Storage room finds – part 2

A few weeks ago I salvaged some equipment from a room clean-out, see the first post to see all the ISA bus parts and loose chips. Today we’ll be looking at the PCI cards, which are unsurprisingly all Adaptec parts. They are different from the older parts from last time in a few ways, firstly their construction is radically different because they use mostly surface mount components. They have fewer component counts with much much more integration on ASIC chips instead of off-the-shelf parts. Finally, Adaptec typically made some of the better SCSI cards with more processing done on the card rather than the host machine, this meant more CPU for applications and higher data through-put.

The first card here is an AHA-3940uw. This card was available for UltraSPARC systems as well as PCs. It doesn’t have RAID capabilities, but will do DMA transfers with-out CPU intervention to save processing on the host.  It requires the host system meet PCI specification 2.1 and that PCI-to-PCI bridges work on the host chip-set. I believe that this is because the middle chip is such a bridge and the chips on each side manage one channel each. The bridge chip appears to be manufactured by DEC (Digital Equipment Corporation also known as Digital for short) which is interesting as they weren’t really in the expansion card market.

This card appears to be from the late 90’s, it even still has MS-DOS support even though that wasn’t relevant at the time.

Here we have an ASC-29160 a 64bit PCI-X card from around 2000. PCI-X allowed for faster transfers to and from the card, which could have made quite a difference. Cards such as these could have something like 15-30 devices connected at maximum, although many fewer in practice (due to physical limitations of having that many HDDs). Many hard drives could easily generate more data than the standard PCI bus could handle (133MB/s) thus making the bus a bottle neck in the data flow.

The PCI-X standard can achieve twice the speed of PCI if running at the standard 33Mhz, but can get much faster with higher clock speeds that were offered. The most common speeds you’ll find are 33Mhz, 66Mhz and 133Mhz, but higher speeds were developed although not widely used.

This is a AHA-3940AUW which is essentially a redesign of the first card. It offers the same number of ports (at the same speed) plus a legacy SCSI connector all from the one integrated chip. It seems from the date codes that it was manufactured about a year later, so it’s probably just an incremental improvement (perhaps just for cost).

Adaptec also made lower end cards, here’s an example of one, an AVA-2906. It was made roughly mid 1999, but only supports the older SCSI standards at much lower speeds (10MB/s). It could have been used in the consumer market for scanners and early CD burners, both devices with lower bandwidth requirements. Whilst not being any faster than the ISA cards from last time, it would most certainly have cost significantly less.

Lastly here is a AHA 2940UW, which is basically just a scaled down version of the 3940 cards shown earlier. Whilst it’s not remarkable, it is a handy card as it supports most of the SCSI standards without being complicated or expensive.

That’s all the PCI SCSI cards that were saved, I did note a few things about them collectively as a group. Firstly most of these cards appear to be similar in both age and features, and they are all Adaptec cards. This was a common practice for a few reasons,  mostly ease of replacement (and fewer spares required) and less hassle when commissioning new equipment. It can save lots of time.

I’ve used Adaptec cards frequently specifically for their RAID feature, none of the cards here have this feature. The original machines mustn’t have needed either the additional space, speed or redundancy that RAID affords, as most of the cards featured also came in a version that supported RAID, but would have been more expensive.

Storage room clean-up

Recently at work we’ve had a little clean out of a storage space that had some old and unused computer parts among other items no longer being used. It turns out some quite interesting and useful old parts were in storage there, some of which I’ve been allowed to keep. This is of course on the condition that any data on any device or media is securely erased.

Most of the computer parts are Adaptec SCSI cards of various vintage, but there are also a few other ISA cards, loose chips, a couple of hard drives, and an amount of tape media that will be useful for some of my tape drives. So too much to show in one post, this time I’ll be looking at the loose chips and ISA cards.

First up we have these spare chips. The four at the top are SRAM chips, although it has been difficult to determine the exact size and pin-out, I think they are 256 x 4bits. The data sheet is proving elusive for these. The bottom three chips are EPROMs, the two left most being hitachi HN462732P chips and the right a NEC chip. The middle chip of the three is a OTP (one time programmable) so is basically of no use unless it is blank, but the other two are UV erasable so may prove useful in the future for repairs.

Here is what appears to be an early SCSI card made by Trantor in 1992. There seems to be little information about Trantor around on the internet, but it seems they were bought out by Adaptec at some point, although this doesn’t seem to be documented. If I remember correctly they were known for the tape drives they made.

This particular card is handy as it will fit an XT class machine and appears to have the chips and ROM necessary for booting of the SCSI disk. There are even still some MS-DOS drivers available for it. The silk screen has the jumper configurations, so I shouldn’t need a manual to use it.

Here we have two IO cards, one with an Acer chip-set and the other with one from UMC. These cards would have been exceptionally useful for 286 and 386 machines as they have most of the IO you would need all on one card. They both have two serial ports, a parallel port, a joystick port, floppy drive connector and a hard disk (ATA) connector. Quite the array of ports indeed for one card! The main problem with having one of these today is finding out what the jumper settings on the board are, there isn’t any silk screen to speak of.

This is an Adaptec AHA-1542b, made in early 1993 from the date code on the main chip. It supports SCSI-2 and can transfer data at about 10Mb/s connecting up to 7 devices. It’s interesting because it has a floppy connecter as well as the usual SCSI connection. You may notice that it has two ROM chips, one is the usual BIOS extension to allow for operating systems like DOS to access the disks. The other is micro-code for the boards controller.

Finally here we have two cards, both Adaptec. One is an AHA-1510A and the other is an AHA-1522A. You may notice that both actually use the same board, just the 1510A has far less components populated. The 1510A is basically a stripped down card with all the extra bells and whistles removed to make it cheaper. The user manual for it says that this cheaper card is “utilizing the CPU’s untapped power to improve system I/O performance.” So I’d say that it doesn’t have DMA support on the cheaper card among other missing features.

I had a look on the Adaptec website for all the cards shown today, and surprisingly despite their age downloads and manuals still exist. I can only applaud them for still offering the downloads, I only wish more manufacturers did the same.

Next time I’ll show the various PCI and PCI-X cards, and yes they are all Adaptec.

Storage Photo Tour 3

This weekend I was having a tidy up, between doing more domestic style cleaning I got out another one of my storage boxes to document its contents. I’ve done this twice before, once here at my place and once at my parents place. You can find the posts here and here.

This time around I found less older hardware and some more modern stuff, in particular a cache of AGP graphics cards of various types.

Storage photo tour 2

This week I came back to my parents place to visit them. I keep some of my old computer parts and books here and I thought I would share some photos.

First up we have some processors, a 20Mhz 386DX along with the much rarer Intel Overdrive and 486sx that is mounted for use in a socket. The fourth processor I couldn’t identify because of a permanently attached heat sink.

This is my Dads old calculator that he bought back in 1975! Its got an LED based display and even came with a power pack for when the batteries ran out. It’s built out of a number of circuit boards and Dad couldn’t remember who made it. The only label on it says made in England.

These two cards came out of a Reply Corporation machine that my Dad bought. I believe they are micro channel architecture cards. IBM introduced the standard to try and make the main bus of PCs proprietary so they could charge clone makers royalties. It didn’t work out for them in the end. The top card is a IBM SCSI card with a card edge connector instead of the conventional one. The bottom one is a Madge Ringnode card, which is a token ring network card. This type of network was popular before Ethernet came along.

This is a Western Digital Hard disk controller from an old Epson computer an uncle gave my dad. I’m not sure if it’s for RLL or MFM drives but at least it seems to have a mask ROM that should have the low level format utility in it.

This is the floppy controller from the same Epson computer, it has a Western Digital chip on it. It also has a serial and parallel port as was common on early PCs.

An EG-3000 EGA graphics card, it must be a later model as it’s quite small. The PCI VGA card is there for size comparison it’s a S3 Trio64.

This is an NEC FD1157 floppy disk again from the Epson computer, it’s an early 1.2Mb drive, you can tell by the stepper and spindle motor being larger and older style.

This is our old 386sx main board from our first PC, a Twinhead Superset 590. It has some damage from a leaky external battery, fortunately the board still works, but some connectors require replacement. The board has a chips and technologies chipset and a Paradise VGA adapter. The chassis was unfortunately ruined by the battery.

Dads floppy disk storage box and some KAO blank floppies in the box. I remember these primarily for how colourful they are.

A Tandon hard disk drive. I’m not sure but I think these drives like these could use different encoding methods, either MFM or RLL. The controller was largely responsible for most effort in encoding. This drive uses a stepper motor like many other old drives and is mounted inside a bracket with a faceplate and activity LED.

The 30 hour basic book is what I first learned programming with, as you can see it was really for the BBC micro computer. I was using gwbasic on the 386sx pictured earlier. The gwbasic reference helped me learn more about the language and filled in the gaps. It meant I could achieve more with the graphics, until I ran out of memory.

Here is the main board manual and the Open Access handbook. The latter is for a office suite that we used before getting Microsoft Works.

Finally here is the gwbasic manual I was lucky enough to get with the Epson computer, it’s basically the same as the previous reference manual. When the Epson was built it was common for machines to come with programming manuals. Many early databases and basic commercial software was written using basic, gwbasic made porting these easier. Usually they came from CP/M machines.

Frame buffer and hard disk follow-up.

I investigated a couple of interesting things hardware wise yesterday that may be of interest.

Frame buffer

Firstly I tried out the Sun Wildcat Expert3d lite frame buffer card I got during this week. To my surprise it does indeed work with my systems, in fact it seems to be hardware compatible with both of them.  For those who haven’t read my earlier posts, I have two more “recent” Sun machines, a Sun Fire R280 and a Sun Fire V440. I tried the frame buffer card in both machines and the ROM not only recognised the cards but displayed boot messages on the screen.

I don’t have any Sun keyboards or mice for either machine so I wondered if I could use a standard HID compliant PC keyboard and mouse. This also surprisingly worked quite well both systems recognised the keyboard as being connected as long as you plugged it into port 0 (the left most bottom connector looking at the back). I wondered what issue people were having with these machines, I didn’t wonder long.

Storage Box Photo Tour 1

A couple of weeks ago I got all my hard disks out and took photos of them. This made me wonder if any of the disks still worked and what was on them. I keep all my PC expansion cards stored in some storage tubs inside static bags with some desiccant packs to keep them dry, which means I would have to have a look in them to search for a SCSI card to test the drives with. I figured why not take some photos of the cards whilst I am at it.

Basic tech cleaning gear

Firstly I have here a picture of two very useful things every computer technician should have around. A can of isopropyl alcohol in spray form for cleaning slots of various types, you could use contact cleaner as well, but that leaves a residue sometimes. The box is full of isopropyl infused medical wipes which are good for general cleaning and can be used to clean card edges and the like. Very useful for cleaning thermal paste up if you ever need to, they are less messy than other methods. Not pictured but also useful is a tooth brush (not used for teeth anymore!) and cotton buds to aid in the cleaning process.

Hard Disk Heaven

Unfortunately my computers main hard disk and back up drive have once again thrown up some SMART errors, that whilst thankfully are non-fatal, have sent me into a flurry or backing everything up. This got me thinking about all the hard disks I have floating around in my collection and so I decided to dust them all off and take some photos of the more interesting ones.

MiniScribe Drives

First up are two MiniScribe MFM hard disks from the early days of PC computing. The one on the right shows the common use of stepper motors to move the heads in early hard disks, the other may be one of the first to use voice coils, we can’t tell from the outside.

MiniScribe as a company is famous for how they went out of business. They were running low on funds, but wanted to make everything look good on paper, so they sent out a bunch of bricks instead of hard disks to some customers who were in on it. They laid off some staff involved shortly after, who of course went to the papers and of course the rest is history. A rival company Maxtor ended up buying their intellectual property.

Full Height 3.5 inch SCSI Drives

The second group here are all full height 3.5 inch SCSI drives, all around the 200MB capacity. In the back left is a Conner cp3200F, a drive that came out of a 486 we had in my late teens, as far as I know it still works. At the back right is a drive made by Digital (Digital Equipment Corporation), this would have most likely been in one of their computing products of the time like a vax station. It may actually be a re-badged drive from another manufacturer such as Conner. You’ll notice they both look very similar. Also in picture is an old Quantum ProDrive 210S and a Maxtor LXT213SY.

Common SCSI Drives

Here we have a Quantum ProDrive LPS SCSI drive (top) that was very common in early Macintosh machines. I had heard that these drives had a problem where the lubrication in the spindle was too stiff for it to spin up until warm. Users were instructed to try to jolt the machine in such a way as to loosen the grease and start the drive. I guess they could have also warmed it up by some means. The story sounds a bit fanciful, but I didn’t use machines with these drives so I don’t know how true it is.

The drive on the bottom left is a Seagate ST31055W SCSI drive that is quite a bit more modern. Drives like it were frequently used in commodity PC servers in the early 2000’s. The drive on the right is a fujitsu MAN3184MC SCSI drive with an all in one connector. This connector was commonly used in more high end equipment and was frequently found in Sun machines. My old Sparcstation 20 will accept these drives and it was developed in 1992. This type of drive didn’t fall out of favour until SAS became the norm for servers, many servers currently still running are using these drives. I’m lucky to have a number of these drives spare including four 10,000 RPM HP drives of this type.

Some smaller IDE drives

Two of these drives are remarkable in the form factor they seem to fit in. Both are Seagate drives, a ST51080N and a ST5660A, which are about 1GB and ~500MB in size respectively. You can see that they are both significantly smaller than the Western Digital Caviar drive, which is about the standard size for most 3.5 inch drives today. The Western Digital drive is a Caviar 11200 10G disk, and as such is younger than the other two. It seems to have an IBM sticker on it, so at some point it came out of an IBM machine.

“Modern” drives

The last three drives are somewhat more modern than the rest, two are laptop drives which are an IBM TravelStar and a SATA Hitachi 120G drive. Both are pulled out of laptops from when I was working in IT support. The third drive is a Western Digital WD3200KS drive which is SATA and 320G. This used to be my main drive for a time until it developed bad sectors and got retired. I keep it pretty much only for nostalgia and the fact it has a nice backup of my system on it from the time it died. Whilst it was purchased in roughly 2006, it is a good physical representation of what modern drives look like today.

This concludes the photo tour of the mostly retired hard disks of my collection. In the time it’s taken me to prepare this post, I’ve managed to run the Western Digital tool on my main disk, and it is now looking healthy again without any data loss. I’ll still need to replace it sooner or later as errors like this are often indicative of more arriving soon. The backup drive is looking distinctly much less healthy, but at least it wasn’t storing anything more than a backup, so I’ll simply replace it when I get some money.

I’ve found it interesting that the older drives I have are more reliable than my modern equivalents, some of the old ones still work with no bad sectors, and I’ve put them to use in my Sun machines. Most however are untested. I’ll be keeping all the older drives as they have historical value even if they don’t work. Some like the small Seagate IDE drives I picked up simply because of the small form factor and are in my collection mostly as a curiousity.

I also have a bunch of modern hard disks that are untested, but are frankly mostly unremarkable, so I may dispose of them if I need space. They might become a good source of strong magnets.