ssokolow wrote:

Would this be a "wicks into cracks by capillary action method" suitable for reinforcing a crack that propagated outward from a h […]
Show full quote

cyclone3d wrote:

Going to have to pick up a new "bottle" of Loctite superglue to fix the board itself. Then on to fixing the traces and resoldering the power plug connections.

Would this be a "wicks into cracks by capillary action method" suitable for reinforcing a crack that propagated outward from a hole for a support pin?

bitmap.png

I had to scrape the solder mask off of and solder-bridge three points on the opposite side in order to get Start and Select working again. (Not bad for a "$5, for parts" controller, but I'd like something more robust than the "Not for use by rage-grippers" sticker I put on the back cover.)

...if so, what are your detailed instructions for doing it?

Superglue soaks into the cracks really well. Just put a bit on and it should wick right in.

To make it stronger I would use some type of epoxy after superglueing it. Plastic repair epoxy should work really well.

My K6-2/V2-SLI system now has a new home:

More photos here: Retro Rig Photo Thread

appiah4 wrote:

Ok, I just had to post this. That poor, poor AT case on the desk, on the far left, hanging on by the skin of its teeth..
🤣

Think of it as the Vintage/Retro computing version of the "hang in there, baby" poster. 😀

Thought I released the magic smoke from a nice 1GB CF card.

I love those backplate IDE-CF converters. Unfortunately aligning card in slot is a pain from the other side of the system. Somehow I managed to shove in the card side-on. That messed up some pins, but I didn't notice it until I had reinserted the CF card correctly and tried to boot from it. That didn't do very much and didn't smell right either...

Fortunately turns out that both card and adapter and IDE controller have survived.

Good news too, as that let me test my Aztech WaveRider32+ cards (both work, even the one I thought had dead MIDI) and NX Pro (works, including Covox hardware emulation). Shame how noisy it is...\

The only thing I wanted to do today that failed was testing my 'new' 8GB SSDs in a legacy system. My AN430TX detects the Promise SATA-150 TX2 controller, but without Molex to SATA adapter it's not going to do much more than that. So ordered a pile of them, as well as some Molex to mini-Molex (floppy) adapters and two good MIDI cables.Time to stock up.

The 286-12 restore is quickly coming to a close. All the cards are in and setup. MS-DOS is installed. Every component got a scrub in the bathtub. Tested interleave values with SpeedStor on the MFM hard drive for performance benchmarks. Interleave setting of 2 was the sweet spot for this setup. Takes a lot of time to do this because you have to low level format the hard disk every time you change the interleave setting and then fdisk and full format after that, but the performance gain is worth the time.

Big difference in performance of course:

Interleave - 1 = 31K/sec
Interleave - 2 = 220K/sec
Interleave - 3 = 160K/sec
Interleave - 4 = 122K/sec

cyclone3d wrote:

Superglue soaks into the cracks really well. Just put a bit on and it should wick right in.

To make it stronger I would use some type of epoxy after superglueing it. Plastic repair epoxy should work really well.

Thanks. 😀

286-12 DONE! Whew... that actually took way longer than I thought it would. On to the next one.

That's old Betsy running a 286 friendly demoscene demo. I have an ESS ES1868F based audio card in there along with a 256K VGA card.

The keyboard, monitor and speakers are just my workbench models and I know they don't look right with this elegant beast.

Speaking of which, the last four CRT monitors I've picked up from free Craigslist and Letgo postings have all been duds (washed out picture) so still looking for a decent local CRT.

I have spent the last couple evenings trying to find out why the 720K floppy drive in my Tandy 1000 TL/2 was not working. I presumed it was simply dead. A bummer since Tandy feeds power through the floppy cable too, meaning any replacement would require I hack the cable as to not bake whatever drive I put in place. I continued poking at it tonight though and noticed that I didn't hear the drive spinning when a disk was in place. A friend suggested I poke at the switches at the front of the drive (circled in the attached picture) and insert and eject a disk over and over again to work the switch. Tada! The first couple times, I had to lightly press the corner of the disk down to get it to detect and spin. But after doing the insert and eject trick over and over, it seems to have worked things loose. I need to get some contact cleaner on a toothpick or something and try to get it to soak inside, but for now, it's working! Getting this drive is important to me because I'm getting an ISA 1.44M floppy controller this weekend but without being able to install the 2M driver to access the floppy controller (and another program for Tandys that disables the onboard floppy controller), I would be dead in the water.

Finally, making progress!

Replaced the LianLie RH17 Mobile Rack in the 486 beacuse the old one died using a Laptop 2.5" HDD in an experiement in installing WIndows XP......so now I have a new one in there. However, I think I learned my lesson about hacking the power-interface key to disable it - DON'T - I took it apart thinking I just broke a wire or something - nope, surprisingly, there is actually LOGIC inside of one of those things (basically an eight pin SMD IC - probably to methodically switch the power lines off the drive before removal). It was one of those "might as well" things through since I DO have a need for some more RH17 trays for future youtube videos.

After getting hired on a new job, and waiting for my start date, decided to take on the data-recovery a friend wanted me to do. The agreement was I get to keep the computer & monitor if I can get the data off of it (it's her Grandma's old PC)....well lookee what I have here, a rare LPX based Socket 5 Pentium Tower - with a Micron branded CRT Monitor - so that's 2 out of the three (That is unless I decide to ditch either the Tandy or the 286....but we all know that's not happening and I might be keeping this PC as well). We were talking about this since may.

The Monitor has some nasty stains on top, but I'm sure I can make those at least look better, if not completely clean them entirely. Also looks like it might need some recapping and possible voltage/focus adjustment eventually, nothing too scary.

As for the computer, pretty cool, forgot NEC made computers. DId some poking around, looks like it has what probably is 16 or 32MB of RAM on 2 EDO SIMMS, it's a Socket 5, so it's definatley a 2nd generation Pentium, pulled the heatsink, no paste, one of those aluminum embedded pads for cooling, can't really read the digits either on the CPU Chip, but it is indeed a Genuine Intel Pentium processor in there. Not sure what clock speed. It has built-in sound, SVGA (with 2 open PLCC Slots for more VRAM - same as my old 486's S3 card), Serial, Parallel, and Game, and 4 ISA Slots and 2 PCI Slots. One ISA slot hosts a full length Modem card of some kind that I'll likely yank out as I'm not likely to ever use dial-up again. Looks all original too, so I might just restore it after the fact and maybe have one actually period correct system - though I tend to lose all resistance and eventually soup the crap out of it in the end into basically a Monster Truck vintage PC.

Data recovery starts tomorrow, might just be able to jam everything on a DVD-RW disc considering that hard disk is likely smaller or at most as big as 4GB in size assuming it's original.

Obligatory Pics....

sweeet NEC, those are really fun machines!

Don't forget to replace the clock battery/RTC Dallas clock since those machines would refuse to boot, even after "saving" the settings in the BIOS. My Packard Bell did that, and I put in a temporary solution until I put in a 2x AA battery holder soldered to the clock battery leads on the motherboard and it fires right up in a split second.

dionb wrote:

That's DRAM not VRAM. And half of it is socketed, so first thing to try is to remove those chips and see what that results in.

So yesterday, I finally got around to working on that PCI video card, and did what you recommended with the "DRAM" (I find that interesting, personally; still learning a lot about older tech).

And, what do you know? No problems. Decided to clean then and put them back in, to see if that would help, but ended up with the same corrupt images. Then, I decided to test each chip by itself, and, from what I can tell, one of them was indeed the culprit (the "-7" one if I am not mistaken). However, I guess given how that expansion memory works and how games utilize that memory, NFSSE was looking like it was sent through a cheese shredder with only the working chip:

Now, a question. I noticed with the bottom memory expansion, um, "bank" that some cracks had started to form in the plastic itself, one on the side of the bank, effectively splitting both sides of the bank, seen in this shot (it's hard to see):

and the other coming from the same side inward on the bottom of the bank:

Is this something I should really be concerned with, or is this only a cosmetic issue?

That -7 chip may not even be defective but just too slow for the board. All of the other chips are -6 i.e. 60ns, but the 70ns one (which may have been put there in the past by someone trying to replace another defective chip) may not be fast enough for the board. I suggest replacing both of the removable chips with others rated 60ns or faster.

(edit) I did a similar thing with a card recently S3 Trio64V+ memory: FPM vs EDO

bjwil1991 wrote:

Don't forget to replace the clock battery/RTC Dallas clock since those machines would refuse to boot, even after "saving" the settings in the BIOS. My Packard Bell did that, and I put in a temporary solution until I put in a 2x AA battery holder soldered to the clock battery leads on the motherboard and it fires right up in a split second.

Yep, that's the problem with it, and the only one. Otherwise boots right up to "no operating system found". It's got a surprisingly modern BIOS for a Socket 5 system, the last Socket 5 I had (GGateway 2000 I rebuilt for a friend whose still using it to play DOS/Win9x games to this day) was more primitive BIOS wise than my 486 is.

Today I cleaned and dusted it out, powered it on, it works. It has 16MB of RAM and a Pentium 100 in it. It also has an IR Reciever on the front (First time I've ever seen one of those headers actually used before). The motherboard is an intel Chipset board with apparently Opti video. The modem must be a hardware modem, not a winmodem, given it's size, not sure what speed thouggh, could be 56K for all I know. The only thing not working is the O/S, and that's likely because of the failed Dallas Clock chip as mentioned (though it's a different brand than Dallas).

The HDD works, Data recovery was a breeze being as it was only a couple gigabytes in size (the original Seagate Medalist), pretty much an all original NEC with the only possible exception being the Modem.

The Micron monitor I got with it, I just finished a focus and voltage adjustment on the flyback - it has a nice access hole in the back of the monitor for it you just pry off with a screwdriver. Rather nice. My NEC MultiSync monitor has 2 holes drilled into the side so I can adjust like that on the outside as the flyback on that is located on the side. My current plan with my vintage stuff is keep the original PSU's and Monitors as long as I can, and learn some electronics repair in the process (already learning to make guitar pedals so I'm finding learning Monitors and Power Supplies a breeze).

It's also nice to use the Micron on the 486 and be able to use 800X600 reliably again, the graphics are sharp as a tack!

the_ultra_code wrote:

dionb wrote:

That's DRAM not VRAM. And half of it is socketed, so first thing to try is to remove those chips and see what that results in.

So yesterday, I finally got around to working on that PCI video card, and did what you recommended with the "DRAM" (I find that interesting, personally; still learning a lot about older tech).

Some people erroneously abbreviate video memory to "VRAM". On modern cards everything is some (G)DDR version of SDRAM, so it's obvious what people are talking about. But back in the day, specifically at the time that this old card of yours was made, there were far more types of memory used on video cards.

The cheapest was DRAM, either FP (Fast Page) or EDO (Extended Data Out) - basically the same stuff as system memory at the time. This could only support one operation at a time - so either writing to it from the system, or reading from it by the RAMDAC. That led to a performance bottleneck, so various other types of memory were introduced that were "dual-ported", i.e. they could be written and read out at the same time. The commonest was VRAM (Video RAM), which was used by most high-end designs in the early to mid 1990s. Others were WRAM (Window RAM) - similar to VRAM but supposedly faster and cheaper. Matrox was the only vendor to widely use this; MDRAM (Multibank DRAM) interleaved lots of small banks to increase bandwidth and approach dual-ported performance. Only Tseng used it in any quantities. SGRAM (Synchronous Graphics RAM) was a special version of SDRAM that was single-ported, but allowed two pages to be opened at once, so behaving much like dual ported VRAM/WRAM unless the data was on the same page. SGRAM had a very short period in the limelight (around 1998), but virtually everybody used it at the time. Finally some rare chipsets (Cirrus Logic Laguna and Chromatic mPact) used RDRAM (Direct Rambus DRAM), of the Caminogate-fame, to very little success.

Here's some general info:
https://en.wikipedia.org/wiki/Dynamic_random- … ry#Graphics_RAM

More in-depth stuff can be found in 1990s reviews and background articles.

Now, a question. I noticed with the bottom memory expansion, um, "bank"

That's not a "bank", it's a socket.

A bank is a virtual organization of memory lines, i.e. a concept, not something you can touch, see or observe cracks in.

that some cracks had started to form in the plastic itself, one on the side of the bank, effectively splitting both sides of the bank, seen in this shot (it's hard to see):

[...]

Is this something I should really be concerned with, or is this only a cosmetic issue?

If the issues occur whenever this socket is populated and not when the other one is, then yes, it's something to be concerned with. I'd be surprised though - plastic has no electric function (apart from insulating pins from each other), so a crack in the plastic is highly unlikely to be a problem on its own. However, it's indicative of rough treatment. That same bad treatment could have damaged something else.

But for now I agree with the suspicion that the main problem is that the 70ns memory can't keep up with the card and that it would work fine with a 60ns chip.

dionb wrote:

Some people erroneously abbreviate video memory to "VRAM". On modern cards everything is some (G)DDR version of SDRAM, so it's o […]
Show full quote

Some people erroneously abbreviate video memory to "VRAM". On modern cards everything is some (G)DDR version of SDRAM, so it's obvious what people are talking about. But back in the day, specifically at the time that this old card of yours was made, there were far more types of memory used on video cards.

The cheapest was DRAM, either FP (Fast Page) or EDO (Extended Data Out) - basically the same stuff as system memory at the time. This could only support one operation at a time - so either writing to it from the system, or reading from it by the RAMDAC. That led to a performance bottleneck, so various other types of memory were introduced that were "dual-ported", i.e. they could be written and read out at the same time. The commonest was VRAM (Video RAM), which was used by most high-end designs in the early to mid 1990s. Others were WRAM (Window RAM) - similar to VRAM but supposedly faster and cheaper. Matrox was the only vendor to widely use this; MDRAM (Multibank DRAM) interleaved lots of small banks to increase bandwidth and approach dual-ported performance. Only Tseng used it in any quantities. SGRAM (Synchronous Graphics RAM) was a special version of SDRAM that was single-ported, but allowed two pages to be opened at once, so behaving much like dual ported VRAM/WRAM unless the data was on the same page. SGRAM had a very short period in the limelight (around 1998), but virtually everybody used it at the time. Finally some rare chipsets (Cirrus Logic Laguna and Chromatic mPact) used RDRAM (Direct Rambus DRAM), of the Caminogate-fame, to very little success.

Here's some general info:
https://en.wikipedia.org/wiki/Dynamic_random- … ry#Graphics_RAM

More in-depth stuff can be found in 1990s reviews and background articles.

Now, a question. I noticed with the bottom memory expansion, um, "bank"

That's not a "bank", it's a socket.

A bank is a virtual organization of memory lines, i.e. a concept, not something you can touch, see or observe cracks in.

that some cracks had started to form in the plastic itself, one on the side of the bank, effectively splitting both sides of the bank, seen in this shot (it's hard to see):

[...]

Is this something I should really be concerned with, or is this only a cosmetic issue?

If the issues occur whenever this socket is populated and not when the other one is, then yes, it's something to be concerned with. I'd be surprised though - plastic has no electric function (apart from insulating pins from each other), so a crack in the plastic is highly unlikely to be a problem on its own. However, it's indicative of rough treatment. That same bad treatment could have damaged something else.

But for now I agree with the suspicion that the main problem is that the 70ns memory can't keep up with the card and that it would work fine with a 60ns chip.

Damn. Awesome history lesson! 🤣

That actually filled in a lot of gaps to my understanding of, *cough*, "video memory" back in the day. Thanks!

dionb wrote:

That's not a "bank", it's a socket.

That's the word that was on the tip of my tongue! Couldn't think of it when I was typing. 🤣

dionb wrote:

If the issues occur whenever this socket is populated and not when the other one is, then yes, it's something to be concerned with. I'd be surprised though - plastic has no electric function (apart from insulating pins from each other), so a crack in the plastic is highly unlikely to be a problem on its own. However, it's indicative of rough treatment. That same bad treatment could have damaged something else.

But for now I agree with the suspicion that the main problem is that the 70ns memory can't keep up with the card and that it would work fine with a 60ns chip.

Okay, that's good. Lines up with what stamasd said:

stamasd wrote:

That -7 chip may not even be defective but just too slow for the board. All of the other chips are -6 i.e. 60ns, but the 70ns one (which may have been put there in the past by someone trying to replace another defective chip) may not be fast enough for the board. I suggest replacing both of the removable chips with others rated 60ns or faster.

(edit) I did a similar thing with a card recently viewtopic.php?f=63&t=62096&p=694590

Thank you both guys. Yeah, when I am not broke, I'll be sure to get some new memory chips for the card. Feels incomplete without them.

Last night I streamed The Magic School Bus Explores the Ocean on Twitch. I cheated and used 86box.

EDIT: https://www.youtube.com/watch?v=alDJGJdTPnI

Tried to fix 2 of my Quadro FX 3450 (Geforce 6800 basically) cards that artifacted, using the oven method. One works, one still artifacts badly, which I'm assuming is a VRAM issue since I tried 3 times to fix it (if anyone knows another fix for these, I would like to SLI them one day).

oeuvre wrote:

https://i.imgur.com/yUHDrbt.gif […]
Show full quote

Great! You won an Intel i386 DX CPU!