Kahenraz wrote on 2023-12-09, 08:14:

Popped the heatsink off of a TNT2 Ultra today. The fan was bad and the heatsink itself was glued on. Luckily, it was an adhesive […]
Show full quote

Popped the heatsink off of a TNT2 Ultra today. The fan was bad and the heatsink itself was glued on. Luckily, it was an adhesive that could be removed with acetone. The labeling came up with it though.

I forgot to take a photo of it after removal, so this is the photo after the adhesive has been cleaned off.

This card has a bad memory chip, so I have a 1/8 chance of fixing the right one each time I try to swap one out. This is going to be a process. I have two TNT2 donor boards. Spare chips will be used on this Voodoo 3, which I suspect also has bad memory but has only a blank output at POST (although it is recognized by Windows).

20231209_031022.jpg

20231209_031132.jpg

20231209_031615.jpg

20231209_034605.jpg

It took me a while to figure out what I was looking at with the TNT2 Ultra text being backwards 😀

You can improve those chances drastically if you use VMTCE! The memory is laid out in 128-bits with 4x32-bit chips and two banks (front & back). Have a read of my thread on this: Accurately troubleshooting video memory faults with VMTCE
I can't find the pinout of the TNT2 so I can't say whether chip 0 is in the upper left or the bottom right, but if you lift one pin on an SGRAM chip - either pin 1 or pin 100, and post the result with VMTCE then I can tell you how it's laid out

That way, it should be possible to find out which RAM chip is bad before swapping any of them out.

That Voodoo 3 not displaying could well be a VGA ROM fault, try reprogramming that first before swapping out memory.

Shponglefan wrote on 2023-12-09, 02:52:

BitWrangler wrote on 2023-12-08, 23:17:

Yah we got dudes that think four isn't even a full computer worth

When the game asks you to choose your sound device, best be prepared... 😉

Hahah it's meant to be an options screen, not a checklist 😜

Thermalwrong wrote on 2023-12-09, 12:53:

I can't find the pinout of the TNT2 so I can't say whether chip 0 is in the upper left or the bottom right, but if you lift one pin on an SGRAM chip - either pin 1 or pin 100, and post the result with VMTCE then I can tell you how it's laid out.

The inside chips or the outside chips? My TNT2 Pros only have 16MB (4 chips) and start on the outside. My TNT2 Ultra has another set on the inside.

VMTCE immediately starts throwing errors with my TNT2 Ultra. How would that change if I lift a pin?

Thermalwrong wrote on 2023-12-09, 12:53:

That Voodoo 3 not displaying could well be a VGA ROM fault, try reprogramming that first before swapping out memory.

I did try to reflash the BIOS first. I'm pretty sure the memory that holds the framebuffer is corrupt.

See here:

Re: Seeking advice on reviving this Voodoo 3

Do you happen to know which chip might be the bad one on the Voodoo 3? I'm going blind in this one, because there is no video at all. I suspect that it's whichever chip holds the framebuffer.

Kahenraz wrote on 2023-12-09, 16:27:

Thermalwrong wrote on 2023-12-09, 12:53:

I can't find the pinout of the TNT2 so I can't say whether chip 0 is in the upper left or the bottom right, but if you lift one pin on an SGRAM chip - either pin 1 or pin 100, and post the result with VMTCE then I can tell you how it's laid out.

The inside chips or the outside chips? My TNT2 Pros only have 16MB (4 chips) and start on the outside. My TNT2 Ultra has another set on the inside.

VMTCE immediately starts throwing errors with my TNT2 Ultra. How would that change if I lift a pin?

Uh, I'm not sure what outside and inside are. How about we call them front and back with the front being where the GPU is?
I haven't done much with Nvidia cards yet but bank0 on my 3dfx cards is on the front.
The reason I suggest to create an error is so that we know where the error is - i.e. lifting a pin on a chip in the lower-right and logging what changes before and after the chip leg was lifted. Then we can see which address area it affects and work out which chip is which on that card.
For example lifting DQ3 on the memory chip on the front of the card in the bottom right - that should cause an error ending with "C" if it's laid out like I think it is with the last bits of the 128-bit data bus being in the lower right of the card - that's how it is on 3Dfx cards but I have no datasheet for the nvidia TNT2 so can't confirm. I don't have any SGRAM TNT cards either.

I did try to reflash the BIOS first. I'm pretty sure the memory that holds the framebuffer is corrupt. […]
Show full quote

I did try to reflash the BIOS first. I'm pretty sure the memory that holds the framebuffer is corrupt.

See here:

Re: Seeking advice on reviving this Voodoo 3

Do you happen to know which chip might be the bad one on the Voodoo 3? I'm going blind in this one, because there is no video at all. I suspect that it's whichever chip holds the framebuffer.

Great 😀 At least that's eliminated. VMTCE is a pretty dumb tool so only runs on the primary graphics card and I don't yet know how to log its output, that looks like the kind of situation where a serial terminal output or logging to a file would be handy.
That card looks like it's been through some stuff already, have you replaced the damaged capacitors?
That graphical mess in the last post looks like how my Voodoo Banshee looked when I tried resoldering the pins on the SGRAM chips and had a pin bridged, like one address pin bridged to another. Causes just complete garbage output. It could also be that on your card maybe an address line is not hooked up somewhere, that would also cause chaos. Troubleshooting data bits and data errors is somewhat easier.

The memory layout on the Voodoo 3, because of the 128-bit bus is shared across all the chips. So address bits are laid out like this:
Chip 1 (upper left) - bits 0 to 31, bits 128 to 159
Chip 2 (middle left) - 32 to 63, 160 to 191
Chip 3 (upper right) - 64 to 95, 192 to 223
Chip 4 (lower right) - 96 to 127, 224 to 255
And so on, all 4 memory chips are responsible for the first megabyte of memory in different places.

Today i replaced the soldered internal battery of my "new" 20 year old Alesis D4 with a CR2032 holder for easy maintanance in the future.

I also cleaned it and removed 20 years of stickers somebody put on it to lable the channels...
I wanted to upgrade the firmware to 1.04 but its already on 1.04 from the factory.
I totally forgot it needs a 9V AC power supply... Because i didnt get one i have to dig out my SR16 supply the next days to test it...

Got my Tauntek IC tester all built up and running! Got it hooked up to my 386 machine and talking to it via procommplus in dos. This is a really great alternative to the retro chip tester pro, and costs about 1/10th the price. It's more limited in what it can do, but for my use case it's awesome.

Kahenraz wrote on 2023-12-09, 16:27:

VMTCE immediately starts throwing errors with my TNT2 Ultra. How would that change if I lift a pin?

presumably the log file contains addresses and banks where the errors come from

Thermalwrong wrote on 2023-12-09, 01:22:

Forcing myself to finish up more projects since I'm packing things away and putting things away as parts means they're more like […]
Show full quote

Forcing myself to finish up more projects since I'm packing things away and putting things away as parts means they're more likely to get lost.

I'm frustrated both with my attempts to fix the Voodoo Banshee and the Soundblaster 2.0 - got the new AT89S51-24JU parts in to replace the DSP and it's *still* not working, so it's just an Adlib + Joystick card right now.

This Toshiba Satellite 4000CDS has been at 80% completion as a project, for a while. It's a DSTN laptop that's otherwise really nice with a built in floppy & CD-drive, and Yamaha OPL3-SA3 audio. It came to me with a screen that had actual rust on it and only the top half of that DSTN screen worked.
dstn-becomes-ips-1 (Custom).JPG
Getting a replacement screen has failed a couple times so I took a different route - jamming an iPad 3 2048x1536 IPS LED-backlit LCD panel into the screen housing which connects into the VGA connector.
It had sat on hold for ages because it needs the screen facia designed and printed, but today I decided to bodge it together to test which was very worth it - made a bunch of changes to how all the parts fit in the case. The controller had to be modified somewhat to fit the flat panel housing and the VGA cable was a special order, but it works 😀 The LCD backlight is powered by the inverter power cable and the screen controller + screen plug into the 5v pins for the DSTN LCD itself.
Being a more modern LCD, which itself is something like 10 years old - it's a lot more responsive than a period correct TFT so FPS games are somewhat more engaging with real 60hz refresh. Lots more contrast too:
dstn-becomes-ips-3.JPG
It works better than I had hoped even if half of it is blu-tack right now and the screen has little structural rigidity currently, it's getting put away in one piece now for finishing in future. I'm probably going to gut the DSTN LCD to use its frame rather than making some big CAD design, the metal frame will be stronger than 3d-printed plastic anyways.
dstn-becomes-ips-2.JPG

And today it got the housing made 😀

This is now together enough that it's going to get put away, or used since it's pretty nice - probably when the weather is nicer I'll put some grey spray paint onto the frame to make it match. Put some more hot-glue onto the frame to make it fit the screen bezel a little better.
Oh and a CMOS battery since the screen only works when it's in "simultaneous" LCD+VGA mode to be able to power the screen and feed it the correct refresh rate - it does not like 70hz. It needs a CMOS battery to save that setting, otherwise each time it's turned after being off for a while, there's no screen until the fn+f5 display switch is pressed a couple times.

This print is v1 and there are some mistakes although I made 2 cut-up test prints first to make sure things like the button measurements were accurate. It's actually too big for my Prusa MK3 to print in one piece so instead what I've done is used a broken 12.1" TFT panel, removed the glass & electronics, cut out the bottom part since that's where the electronics are. And made this 3d print to fit into the space where the glass went before, it's literally 1cm off the full y-axis space and uses 100% of the x-axis space on the print bed.

The only clue from the outside is this cable poking out the back:

I was about to start taking chips off of my TNT2 Ultra when I spotted this. I inspected the board thoroughly before working on it. This must have already been cracked and came loose as I was cleaning it in the sink in preparation for soldering.

I use a very delicate brush with light pressure, so there is no way that I could have inadvertently cracked and dislodged this capacitor on my own without it already having been loose.

Edit: No such luck. There are still memory errors.

Thermalwrong wrote on 2023-12-10, 02:05:

Good job! It looks nice. Well done!

Thanks, I'm gonna be moving soon and am putting stuff into storage so getting as many things done as I can before the move happens. Right now I'm stuck since I live 4 floors up but the lift is broken and everything's on hold til that's fixed.
I'm not 100% happy with this first print and probably should have used the front bezel as the part that holds the new screen in place, it might be nice to make another one of these in future with better Toshiba colour matching plastic.

I'm taking the shotgun approach to getting things done - working on the Voodoo Banshee again today and took another look at the Voodoo 3 I broke which you can read about this thread: Accurately troubleshooting video memory faults with VMTCE
I pulled off its RAM chip that I knew was the faulty point and figured the RAM chip can't be bad since I know that overtightening a heatsink that was nowhere near that area, caused the memory fault to start happening. I've always written it off as 'easy come, easy go' since this V3 3000 was found in a scrap lot and was already in rough shape.

Pulling the RAM off and having access to the full databook for the Voodoo 3 which lists its BGA pinout, as well as knowing exactly which data pin is not working lead to something ridiculous. I've just successfully reconnected a broken trace that's under a BGA chip without hot air or desoldering / reballing the chip, check this out - you'll need to zoom in:

I knew from counting the pins down from 63 (0 to 63) that the damaged trace was Memory data 58 - which just so happens to be on the very edge corner of the chip. That's got to be what I damaged when I screwed the heatsink down too far. Probing around with a needle stuck to my multimeter probe to get under the chip I could tell where the trace was and the ball was not hooked to it.
The via for that trace is under the chip and attempting to heat it didn't really do anything so looking at it from the side like this with the PCB backlit I decided to solder a bit of enamel wire to it and eventually it stuck (not well, it is loose still) well enough that I was able to run this enamel wire through the next nearest via at the edge of the chip and hooked it up at the back.
There's only enamel protecting Memory data 58 from whatever hole it's going through but I'm gonna put some soldermask on the back to hold it in place.

With that trace fixed, now video memory test passes and the card works in both 2D and 3D again. Do I think it's going to be a reliable repair? Nope, it's already stopped working after I cleaned up flux and I don't want to resolder it for fear of damaging the enamel, but it works for now and I'm happy with that.

Played around with my ECS 848P-A v2.0 board, found out ecs dropped all the bios and manual files off their website so uploaded what I had to archive org. A few years ago they had a lot of files but now only drivers 🙁
will up to our library after chores...

I finished repairing, and consequently doing maintenance on a CHUNON FZ-502 Rev. A 360KB 5.25" floppy drive. It is cosmetically a bit beat up but it works 100% and now I have two of these!

I swapped out the original external battery from my Samsung S550 286, replaced with a 4xAA battery pack, patched to 3xAA. Will likely replaced with CR2032 when I get some sockets.
Moved it from my XT-IDE CF card to a real hard disk - managed to break a pin on the EEPROM but flashed a new one to a EPROM I had around.
Also fitted a 10MB network card also holding a XT-IDE AT rom. Was intended to fit a 3C509B but accidently enabled the ROM at the wrong address and that disables the VGA card so swapped to DM8801F with the ROM until I figure out how to reset the 3C509B.
520MB HDD didn't want to play, so using a 1.2GB HDD.
Didn't get time to setup RetroNAS on a PI4 have floating around, hopefully will get to this evening.

A few days back I thought I blew an AT PSU, so I tried another PSU. Thankfully the short circuit protection kicked in. So after some inspection of the board, I saw this.

Some corrosion from 3x AA batteries (30+ years old) hanging above the mainboard.
After some cleaning and testing the values were good.... Luckily I saw this as well:

Yep, a shorting C between GND and +12V.
Let's replace this little capacitor with a non-SMD part and test again:

Happy days.

Upgraded Tiny Core Linux on my Pentium MMX build (dual-booted with Win95) to release 14.0 (kernel 6.1.2).

Works fine.

PD2JK wrote on 2023-12-11, 09:30:

A few days back I thought I blew an AT PSU, so I tried another PSU. Thankfully the short circuit protection kicked in. So after some inspection of the board, I saw this.

Why are there so many clock chips on that motherboard? I count no less than four from just that one angle.

Kahenraz wrote on 2023-12-11, 12:38:

PD2JK wrote on 2023-12-11, 09:30:

A few days back I thought I blew an AT PSU, so I tried another PSU. Thankfully the short circuit protection kicked in. So after some inspection of the board, I saw this.

Why are there so many clock chips on that motherboard? I count no less than four from just that one angle.

I don't know. Maybe it's a (WD) Paradise thing, one some cards here I count just as much.

PD2JK wrote on 2023-12-11, 13:08:

Kahenraz wrote on 2023-12-11, 12:38:

PD2JK wrote on 2023-12-11, 09:30:

A few days back I thought I blew an AT PSU, so I tried another PSU. Thankfully the short circuit protection kicked in. So after some inspection of the board, I saw this.

Why are there so many clock chips on that motherboard? I count no less than four from just that one angle.

I don't know. Maybe it's a (WD) Paradise thing, one some cards here I count just as much.

It should be the on-board video card, that should be how different refresh rates were done before PLL clock generators were a thing 😀 It's common to see a bunch of crystals on early VGA cards too, kinda weird to see it with a WD90C11 though

RandomStranger wrote on 2023-12-09, 18:49:

twiz11 wrote on 2023-12-08, 22:48:

RandomStranger wrote on 2023-12-08, 20:57:

A couple of days ago I started playing Need for Speed Underground 2. I definitely have to re-learn the hitboxes. I chose the Cor […]
Show full quote

A couple of days ago I started playing Need for Speed Underground 2.
I definitely have to re-learn the hitboxes. I chose the Corolla as my starter car. I remember late-game that being one of the best control car without sacrificing top speed and after some upgrades it definitely lives up to my memories. For now I have that and a stock Escalade. When I unlock my third garage slot, I think I'll get a rice mule. If I'd follow my heart, I'd use the Escalade for that, I have no respect for that car class and don't mind making them ugly(er), but I'm they drive like a brick and rather use them as little as possible.

how did you get nfs u2 working again, that game has been borked for years

I don't know what curse did you put on my game, but today it refuses to launch 🙁

It looks like the optical drive have failed in my nostalgia XP build. Today I got some new games and tried a surface test. All came back with something like 25% problem sectors so I tried a game I know to be completely flawless and it also came back with similar results. It was a DH-16AES. Replaced with a GH22NS40 I recently salvaged from a faulty PC at work which was sent to recycling and now it's OK.

Looks like I'll have to get some optical drives while they are still cheap.

PD2JK wrote on 2023-12-11, 13:08:

Kahenraz wrote on 2023-12-11, 12:38:

PD2JK wrote on 2023-12-11, 09:30:

A few days back I thought I blew an AT PSU, so I tried another PSU. Thankfully the short circuit protection kicked in. So after some inspection of the board, I saw this.

Why are there so many clock chips on that motherboard? I count no less than four from just that one angle.

I don't know. Maybe it's a (WD) Paradise thing, one some cards here I count just as much.

They even explain it at the bottom of the page. Very cool.

Paradise/Western Digital Crystals Most Paradise and Western Digital SVGA graphics cards come with 3 or 4 crystal oscillators, wh […]
Show full quote

Paradise/Western Digital Crystals
Most Paradise and Western Digital SVGA graphics cards come with 3 or 4 crystal oscillators, which are defined as follows:

MCLK is the memory clock and determines the DRAM read/write access timing as well as the system CPU I/O and memory timing. MCLK should be 36 MHz if 120ns DRAMs are installed, and 42 MHz if 100ns or faster DRAMs are installed. This crystal should be of a higher value than VCLK0 and VCLK1.
VCLK0 is the video display clock (aka, the dot clock) for both text and graphics modes. At 25.175 MHz, this displays 640 pixels per horizontal display line.
VCLK1 can either be a second video display clock input or an output to an external clock selection module. If used as an input, 28.322 MHz is used to display 720 pixels per horizontal display line.
VCLK2 (if a fourth crystal is present) can either be a third video display clock input, or an output to an external clock selection module. If present it is usually 42.000 MHz or 44.900 MHz - don't confuse VCLK2 with MCLK. If the card also has a 36 MHz crystal elsewhere, that one is likely to be your MCLK crystal, and the 42 or 44.9 MHz one would be VCLK2. Your card might have a jumper to enable this fourth crystal, i.e. VCLK2.

For 320 or 360 pixel modes, VCLK0, VCLK1 or VCLK2 is divided by 2 by the chipset to derive the pixel clock (PCLK).