brassicgamer.bsky.social
UK-based vintage gaming and computing fanatic. Curating, restoring and using ancient PC tech since 1993 for that nostalgia hit. He/him, ADHD/BPD.
425 posts
603 followers
177 following
Getting Started
Active Commenter
comment in response to
post
31/ Everything is back where it should be, more I can install the cache chips. I needed an extra SRAM and ordered it from silicon-ark.co.uk. It came really quickly and was packaged very well. Fingers crossed it works! Now I need to program the flash BIOS.
comment in response to
post
30/ Okay, so the kapton came away a bit and didn't entirely do its job... but the socket is out! 😅
comment in response to
post
29/ New socket carefully and successfully removed. Board has been prepped with kapton and fresh solder on each pin to make removing the old socket as smooth as possible.
comment in response to
post
28/ I don't have any spare 32-pin sockets, so I need a donor board. This Rockwell modem will do.
comment in response to
post
27/ I just checked the motherboard to see if there is physical space for a 32-pin chip and, interestingly there is an existing footprint on the silkscreen. The multimeter confirms the Vcc pin is connected to 5V so... maybe I can replace the socket with a larger one and not modify the chip?? 😮
comment in response to
post
26/ I have spare flash ROMs but they're 32 pin. Thanks to JEDEC, the pinout is the same, so I only have to do a little hack to get it working (Vcc and write enable are on a pin outside the 28 pin footprint).
comment in response to
post
25/ *sigh* I forgot you can't program an EPROM using a motherboard or a NIC, only flash chips. I will have to scavenge a flash BIOS chip from a later board and use that instead of an EPROM. 🙄
comment in response to
post
Ffs 🙄
comment in response to
post
24/ It's moot as the programmer has stopped working. 😠So I think my only option now is to hot swap on another motherboard. I don't have any flash chips or EEPROMs of the right type, only EPROMs, so I can't use the network card trick AFAIK.
comment in response to
post
23/ The only obstacle to testing this board is programming the BIOS. How do we feel about using this to power my USB hub? Best case scenario, it works perfectly. Worst case it fries the hub and my '90s Toshiba laptop. Maybe I should test what it's outputting, and check for ripple.
comment in response to
post
Yeah I've got a box in the loft containing my DVDs in their original fancy boxes, but if it was a just a film I've just taken it out of the plastic case and kept it on a spindle. For when the EMP bomb destroys all the digital copies!
comment in response to
post
If it was any other generic laptop I totally would but I kinda want to add it to my 'retro' armada as I was totally envious when she got it.
comment in response to
post
Well done! 😅
comment in response to
post
22/ I think the main issue with playing with unconventional hardware is the unconventional obstacles. But there are also everyday obstacles, like the power supply for my EPROM programmer deciding to just stop working. Now I need to find a 5V supply capable of delivering 3A.
comment in response to
post
5/ The main difference is the chipset, with the TMC using the OPTi 82C682 (with "50 MHz" seemingly scratched into the surface of the chip) and the ECS using the SiS 85C406. Just noticed that the ECS board is obviously later, dated week 46 of 1993, while chips on the TMC are a whole year later.
comment in response to
post
4/ The TMC uses a clock generator for frequency control, whereas the ECS appears, until I test it, to use a discrete crystal as a reference for whatever trickery the chipset uses to discern the CPU clock.
You can see from the photo that some models of the TMC board used a pair of XTALs instead.
comment in response to
post
3/ Both boards use SMD ICs exclusively for the logic, which I think is a more high-end characteristic for a board in 1992. But then the presence of EISA is the main indication of this. The ECS has 8 slots, however, while the TMC only has 7, one making way to accommodate the 72-pin SIMM banks.
comment in response to
post
2/ The ECS uses a ZIF socket, while the TMC uses a LIF. This could suggest the ECS is a later board, but the presence of 72-pin SIMMs on that board suggests the opposite. It could also suggest the TMC is a higher-end board, however, further hinted at because the TMC supports up to 512KB of cache.
comment in response to
post
Yeah this is exactly the kind of example that comes up regularly: in the States and expensive!
comment in response to
post
21/ Found some, that was easier than expected. 😅
comment in response to
post
20/ Next problem is the cache, which is a fucking nightmare. It requires three 16k x 4 chips for the tag etc. except they need to be in a 22pin package with at least 20ns access time. This is a stupidly specific type of SRAM and I don't think I'll be able to find any easily. 🙄
comment in response to
post
19/ Well 3) is the scenario that's looking most likely. There are very few jumpers on this board and all of them are in the manual. So there actually are no jumpers for setting the frequency, but it also doesn't mention this in the manual *at all*.
comment in response to
post
18/ 3) the absurd option which is that it somehow autodetects the CPU. This seems somewhat unlikely if not impossible for 1992. I don't know the pins of the 486 inside out, but I'm pretty sure there isn't one for signalling the native frequency.
comment in response to
post
17/ The problem with that is that there is no datasheet for this chipset. My options are: 1) identify all the jumpers that are documented and see if any jumpers remain that are not, 2) probe the chipset until I find a clock signal, then follow the traces, or...
comment in response to
post
16/ Normally in this situation (I say 'normally'; it's never happened to me before and it shouldn't happen) you would find the clock oscillator, look at the datasheet and follow the traces to find the jumpers that change the clock. The oscillator appears to be integrated into the SIS chipset...
comment in response to
post
Yes. Also fuck Varta more.
comment in response to
post
15/ I have come upon an absolutely bizarre situation with this board. The method for adjusting the front side bus speed is completely undocumented. Considering there are two jumper guides on The Retro Web and the original manual available this is just... insane. I guess I'll have to experiment?
comment in response to
post
14/ Not perfect but it'll do.
comment in response to
post
I know that's generally considered bad practice, but I want to make sure it works before I make any further modifications. I don't know the internal structure of these 'RAMified' packages, and I've not had any problems before so this will do for now.
comment in response to
post
This flux dispenser is super crappy. It just stopped working not long after I started using it and the plunger became warped. The flux is now coming out of the wrong end.
comment in response to
post
Once the package was out I used a manual pump to clean up and remove the remaining solder. Worked perfectly!
comment in response to
post
13/ My preferred technique is to use a circular file to wear down the case until the contacts are visible.
comment in response to
post
12/ Dallas chip has been removed, socket is in its place, now for the coin cell mod.
comment in response to
post
Ended up using the heat gun. Should have done that in the first place. Lost a tantalum and a couple of headers in the process but they'll be easier to refit than the Dallas was to remove.
comment in response to
post
Yeah but that shit is expensive!
comment in response to
post
😅
comment in response to
post
The end of the paper fastener (I do have some as it happens) it quite a bit larger but yes, the similarity is there.
comment in response to
post
11/ Another annoyance is that more aggressive timing cannot be used for the cache with a 50MHz CPU. Technically it can with 15ns chips, but it seems these didn't exist when this board was designed, plus the smaller capacity chips are impossible to find at that speed so maybe they were never made.
comment in response to
post
10/ It's a shame the cache chips were nicked, because the implementation is a pain in the butt. Later, simpler versions used an even number of cache chips plus one of the same type for tag SRAM. Earlier boards like this need 2 differing chips for tag and an additional chip for dirty bit. 🙄
comment in response to
post
Yeah that's what I was thinking, or like a push switch in an old LED calculator.
comment in response to
post
9/ Weirdly, after doing a microscopic inspection of the board to check QFPGA pins and the like, I found this little guy. It's slightly domed, but there isn't anything like that on this board so I'm perplexed. Glad I found it though, it was shorting some pins.
comment in response to
post
🎵 "Ain't nobody dope as me, I'm just so fresh, so clean" 🎵
comment in response to
post
I had my CRT TV and Dreamcast set up over Christmas for House of the Dead 2 with dual light guns, it was awesome.
comment in response to
post
7/ Another exciting thing I found in the manual is that there are many configuration options for the bus divider. This will allow me to tweak the EISA bus speed optimally. Normally stability is preferred by systems, so this is a nice feature to have.