though.

No prizes for guessing – it’s also a Pioneer drive, a DR-504S 32X rated drive. The drive was manufactured May 1998 in Japan, making it the most recent component in the whole computer and probably the last upgrade to be performed.

As expected, the drive is a full-length drive … they hadn’t quite mastered the art of making short 5.25″ drives yet. From the specification sheet, it claims to read from 14X-32X using a CAV/CLV combination mode with an average random seek of <60ms and access of <70ms.

There was a piece of loose rattling plastic inside which I shook out. I’d be interested to see if this drive still works – and if it does, how fast (or slow) it is.

Expansion Cards

The computer had just two expansion cards. The first was the VGA graphics adapter, a rather “vanilla” S3 Trio64V+ with upgraded EDO Memory totalling 2MiB. Based on the FCC ID of ICUVGA-GW503, this card originates from Gainward, a manufacturer that still exists to this day.

Like the RAM we saw earlier, this card has a label from TPG on it, with an ID number of 9606114426 suggesting it may have been upgraded in 1996. Consequently the blanking plate that didn’t match the rear may have been where the original CGA/MDA Hercules ISA card had been removed.

The second expansion card is a Creative SoundBlaster AWE32 Value IDE card (CT3910).

This card is a fairly common card, not as cheap and nasty as the Vibra 16 that came after but also not as flashy as the AWE64 that came after it. It offers both wavetable synthesis and Yamaha OPL synthesis for MIDI. This one also offered an additional IDE controller on board, and was often bundled with CD-ROM drives to add them to 486-class machines. As the IDE controller was not disabled on the board (possibly because it could not be owing to the lack of jumper), it may have caused a conflict with the secondary IDE controller on the motherboard by overlapping I/O addresses by default. As a result, a computer novice may have installed the CD-ROM on the primary IDE bus as they couldn’t work out why the secondary didn’t work. Or maybe they just didn’t have a second cable for it.

Surprisingly, given all the work done by TPG on the machine, it was surprising not to find an internal modem shoved into the case.

Conclusion

The Techway Endeavour II PC is an old vintage computer with a story to tell. Built by formerly successful Australian company Techway and seemingly upgraded by Total Perhipherals Group, now TPG Telecom, it paints a story of how the money in computer hardware dried up, leaving companies to shutter their doors or move onto other service-based industries.

Built using relatively “standard” off-the-shelf components, the computer seems to have undergone a number of life-extending upgrades including a CPU upgrade, addition of a CD-ROM drive, a RAM upgrade and replacement of the hard drive and graphics controller. Originally built on 23rd May 1995, it is now 23 years old.

But will it run? What’s on the hard drive? What was it used for? Will we ever know? How do we get it going? Find out when the next part is posted … hopefully soon!

About lui_gough

I'm a bit of a nut for electronics, computing, photography, radio, satellite and other technical hobbies. Click for more about me!
This entry was posted in Computing and tagged , , , . Bookmark the permalink.

11 Responses to Salvage: Techway Endeavour II Computer (Part 1)

  1. rasz_pl says:

    >a computer novice may have installed the CD-ROM on the primary IDE bus

    According to CMD-640 bug description (fascinating read) this controller was incapable of simultaneous primary/secondary IO anyway.

    Looks like a perfect computer for year 1996, just in time for Quake.

    • lui_gough says:

      Indeed, it seems the design of the chip was very much problematic, but they were actually quite widely used amongst lower end designs. Probably owing to the nature of OS workarounds, I wasn’t even aware of the bugginess until researching it recently, only to find that I had probably lived through that entire generation of computer with sub-par storage subsystem performance because of it.

      I suppose their legacy does live on through Silicon Image – although they too occasionally get a bad rap …

      – Gough

  2. sparcie says:

    Given the age of the machine I’d say the original graphics card is unlikely to have been a CGA/MDA card, as those standards had long been replaced by that stage. Hercules as a company actually lasted quite a while, well into the 3D accelerator era. They didn’t make any chipsets themselves, instead usually licensing or buying chips from other companies for their cards.

    Hercules are really best known for their original monochrome cards, which are a standard unto themselves, but at the point this was made VGA would have been a minimum with SVGA and VESA cards beginning to become more common place.

    • lui_gough says:

      Indeed, that would be my expectation. But you never know – it was at a university, so I have sometimes seen very strange configurations as the computers may have been used for industrial control or maybe a very specific application which might have required it.

      Even my 486 had a Trident VLB SVGA card and my first Pentium had an integrated ATi Rage of some description. But that leaves a curious wonder as to why it would be filled in on the configuration label as Hercules? The card is a Gainward, it’s based on an S3 chipset … so maybe Hercules “compatible”? The other thought I was thinking was maybe it was the result of a “trade-in” as in the 90s, it was still somewhat profitable to refurbish trade in parts and sell them on second-hand. A 528MB hard drive in a Pentium-class machine wasn’t particularly big either as I did have a 480MB and a 528MB on my 486DX2/66 of the time.

      I’m still working on it but another posting should come over the next couple of weeks where (hopefully) I find out what is on it/what it was used for/who was using it …

      – Gough

      • sparcie says:

        I’d guess it was a Hercules branded VGA card of some type rather than a MDA compatible card. What tipped me off is the 1MB of video memory on the label, original Hercules cards only carried 64k for the mono display. 1MB was very common for many VGA and basic SVGA/VESA cards. Perhaps they upgraded to the S3 card because of its basic acceleration features?

        Although I have heard of MDA compatible Hercules cards being used quite late as they offered a way to connect a second screen before that became common place. There wasn’t really a way to have two VGA cards in old systems because of the fixed address memory mapped frame buffer, but MDA and Hercules used a different address space meaning they could be used alongside EGA, VGA and some later cards. From what I understand this was often used in debugging, using the second display to show debugging information. It could have been used in this machine for a similar purpose.

        • rasz_pl says:

          Here is a list of non HGC Hercules branded graphic cards sold before 1999: https://pclab.pl/art6767-2.html (article from 2003).
          TIGA compatible cards ran you >$1K, 8514 ones ~$400.

          First on the list, Hercules Chrome started at 1295 British pounds in 1992! with top model listed at 1495£.

          The only low end 1MB cards on there are Hercules Stingray series, based on one of _the fastest_ DOS VGA cores made by ARK. Even in Windows 95 S3 Trio64V+ is in the same league according to PCMag tests from December 1995.

          The only reason I can think of for “upgrading” is video memory size to support 1024x768x16bit mode. Either original Hercules Stingray came with no sockets for ram expansion, or technician doing the upgrade saved the card for himself.

          • lui_gough says:

            Thanks for the research – indeed, the TPG label on the back of the S3 is the proof that it didn’t (likely) originate from the original builder (Techway). Short of a failure, my suspicion was a greedy tech :). Those prices are eyewatering … but then I guess something we don’t remember (so readily or fondly, or may have entirely forgotten) was just how expensive Intel CPUs were and computer parts in general in that time-frame. Getting 1024×768 in 16-bit would have been nice – I myself struggled with 256 colour from memory, as the resolution was more useful than the colour depth and most applications were happy enough with 256 colour as well. Otherwise it was 800×600 …

            As for the mystery of what this machine was used for … I’m still making progress towards a full unveil :).

            – Gough

          • Krzyś says:

            > Either original Hercules Stingray came with no
            > sockets for ram expansion, or technician doing the
            > upgrade saved the card for himself.

            Indeed – there were such cards with 1 MB only and no memory expansion sockets like this one:

            http://www.a1usedcomputers.com.au/shop/prodView.asp?idproduct=32

            Also VGA connector on that one seems to line in good with that “Video” label”.

            Maybe this particular card from the link was the one that was saved by TPG’s technician 😉 (just joking)

        • Krzyś says:

          Also, Hercules Graphics Card used ISA slot. I don’t know of any HGC-compatible graphics adapter with PCI connector. HGC/MDA cards and monitors were extinct by the time that PCI stardard got popular.

          Also, please don’t confuse MDA/HGC with CGA/EGA. Frequencies were different, synchronization polarity was different, resolution was different. There was also no compatibility between HGC and CGA graphics modes (however there was a TSR to emulate CGA on HGC).

          Therefore I agree with other comments there, that this was just VGA card made by Hercules. I would say that this S3 card could have been the original one (without memory expansion fitted), if not for that datecode (Also I don’t know from which year is this S3 chipset). For me it does not seem to be a big upgrade from stock configuration (would this double amount of RAM enabled someone to use more colors at some big resolution on 14″-15″ CRTs of that era? Or as others suggested – was this a thing with 3D acceleration?), but back then certainly it was bigger than my impression right now (my first PC was late ’99 [email protected] AGP 2x S3 Savage3D w/8 MB kind of thing).

          I don’t know of any HGC-compatible PCI VGA card (I mean hardware compatibility) either, though this does not necessarily mean that there was no such thing.

          (The other way around is represented by, for example DIAMOND FLOWER’s VG-7000, which had also TTL DB9 output and could be configured to wide variety of monitors, including CGA, EGA, MDA/HGC. For some details see end of my post here: https://www.vogons.org/viewtopic.php?p=548894#p548894 )

          ps. dual PCI VGA configuration was AFAIK easily possible as it is what Windows 98 supported (among other combinations). There was just no software support prior to Windows 98. It wasn’t possible with VGA/EGA/CGA ISA cards though, as these were hardwired to certain addresses (the same location of framebuffer, the same location of extension bios in case of VGA/EGA, whereas on PCI bus extension BIOSes are mapped in unique memory range and then copied to UMA region before execution). As sparcie said, mixing ISA VGA/EGA/CGA with MDA/HGC was possible and it was a common configuration for debuggers, but also it was said to be used by some CAD programs, like AutoCAD which displayed its menu on monochrome monitor (that’s what I remember from contemporary computer press).

          ps2. while at it – have a Happy New Year! I wish you as much joy and inspiration as I have every time that I’m reading your blog 🙂

          • lui_gough says:

            Thanks for your reply – always good to hear from technically like-minded people. Indeed, the label seems to give away that the card was not the original installed card at the time of supply of the machine from Techway, but it seems that the RAM upgrade and PCI video card upgrade both were done by TPG in ~1996.

            I guess it was just wishful thinking on my behalf to come across a “real” genuine Hercules card in a salvage nowadays – it seems highly unlikely as 80286/80386 era machines have mostly been junked or melted for scrap. Even if I had one, I wouldn’t have a monitor to use it with … but I guess the only reason I am fond of them was that my first computer (an 80386SX-16 with 4MB RAM and a 40MB Miniscribe MFM HDD) used a Hercules card in MDA mode with a Samsung green-on-black CRT. Too bad I was too young at the time to save it from the scrap heap.

            – Gough

  3. Daniele says:

    Seems my first P133 computer, with the same S3 Trio64V+ that I upgraded with a DRAM chips from a dead S3 Virge. Of course I had an ISA audio card (if I remember well an ESS ES1688)

Error: Comment is Missing!