Motherboard: MSI MS-5156 Socket 7

It’s been a little while since my last post, partly because I’ve been quite busy, but also partly because I’ve been putting much of my energy into making youtube videos. I’ve been meaning to write something for the blog, but have been procrastinating because I’m having difficulty with the floppy drive repair I’m attempting.

I recently received a socket 7 machine, which was used as a data logger for physics experiments in a lab. It came in a chassis that was in poor shape so I’ve stripped the useful parts and taken some photos. Lets take a look, first at the motherboard.

It is an MSI MS-5156 socket 7 and appears to have been manufactured in 1997. It has an Intel 430TX PCIset chipset which from what I’ve read seemed to be fairly good for the most part with some disappointing but not deal breaking issues. My board has a Pentium @ 120Mhz and 16Mb of fast page RAM installed and also came with some cards I’ll show later.

Looking at the details of the chipset on Anandtech I found that it looks like a mid to lower range chipset. It does seem to support what were the latest and fastest Intel, AMD and Cyrix processors from the time. Although there are a number of technical issues where it is worse than it’s contemporaries, such as the range of FSB speeds, which doesn’t go as high as super socket 7 boards peaking at 75Mhz. Like the Elitegroup mainboard in a previous machine I looked at, it can not cache all the memory, with only 64Mb of a possible 256Mb being cachable. For most uses these wouldn’t have been serious problems at the time.

Some of it’s features are also forward looking supporting newer standards that were gaining traction at the time. It supports ATX power supplies which had some safety and design improvements and also supports SDRAM which is faster than the 72pin EDO and Fast Page RAM.

Like the motherboard, some components are not very high spec for the time this machine was originally built. The Pentium running @ 120Mhz is half the speed of the high end parts available at the time, and whilst 16Mb was still common in 1997 it was a bit small and slow being fast page RAM.

The video card on the other hand is a Diamond Multimedia Stealth 64, it has the S2 Trio64V+ chip that was very common on 2d graphic cards of the time. I’ve found the S3 based cards are usually quite good for anything that doesn’t use 3d acceleration (although they do pair nicely with a Voodoo 2). This particular example comes with expanded video memory (2Mb?), so it can display at higher resolutions.

The other board that was present is the analog data acquisition board, an Compuscope 220. It appears to be basically a 2 channel 25Mhz digital oscilloscope from what I can read on the web. It has 4 BNC connectors two of which are the input, I assume the others are either trigger inputs (or outputs) of some description. There is a sticker on it indicating that it was last calibrated in 1991, so it’s significantly older than the rest of the PC. Unfortunately without the software this is likely completely useless.

Looking back at the motherboard we see that the jumper settings for adjusting the FSB and multiplier are documented nicely on the silkscreen. There are additional tables listing settings for specific CPUs from each of the manufacturers if you don’t know the FSB and multiplier for your particular CPU. This leaves no room on the silkscreen for documenting the other jumpers, which leaves you reaching for the manual to hook up the front panel or adjust the memory voltage for instance. Look here for the manual, settings are also listed at stason.org.

This machine was clearly not meant to be fast, I suspect it was built simply to contain the data logger card and run the software that was likely in a much older machine before this one. Perhaps the older machine had died, and this was a solution to keep the equipment running. So in that sense it was probably more than fast enough for what it was used for.