brostenen wrote:

When documentation is wrong.... I mostly see it on Socket3 systems.

Oh yes! The dreaded: "loose page (long lost by now) inside the manual describing some late change" comes to mind.

LHN91 wrote:

Things working *almost* perfectly.

I second this.

I just finished working on my Cyrix MII 90s gaming PC- Only one problem, the OS (Windows 2000) never detects my DVD drive so I can't really use it at the moment. Don't have this problem with 98 or XP.

Chinese srams(DIP) - waste of money and time.

amadeus777999 wrote:

Chinese srams(DIP) - waste of money and time.

I tried to install L2 cache in my Packard Bell Pack-Mate 28 Plus to no success (damn Chinese manufactured shit) as it wouldn't POST at all (I even put the chips in the right orientation).

For me: simple things that should make no problems, but do. (e.g. install a 40 GB IDE drive only to find out that the BIOS does not support LBA)

Apart from that, faulty hardware that seems to work at first glance.

bjwil1991 wrote:

amadeus777999 wrote:

Chinese srams(DIP) - waste of money and time.

I tried to install L2 cache in my Packard Bell Pack-Mate 28 Plus to no success (damn Chinese manufactured shit) as it wouldn't POST at all (I even put the chips in the right orientation).

I have blown close to 200$+ on Chinese "srams" and thus far I can say - stay away from them in general. In essence they're junk. I'm still waiting on a batch of supposed "Alliance" ICs and if these fail too then it's on to seek out a source on Kynix. They have way higher prices(up to 700%) but there's a chance that these could truly be new/old stock... hopefully the real deal. BUT I can't, of course, confirm any proof for a valid product before having tested it myself.

Feiopa bought fake 10ns generics a few years ago, but these came from a good factory/batch and were most likely able to withstand "12ns requirements". A few were bad but all the rest(256/512/1024) have turned out to be fully functional.

I have also bought around 70+ original srams(ISSI, Winbond, UMC and Cypress with ~95% being 15ns and the marginals 12ns Winbonds) and these hardly fail albeit they are also picky about which batch they stem from.

There's only one source on ebay for authentic 12ns caches and they're unfortunately small 16K ICs which can at most be tag rams(should the board support these).

A delicate issue(especially since sram testers are EXPENSIVE) and really shitty situation in general for folks who like to have fast, late era 486 boards running at full steam.

amadeus777999 wrote:

I have blown close to 200$+ on Chinese "srams" and thus far I can say - stay away from them in general. In essence they're junk. […]
Show full quote

bjwil1991 wrote:

amadeus777999 wrote:

Chinese srams(DIP) - waste of money and time.

I tried to install L2 cache in my Packard Bell Pack-Mate 28 Plus to no success (damn Chinese manufactured shit) as it wouldn't POST at all (I even put the chips in the right orientation).

I have blown close to 200$+ on Chinese "srams" and thus far I can say - stay away from them in general. In essence they're junk. I'm still waiting on a batch of supposed "Alliance" ICs and if these fail too then it's on to seek out a source on Kynix. They have way higher prices(up to 700%) but there's a chance that these could truly be new/old stock... hopefully the real deal. BUT I can't, of course, confirm any proof for a valid product before having tested it myself.

Feiopa bought fake 10ns generics a few years ago, but these came from a good factory/batch and were most likely able to withstand "12ns requirements". A few were bad but all the rest(256/512/1024) have turned out to be fully functional.

I have also bought around 70+ original srams(ISSI, Winbond, UMC and Cypress with ~95% being 15ns and the marginals 12ns Winbonds) and these hardly fail albeit they are also picky about which batch they stem from.

There's only one source on ebay for authentic 12ns caches and they're unfortunately small 16K ICs which can at most be tag rams(should the board support these).

A delicate issue(especially since sram testers are EXPENSIVE) and really shitty situation in general for folks who like to have fast, late era 486 boards running at full steam.

I cannot seem to find the 64kx1 15ns 22 pin by WinBond, ISSI, UMC, or Cypress, only AT&T chips have that on eBay as my Packard Bell requires 5x 28-pin (or 4x 32-pin) 20ns (4x) and 15ns (1x) TAG SRAM chips, and 1 22-pin 15ns DB SRAM chip.

Can't comment on the 22pin chip, but standard srams are in general a "per seller item" which means the search is not really useful. If you have a good seller on ebay, bookmark him and see what updates he has. All srams I got were mostly from seller updates and other private sources. Cypress SDIP srams are basically non-existent.

krivulak wrote:

For me it is making memory in DOS. It takes me forever, as a matter of facts I have never ever finished any build because I can't run anything because of lack of memory. Not that there is too little, but because I can't manage it.

There's a science to find compatible drivers that use less memory. Try digging for really old drivers - especially for CD ROM and mouse.

Also: for me, the frustrating part is incompatibility issues. Early 90's hardware was notorious for all sorts of weird compatibility issues of which there was NO documentation on how to fix. PC magazines were full of people complaining about how sound card X wouldn't work with brand Y or game Z. I grew up with IBM who used its own custom IBM DOS which had some incompatibility issues as well with certain games.

I agree about molexes that are too tight. I've had plenty that left the side of my finger bruised, and sometimes scratched a knuckle on the IDE pins.

How about chassis bare metal frames with edges like a hacksaw blade. I've left blood stains on ribbon cables

CD drives that are so long they interfere with cabling, and cards that are too long to fit beside the cpu heatsink

Dracolich wrote:

CD drives that are so long they interfere with cabling,

This why I try to use new-ish (post 1998) CD drives whenever possible, however especially with some older and more obscure computers this isn't possible.

For me the most frustrating part is not about working on the computer, but finding software for it, games in special,sometimes gog games don't work properly on a real ms-dos machine, and most games are pretty much impossible(even for a high price) to find here, so downloading is the only option, and all the copy protection schemes that are based on key disk.

My top "jimmy rustlers" when it comes to working on old PC's:
1. Almost broken stuff, especially laptops. You spend tons of time reinstalling it, just to find out something like a usb/serial port/etc. is busted. Or boards with few defective RAM/PCI slots.
It's so infuriating that I throw that kind of gear to the garbage bin 99% of the time.
2. Finding BIOSes and drivers, I still need to update my DFI UT NF3 UltraD to support the Opteron 185 correctly but there are no BIOSes to be found anywhere.
3. Weird RAM compatibility issues.
4. Swollen caps, replacing through hole components in a multilayer PCB is a nightmare.

martin939 wrote:

My top "jimmy rustlers" when it comes to working on old PC's: 1. Almost broken stuff, especially laptops. You spend tons of time […]
Show full quote

My top "jimmy rustlers" when it comes to working on old PC's:
1. Almost broken stuff, especially laptops. You spend tons of time reinstalling it, just to find out something like a usb/serial port/etc. is busted. Or boards with few defective RAM/PC slots.
It's so infuriating that I throw that kind of gear to the garbage bin 99% of the time.
2. Finding BIOSes and drivers, I still need to update my DFI UT NF3 UltraD to support the Opteron 185 correctly but there are no BIOSes to be found anywhere.
3. Weird RAM compatibility issues.
4. Swollen caps, replacing through hole components in a multilayer PCB is a nightmare.

Wow... I think this is my list.

I would add physically weak or poorly designed parts that break easily or fail in such a way that it makes repairs impossible or extremely time consuming. I'm thinking specifically of peripherals that are worth using\saving because they are old, rare and unique... but have terrible flaws that make repairs a nightmare. Like the IBM Model M2 keyboard. Compact, awesome clickety-clack feel and sound, standard layout... but the internals commonly suffer failures and the keyboard is an absolute nightmare to dismantle, every step of the way.

I cannot seem to find the 64kx1 15ns 22 pin by WinBond, ISSI, UMC, or Cypress, only AT&T chips have that on eBay as my Packard Bell requires 5x 28-pin (or 4x 32-pin) 20ns (4x) and 15ns (1x) TAG SRAM chips, and 1 22-pin 15ns DB SRAM chip.

For this type of SRAM, your best bet is probably the Micron parts MT5C6401. They seem to be relatively common, and I believe they are the only manufacturer that made a 12ns part.

Too-stiff Molex 4-pin connectors... especially in conjunction with certain old HDDs...

...yeah. Thankfully I've yet to actually break one.

Someone has mentioned carnivorous cases that bite and tear at you. I have an inch-long scar on my hand from fending off a Thermaltake beast a few years back.

Fortunately I haven't had too much bad experience with carnivorous cases- Most of the time I only get a small cut or two from fighting and taming them.

Thermaltakes are a special kind (in a negative way), that aluminium is sharp as a knife and they sometimes include what looks like brass strips for groudning at the expension slots - 1 swipe with finger on them and you'll be bleeding like a butchered pig 😵

- The neverending OCD battle between the part of me that wants to max out everything on a system and the part of me that tries to keep it period correct (I keep upgrading my GX110 to a Radeon 9250 then downgrading it back down to a Radeon 7000 or G450 every other week)
- Finding jumper settings for older hardware that are not on TH99
- Cutting my hands on stuff whenever I try to plug something in or out of a computer from before 2000