It’s new year’s eve, and I’m back with the last part of the salvage season posts. Just as well, since salvage season is coming to an end as the year is coming to an end. This post looks at some old salvaged I/O boards and fluorescent tube starters.
Fluorescent Tube Starters
I mentioned in the previous part that I had salvaged a batch of starters, and the Crompton starter was somehow anomalous. As a result, after cracking them open, I lined up the three different types of Philips S10 that I had salvaged.
The newest one of the batch is to the left, and has a green base and line-art drawing on it. It also has the modern Philips anti-counterfeit measures of a customized starter top that curves inwards, and pins that are hollowed out at the bottom.
The remainder are older Philips S10s from the early 90’s, and have bakelite bases rather than the green plastic base of the new one. These older starters aren’t all the same – note the can of the rightmost one has an underline under S10, whereas the identical four middle ones do not.
The older units had a horizontally shaped RFI capacitor with neon green sealing at the ends, whereas the newer ones have a vertically shaped capacitor.
In all instances, all six starters seem to have had some decent level of service in their lifetimes, with the insides of the glass coated in so much deposited metal that even when photographed in front of a diffused flash, the light fails to show the internal bimetallic strip structure. I can say that the structure is a symmetrical two-leaf design, but it just can’t be seen directly. The deposited metal is likely due in part to the sputtering that occurs when the glow is running (essentially cold-cathode) in the starter, and at the arcing events that occur during the connection and parting of the contacts. The amount of deposited metal seems to vary, and might be a hint as to how much wear the starter contacts have seen (although would not indicate other causes of starter failure – for example, fatigue in the bimetallic strip, loss of the fill gas, broken internal joints, etc.)
While these are all Philips starters, it seems they might not all have the same fill-gas combination or pressure. The first one I tested had a dim purple glow in the gas, common for some other brands, and a 1mA breakdown voltage of 132v DC. The remainder had a more orange-pink glow, but their breakdown voltages varied from 140v through to 170v DC. All of these values are acceptable, and they might represent some degradation over time from manufacture, but it’s interesting to see that there is some variation within the brand over time.
The Crompton starter breakdown voltage was much higher than this, and was probably the reason why the starter failed to function as expected.
These boards were technically not salvaged at the same time as the rest of the items, although they were salvaged this year. These also came from a tech clean-up and were sitting in a box awaiting documentation. At long last, the impetus and time both simultaneously appeared, so here are some older I/O boards along with their ROMs.
The first is an Adaptec AHA-2940 SCSI card dated Week 39 of 1995 with BIOS 6E00. This is a newer narrow SCSI board with an HD-50 connector on the rear, rather than the formerly more common DB-25.
Removing the label revealed a type of IC that the MiniPro could handle, so the BIOS was dumped for back-up to allow restoration in case the flash loses its charge over time.
The beginning clearly shows the BIOS identity.
The BIOS is pretty advanced, as it has a SCSISelect utility that can be accessed with Crtl+A. A few error message strings are seen here.
A much more extensive list of strings is seen in the two images above. As a result, it seems the adapter had quite a lot of internal diagnostic capability, and also the ability to format disks. Some early hard drive controllers did have such ROM based utilities, however, to invoke them required running debug in MS-DOS and jumping the execution into the ROM space at a particular address noted in their documentation where a short machine code segment in the ROM helped you configure things such as the drive geometry, interleaves, etc.
A second SCSI adapter was part of the lot – this is a Future Domain TMC-3260 based around a 9405AV chipset. It seems there are three soldered down programmable array logic chips – very much a rare thing to find nowadays since FPGAs have taken over that area of the market. The ROM is 36C70 V1.3 dated 1994 and is on a UV-EPROM.
Careful use of a knife allowed me to obtain the part number without destroying the label entirely.
By comparison, the BIOS is much smaller and more simpler. Interestingly, the ID seems to have IBM F1 P264/32 in it, along with V1.00 dated 18th August 1993. It has some readable strings such as ATLAS SIM-19xx-PCI and FDC ROM/CAM 1800 FDC TMC-800. Might this indicate some code-reuse between different PCI based adapters?
Towards the end, we find some short basic strings which are probably used during the start-up. There seems to be a message for in case users have some older ISA Future Domain controllers installed as well.
The next card is a Hong Kong made, generic floppy and single-channel IDE hard drive combination controller. This was a pretty common and interchangeable part in earlier PC/ATs, and because they all emulated the IBM behaviours, they didn’t need any drivers and theoretically any board would work nearly as well under MS-DOS (barring physical design faults and chipset errata). The truth was that there was some subtle variations in the capabilities of various floppy controllers which would impact some more sophisticated “higher” density software or “multi-format” disk copying software such as Teledisk. This one is based around a UMC controller, so it was probably a cheap and nasty unit. Judging by the soldering on the rear, it looks pretty crudely hand-soldered to me.
Another card that was customary to have within a PC/AT is the multi-I/O board like this. This would have been used in the “intermediate” era AT’s, as it had two serial ports, a parallel port and a game port. Earlier ATs had the parallel port supplied by the graphic card, and often just had a dual-port serial adapter as another card. Anyhow, this was not in any way sophisticated, since the serial port is being driven by a W86C450 for the serial ports, equivalent to the 8250A which isn’t capable of really high speeds (above 9600bps) as its small 1-byte buffer means that CPU intervention was continually required to avoid losing characters.
As I mentioned, these cards are fairly generic, so here’s another made around the same chips but with a different layout and different supporting circuitry choices. Functionally, they would be equivalent, save for the missing serial header cables.
The card above would probably elicit fond memories for some, being a VLB card. This stood for VESA Local Bus and was a higher performance way of connecting bandwidth hungry devices to a 486-based CPU. This was basically done by adding a connector at the end of the 16-bit ISA bus slot, formerly used by MCA and similar to PCI except in brown. These long cards often caused issues with poor contacts, difficult installation and compatibility issues especially if multiple cards were used (e.g. a high performance storage controller plus a graphics board).
This one above is a Tseng Labs ET4000/W32P, one of their later efforts which is also PCI capable. The date is Week 12 of 1994 on the chip, and it comes with four Samsung KM416C256AJ-7 VRAM chips for a total of 2MiB. This board is branded ALTOS/VESA BN1000.
The BIOS did have an interesting distingishing feature – the disclaimer at the beginning [This is not a product of IBM (IBM is a trademark of International Business Machines Corp.) and is dated 25th October 1993, V8.00N. It seems it could support anywhere from 256kiB to 4MiB of display memory. I suspect the disclaimer was because they wanted to use the IBM letters to pass VGA BIOS validation/compatibility checks, but didn’t want them to be sued by IBM for misuse of their trademark. For byte alignment, the punctuation includes a double-space. Clever.
While VLB cards are pretty rare to come across now, I managed to get two – this is a Cirrus Logic 2283RAV-B based on the GD5426 chipset. The chip date is Week 21 of 1993, and it has pre-soldered a total of 8 x Mitsubishi M5M44256BP321GE0C-7 DRAM for a total of 1MiB of video RAM. As the type of BIOS chip was not known, I didn’t risk damaging the chip by reading it out.
Onto somewhat more modern boards, this is a Matrox MGA G200. The chip is slightly discoloured in the middle due to heat, and it’s dated Week 32 of 1999, and uses the AGP bus. This one seems to be fitted with standard SDRAM rather than dual-ported graphics RAM, so isn’t going to have the best performance. I do have quite a few old Matrox cards, and other AGP cards for that matter, but I never got around to posting about them, and I don’t think I will anytime soon either.
Cheap, cheerful, and not very high performance – the SiS 6236AGP was one of the cheap options for someone looking for an AGP graphics adapter. This one seems to be the “cheaper” model without the luxuries such as TV out, BIOS socket or more RAM. Nothing too interesting at all.
This is a Diamond Stealth 3D 4000 AGP 4Mb card, based around the S3 VIRGE/GX2. Notice how the chip has the words “On Board” on it … an interesting choice. I did find some of their GPUs “onboard” some motherboards, and they were acceptable but again, nothing too special in my eyes. This one is missing the luxuries as well, but at least it does have proper video ram for better performance and a socketed BIOS. This card is dated Week 9 of 1998 on the PCB.
The BIOS is V1.04 and dated 24th March 1998.
I don’t normally deal with Apple hardware, so this came to me as a bit of a dusty surprise. This is an Ethernet LC Card, dated about 1991, and connects to the processor-direct slot of a (presumably) Macintosh LC to provide Ethernet connectivity.
The rear, instead of offering a direct media connection, instead featured an AAUI port to connect to a media transceiver so as to offer flexibility between different sorts of networks (e.g. Coax and Twisted pair) as none were particularly dominant at the time.
The whole act of salvage is rather interesting, and people get motivated to do it for different reasons. Sometimes, it turns into hoarding, but for me, it’s mostly about fulfilling my interests, learning, experimenting and playing with gear that I wasn’t able to otherwise obtain. It’s also a good source of items that money can’t buy, and broken things which might be repairable (and educational in the process). Considering the gear would otherwise end up in landfill, salvaging things has a mostly positive environmental impact (although, sometimes this might not be the case where very inefficient devices remain in use for longer periods). I find it a very enjoyable activity, although one that some people seem to look down upon. If done properly, it can be fairly safe, although I wouldn’t advocate doing silly things like “just plugging it in”, because it’s not worth risking your life over someone else’s rubbish!
Through a variety of channels, many different sorts of items were salvaged. As my interests primarily lie in computing, electrical, lighting and electronics; those are the products that were salvaged. However, there are many other things out on offer – furniture is a common one.
I suppose the only beef I have with some salvagers is that some people are doing it for profit. They go around seeking items with metal (e.g. computers) for gold or scrap metal recovery, and in the process, destroy items which could have much more value to a retro computing enthusiast. Often, they indiscriminately rummage through the neat piles, transforming them into disorganized heaps where certain light-weight items are left to blow across the road with the slightest wind, having no consideration for the environment or fellow motorists. Aside from this, they pick up an inordinate amount of items far beyond what an individual would have use for, in order to run their own garage sales from home for their own profit, in essence depriving the community of a “communal” pool of unwanted objects in order to boost their personal gain. I don’t ever salvage more than I can play with, and I don’t make a business out of selling salvaged items because I don’t think that’s right, but other people do and it really annoys me.
I suppose what I’m saying is that proper and responsible salvage is good for me (as I get the item, educational benefit from a teardown/repair, free components to repair other things), good for the council (that doesn’t have to transport so much rubbish and pay for its disposal), good for the item (which might see more use than it otherwise might have) and good for the environment (which avoids having more things thrown into landfill, at least temporarily, and potentially could displace a new item being bought in some cases).