Just because I want the factory preinstall available to check out, since I had the recovery CDs with it. The more I use Windows Me the less I like it, but it's nice to check out now and then.

Kahenraz wrote on 2023-05-02, 15:02:

It sounds like the drive is having trouble spinning up to speed. Does it work intermittently or not at all?

Good question ! Have not had time to check it more thorough as there are 4 HDs and 3 case fans so with side cover off cannot really tell.
One thing: BIOS see's all 4 HD but one is not shown under Device manager and it's drive letters are missing. Pulled cable off it and xp starts like expected.
That is as far as I got. Will pull it out and see if it spins up on the quieter bench machine.

Update: moved HD to another computer and it gets detected by BIOS but also causes Windows 7 to hang. It does spin up and sounds normal. Using a Linux based boot CD was able to grab a few non-replaceable files off it (that took near an hour for a few 100Mbs).
So a drive can die but still be detected by BIOS and appear to be functional but not... it is some odd thing I never seen before. Luckily had just a bunch of replaceable files on it.
Guess I should have mirrored it to a newer drive a few years ago or made a back up image. Just noticed it been in near daily use since before 2009....so a lot of time on the drive.
Last check with Crystal DIsk it was still Green, no issues....

Uhg... I've got this "problem"...

I have a hard time figuring out how many old floppy disks to keep.

I have finally managed to figure out what I have, copy the contents of anything that has any chance of not being available online, putting the good disks in a box to get rid of and chucking the obviously bad disks... but I still have a ton of them! (And don't worry, any that are good that I don't intend to keep will be going in a lot of used disks to sell... not in the garbage. I just need to clear out the space).

I am planning to keep all of the 720K 3.5" floppies because I don't have that many and I find that they are super robust and are far less likely to be defective (probably due in part to the less dense data, but they are just plain heavier and rarely make bad sounds while spinning).

I deleted the contents on probably 25 360K 5.25" disks to keep them as blanks, but then when I realized that like 60 more 5.25" disks were almost all filled with partial .arc files of very common programs (dbase, autocad, orcad, etc.) that didn't need to be preserved AND they were basically all readable except for maybe 3... I'm having a very hard time getting myself to just dump these in a lot of disks on ebay when they are apparently built well enough to last 40 years in storage (all written in the mid 80s), and at some point I just won't be finding more of these.

I have probably 15 or so brand new 5.25" floppies. What has me wanting to keep all the used ones though is that up till now 98% of my "retro computing" has been with systems that have hard drives and at least the option of 3.5" drives (or other storage), so there's a good chance I'll be using far more of these as I start tinkering with older PCs. I own a couple of Apple IIe systems that I intend to dabble with more in the future... if an Apple IIe can use the same disks as a PC (after formatting of course), then I definitely feel like I should keep the extra disks so I can copy games to them. The same goes for any other types of machines I may get in the future. I know there are differences between single sided and double sided, but can older PCs with 5.25" drives basically all use the same disks that a PC would format as 360K? I believe every one of the disks I have tested recently is DS DD or at least formatted to 360K.

As far as 1.44MB 3.5" floppies go, I have... *counting*... oh wow... 175 Imation disks brand new in package, 50 Fujifilm brand new in package and a largish (maybe 200 count?) box of really generic looking ones that are also brand new. I will never ever use this many... but part of me is also concerned that at some point the materials these newer disks are made of will simply deteriorate and they'll all be dead, whether they're used or not. I know that the quality of floppies dropped over time, and these are all definitely newer ones... but when I look at the space taken up by yet more old used floppies with writing all over them, it seems kind of ridiculous to hang onto them while also hording almost 500 brand new ones. 🤯

And just to be clear... I don't buy these things. Almost every one of these disks I'm talking about now have been given to me in the past 7-8 years since I started collecting this stuff. Both the new ones and the old ones.

What I'll probably do:

Keep a small pile of the best looking brand name 3.5" used disks just so I don't feel incredibly stupid later if all the new ones end up being junk.

Keep as many of the 5.25" disks as I can reasonably store in one container without taking up more space (looks like about 90 used ones in sleeves, and 15 new ones)... as well as a bunch of extra sleeves, because... I have no idea. 🙁

... ugh, and all this will leave me with a few really nice but totally empty disk containers if I sell a whole bunch of disks online. 🤯 (Obvious solution is to then get rid of the containers... but they aren't exactly worth much, and why would any self respecting retro PC enthusiast want to get rid of floppy disk containers that could be filled with game disks later?? Amirite??)

Ozzuneoj wrote on 2023-05-04, 04:42:

Uhg... I've got this "problem"...

I have a hard time figuring out how many old floppy disks to keep.

i admit i keep only ones i use as boot disks and the rest are fading away, when they don't work they go

i probably have enough to see me through however many boot disk moments i will need

i just dont have that many PCs that don't also have at least one usb, which makes file transfer easy

more a reflection of my PCs really, a relative lack of pentium and earlier era machines

I updated a BIOS on a G33M02 board, not in itself exciting nor all that retro - but it is so much easier on 'newer' boards than it was in days gone by

the bios update enabled later 775 CPUS to be used incidentally, beyond wolfdale / allendale - so i replaced the e2140 (1.6ghz) with an E5300 (2.6ghz)

extra L2 cache and the extra pace makes a difference but still has ddr2 ram of course

it was just an impulse to upgrade a bit, without actually having a specific role for that machine as yet 😀

Pimped (if you can call it that) the Medion machine I got last Sunday.

List of upgrades:

- Spire SP-ATX-420WTN-PFC PSU in place of the 250W FSP it had, due to the new GPU (Radeon 9700) eating more power than the 250W unit can provide
- 3D Hercules 9700 Pro 128MB in place of the FX5600 card (which turned out to be a "SE" variant, of all things)
- 2GB worth of DDR400 in place of the crappy Siemens 2x256MB sticks it used
- replaced Creatix combo card and WLAN card to a SB Audigy SE and an ASUS WLAN card respectively (the Creatix couldn't connect to WPA2 networks)
- replaced Pioneer DVD-RW and Ultima DVD-ROM drives with a single Lite-On DVD-RW drive
- installed a stock copper-core Northwood HT/Prescott cooler in place of the funky mounting heatsink it had w/ some strange black goop
- updated the BIOS
- replaced original case (which was in a pretty nasty state) w/ a Linkworld mATX case

Well, this is a project i've been working on for quite a while now, a Slot-A to Socket-A adapter. This is currently revision 0.8 (the 7th major redesign) and i hope to be able to make the PCB order this weekend:

It should support CPU's all the way to the later Barton-based Athlon XP's with some bios modding for additional CPU support. It's got some dipswitches for Vcore, Multiplier and Cache control. (Multiplier/Cache control requires connecting a few L bridges on the CPU) There are AM4 mounting holes as well for better cooler compatibility, water blocks in particular. This is kind of a learning project for me to get more familiar with complex/high-frequency PCB design and it combines nicely with the retro-hobby.

AMD K6 wrote on 2023-05-04, 20:21:

Well, this is a project i've been working on for quite a while now, a Slot-A to Socket-A adapter. This is currently revision 0.8 […]
Show full quote

Well, this is a project i've been working on for quite a while now, a Slot-A to Socket-A adapter. This is currently revision 0.8 (the 7th major redesign) and i hope to be able to make the PCB order this weekend:

It should support CPU's all the way to the later Barton-based Athlon XP's with some bios modding for additional CPU support. It's got some dipswitches for Vcore, Multiplier and Cache control. (Multiplier/Cache control requires connecting a few L bridges on the CPU) There are AM4 mounting holes as well for better cooler compatibility, water blocks in particular. This is kind of a learning project for me to get more familiar with complex/high-frequency PCB design and it combines nicely with the retro-hobby.

Wow, that's amazing! Nice job!

I will say though, Slot A boards are pretty hard to come by compared to Socket A boards.

Ozzuneoj wrote on 2023-05-04, 20:29:

AMD K6 wrote on 2023-05-04, 20:21:

Well, this is a project i've been working on for quite a while now, a Slot-A to Socket-A adapter. This is currently revision 0.8 […]
Show full quote

Well, this is a project i've been working on for quite a while now, a Slot-A to Socket-A adapter. This is currently revision 0.8 (the 7th major redesign) and i hope to be able to make the PCB order this weekend:

It should support CPU's all the way to the later Barton-based Athlon XP's with some bios modding for additional CPU support. It's got some dipswitches for Vcore, Multiplier and Cache control. (Multiplier/Cache control requires connecting a few L bridges on the CPU) There are AM4 mounting holes as well for better cooler compatibility, water blocks in particular. This is kind of a learning project for me to get more familiar with complex/high-frequency PCB design and it combines nicely with the retro-hobby.

Wow, that's amazing! Nice job!

I will say though, Slot A boards are pretty hard to come by compared to Socket A boards.

That's a valid point indeed... But 800mhz+ Slot-A cpu's are also becoming pricey and hard to come by. For reference, i have 80 Socket A cpu's and 1 Slot-A cpu. (and 4 Slot-A motherboards...) I could go above and beyond to find a 1GHz Slot-A cpu but then it's only usable on my Slot-A boards... Or use a slotket and have a large amount of Athlons to choose from that can still be used in other boards. And it's kinda nice to just have some new stuff to mess around with, maybe building a nice competitor for the Slot-based Tualatin systems with something like a Palomino based cpu.

AMD K6 wrote on 2023-05-04, 21:34:

Ozzuneoj wrote on 2023-05-04, 20:29:

AMD K6 wrote on 2023-05-04, 20:21:

Well, this is a project i've been working on for quite a while now, a Slot-A to Socket-A adapter. This is currently revision 0.8 […]
Show full quote

Well, this is a project i've been working on for quite a while now, a Slot-A to Socket-A adapter. This is currently revision 0.8 (the 7th major redesign) and i hope to be able to make the PCB order this weekend:

It should support CPU's all the way to the later Barton-based Athlon XP's with some bios modding for additional CPU support. It's got some dipswitches for Vcore, Multiplier and Cache control. (Multiplier/Cache control requires connecting a few L bridges on the CPU) There are AM4 mounting holes as well for better cooler compatibility, water blocks in particular. This is kind of a learning project for me to get more familiar with complex/high-frequency PCB design and it combines nicely with the retro-hobby.

Wow, that's amazing! Nice job!

I will say though, Slot A boards are pretty hard to come by compared to Socket A boards.

That's a valid point indeed... But 800mhz+ Slot-A cpu's are also becoming pricey and hard to come by. For reference, i have 80 Socket A cpu's and 1 Slot-A cpu. (and 4 Slot-A motherboards...) I could go above and beyond to find a 1GHz Slot-A cpu but then it's only usable on my Slot-A boards... Or use a slotket and have a large amount of Athlons to choose from that can still be used in other boards. And it's kinda nice to just have some new stuff to mess around with, maybe building a nice competitor for the Slot-based Tualatin systems with something like a Palomino based cpu.

Yeah, it's definitely an interesting project. Slot A CPUs are pretty abundant compared to the motherboards, but anything above 700Mhz or so tends to be quite pricey, and anything over 900Mhz is rare. I would be interested to see how a later model Athlon XP runs in an old board if it's possible. Though I would imagine that you'd be limited to 100Mhz FSB, since the slot A chipsets never supported 133 or higher. If there was a way to get it to support the proper voltages for mobile Barton chips you could get a 16x or 16.5x multiplier and 512KB of cache with very low temperatures. That'd be a pretty cool setup. 😀

That's a super amazing project indeed! I hope it works in the end 😀

PcBytes wrote on 2023-05-04, 17:23:

List of upgrades:
- Spire SP-ATX-420WTN-PFC PSU in place of the 250W FSP it had, due to the new GPU (Radeon 9700) eating more power than the 250W unit can provide

I really doubt the whole system would ever come close to 200W in peak load. Those Spire PSUs are bottom of the barrel trash.

Munx wrote on 2023-05-05, 06:28:

PcBytes wrote on 2023-05-04, 17:23:

List of upgrades:
- Spire SP-ATX-420WTN-PFC PSU in place of the 250W FSP it had, due to the new GPU (Radeon 9700) eating more power than the 250W unit can provide

I really doubt the whole system would ever come close to 200W in peak load. Those Spire PSUs are bottom of the barrel trash.

This one is worth of at least 300W, from what I checked inside it (and it's been recapped ever since I got it). The golden cased units are decent though - it's the bland grey ones that suck ass.

That, and the FSP it had was literally rigged to burn any capacitor installed on the 12v rail - issue is, I cannot figure the load resistor that's installed on there (it got so hot, the color rings legitimately vanished.).
My other options would have been a CWT built ISO-500PP, an AOpen Z500-12AE3 (which besides being yet another FSP, it's way too overkill), and a HKC SZ-430PDR unit that looks decent enough for about 350W or so. (despite being rated for 430W)

Munx wrote on 2023-05-05, 06:28:

PcBytes wrote on 2023-05-04, 17:23:

List of upgrades:
- Spire SP-ATX-420WTN-PFC PSU in place of the 250W FSP it had, due to the new GPU (Radeon 9700) eating more power than the 250W unit can provide

I really doubt the whole system would ever come close to 200W in peak load. Those Spire PSUs are bottom of the barrel trash.

i'd agree that a lot of people choose PSUs with much more wattage than will ever be consumed

i'd also say that people confidently dismiss poorer quality PSUs when they actually will often last a long time and be trouble free, surprisingly so given their reputation and the willingness of some to buy very expensive highly rated PSUs rather than 'risk' a cheaper one

and some will add that lower quality / cheaper PSUs will destroy the system, saying this a lot more than actually happens

Following watching this video, I just converted 2x sticks 72-pin 4MB EDO memory into FPM: 386 with EDO memory - by Bits und Bolts on Youtube

Essentially, EDO can often be made to operate as FPM on a per-SIMM basis by connecting the !OE pin of each memory chip to !CAS. Normally those are connected to Ground which makes this mod a little tricky since there's not usually an easily accessible trace that you could cut for the !OE pin - here's the pinout of the MT4C4007JDJ chip I'm working with for reference:

It requires sticks with 20-pin SoJ chips that are spaced fairly far apart and I tried two methods, one was heating the pin while pulling it out using some sharp tweezers as a pick. This was tricky but it worked for the first module. Once the pin is lifted from the Ground pad, bridge it to the adjacent !CAS pin.

The second method I tried is almost better and much quicker, but I ended up with one cut pin ending up underneath the chip and I had to refit it with hot air after cleaning that up. Anyway, this method is to use a sharp knife to cut that pin at a 45 degree angle by rocking the knife back and forth until it's through and hits the plastic. Push the solder side of the cut pin down a little bit to make space, then solder the cut !OE pin to the !CAS pin.

The board I'm testing with is an old MSI MS4143 which definitely does not support EDO memory, in fact if I install an unmodified EDO stick in the board, then it hangs with POST code C2. With the 2x modified sticks though, it boots and detects 8MB - time for some memtest 😀

I have a Slot A machine with a 800 MHz Pentium 3 + GeForce 4 Ti 4800 SE, a nice CRT Screen, a voodoo 2, and an ISA SB16 with a real OPL3, and 512 MB of RAM.

I'll never part with it. It took me years to build the damn thing. 😜

xcomcmdr wrote on 2023-05-05, 21:21:

I have a Slot A machine with a 800 MHz Pentium 3

Wait, what?

(Checking calendar to make sure it's not April 1st)

Nevermind, Slot 1. My bad. 😀

Got the delivery for a dead LS-486E Rev C1 board from Romania today. The board was sold as broken, as the SiS497 chipset QFP chip was damaged, with a pad (for the KBROMCS# signal to the VT82C41N chip) having been torn off.

Spent the rest of the evening working on it, I removed the chip with hot air, and bend the pins back enough for them to be able to be soldered to their respective pads, without shorting on one another.
I didn't go further than necessary, to avoid damage to the legs.

Then I laid down a jumper wire for the missing pad from the trace underneath, and began the tedious process of aligning the QFP chip and soldering everything in place.
My DM-9 digital microscope really does help for this fine pitch soldering work!

A couple of hours later, this is where we're at, all pins are visually inspected for shorts, and each pin given the push test with the tip of my tweezers, to ensure they're solidly soldered.
It still needs a good clean and some protection for the jumper wire, but it should all be good.

Now, I continued to chase my tail for a couple of hours, as the board was not getting beyond the 0A POST code (Set up interrupt vector table; Initialize first 120 interrupt vectors).
Tried multiple things, like swapping memory and testing connections from the SiS497, but it ended up that the Intel 486DX2 I was using might not be good, as switching to an AMD 486DX4 finally got the board to beep for a video card.

I put in a PCI video card, and sure enough, there is life again!

I'm very pleased with the result, as this was the first true QFP chip de- and resoldering job I've done, and it all went to plan!
Again, that microscope is a huge help, I don't think I could've done it without!

Daniël Oosterhuis wrote on 2023-05-05, 22:31:

I'm very pleased with the result, as this was the first true QFP chip de- and resoldering job I've done, and it all went to plan!
Again, that microscope is a huge help, I don't think I could've done it without!

you should be pleased 😀 thats a great job, interesting to read/see