Sprite_tm

2026-08-31 02:30:01

Sprite_tm

sprite_tm@social.spritesmods.com

Hi all! A headsup: in the next week or so I'm gonna try moving from this account to @sprite_tm . I'll try to migrate everyone over automagically but I'm not sure how well that works; feel free to manually follow me on the new account if you want to make sure we're connected.

Details, for those who are interested: I've been running this account on a private Hubzilla instance for like forever now. Hubzilla never really was a good fit for me: it's a multi-user multi-channel beast that supports a whole slew of network protocols, while all I need is a single-user instance and ActivityPub. Migrating away from it is hard, though, since it doesn't really support the standard ActivityPub/Mastodon migration options. I've looked at GoToSocial as a nice and simple alternative, but since they moved to running ffmpeg in wasm, they need the CPU to support an instruction set my ancient machine does not have. Then recently snac2 (#^https://codeberg.org/grunfink/snac2) came to my attention, and I like it: minimalistic, written in C (which is an... interesting choice for a web application, but hey, I know C a million times better than I know Go) and pretty Unix-y. So I decided to bite the bullet, hack together some migration scripts and move.

Anyway, wish me luck and hopefully I'll see you on the other side!

2026-08-31 02:48:34

morgan

morgan@leds.social

@sprite_tm@social.spritesmods.com @sprite_tm@fedi.spritesmods.com cool! interesting in seeing how that works out. I'm in the midst of trying to setup a single user instance with mastodon and it's..... complex. And I'd far rather suffer C than suffer Ruby.

2026-07-24 15:06:07

Sprite_tm

sprite_tm@social.spritesmods.com

Ugh yakshaving. It started with 'I want to print out a case for one of my 486 mainboard things'. I'm now at the point of 'why is my desk full of crap and why am I researching readout protection of GD32 chips?'

2026-07-24 15:40:40

Sprite_tm

sprite_tm@social.spritesmods.com

Ah, small blessings. I found my FT2232 JTAG board thing, so I at least don't need to go even further by figuring out some way to make an ESP32 do SWD.

2026-07-19 08:25:35

Sprite_tm

sprite_tm@social.spritesmods.com

After writing my retro 486 / VapourDeck article, I thought I'd be done with it for a bit... and then I remembered I (re)started the project originally to make something to put underneath the cute li'l 9" CRT I found. I can probably use one of my old prototypes for that: aside from a few bodges and the fact that they don't run off LiIons and don't have the LCD interface, they're perfectly good li'l PCs. All I am missing is a case to put them in, so I decided to fire up yer olde OpenSCAD to design that. 3mm thick walls with lots of ribs, hopefully the PLA is able to hold the monitor that way.

(Alt: OpenSCAD view of a little desktop PC case, 20x20x7cm in size. It's open, with the top removed so you can see the mainboard inside. The front has vent holes as well as cutouts for a power, turbo and reset button as well as a LED-screen indicating the MHz the CPU runs at.)

2026-07-19 18:17:32

Chip

bytex64@awesome.garden

@sprite_tm Love the front panel accents.

2026-07-05 11:18:54

Sprite_tm

sprite_tm@social.spritesmods.com

I've posted an article about the VapourDeck I built on my site: #^https://spritesmods.com/?art=vapourdeck

For the people who didn't see the Hackaday article written about the Supercon presentation I did on this subject earlier: it's a built-from-scratch handheld running DOS games on an actual AM486DX5-133 using a period-accurate VGA controller and a FPGA/ESP32 for everything else. Probably one of my more involved projects up to now.

2026-04-20 04:43:21

Sprite_tm

sprite_tm@social.spritesmods.com

Seems the design is mostly functional. I manager tot swap scl andere sda on the flat panel connector, which I'll bodge on the other PCB. Als one of the feedback resistors for one of the voltage converters was wrong. Biggest oopsie up til now: the headphone jack footprint is inverted, so it needs to be soldered on the other side...

(Alt: PCB with a Lattice FPGA, ESP32S3 module, C&T VGA chip, AMD 486dx5-133 plus various memories etc. One USBC connector, two USBA ones, micro-sd card, VGA connector.)

2026-04-11 13:02:36

Sprite_tm

sprite_tm@social.spritesmods.com

Always scary to send a big and complicated and expensive PCB design off to be fabbed and partially assembled... did my best to make sure there are no oopsies, but the best I'm hoping for is that the amount of bodge wires required will be limited.

(Alt: PCB artwork in Kicad. Pretty complicated design: large PQFPs, a BGA, TSOP packages...)

2026-04-11 13:16:59

feuerrot

feuerrot@chaos.social

@sprite_tm will it run doom?

2026-01-11 02:26:44

Sprite_tm

sprite_tm@social.spritesmods.com

PSA: if your LCDs look like this, yyou can just buy new back reflector polarizer film and make them look just like new.

Just make sure you get the orientation right by overlaying the LCD (with peeled-off back) and seeing which angle gives you the lightest image. These were oriented at 45 degrees...

2026-01-11 02:40:06

Sprite_tm

sprite_tm@social.spritesmods.com

By the way, this issue is known as 'vinegar syndrome', as it's caused by the glue that adheres the polarizer breaking down into acetic acid. Can confirm that the LCDs smelled highly of vinegar when I pulled off the polarizer.

2026-01-06 12:24:34

Sprite_tm

sprite_tm@social.spritesmods.com

Always nice to finally get those PCBs that took so much time to design in your hands...

Now all I need is time to put the components on them... And develop gateware for them... And develop firmware for them... And test them...

2025-12-03 05:25:40

Sprite_tm

sprite_tm@social.spritesmods.com

So, last update on the little CRT. I kinda didn't do it on the fly as it's all kinda... unfulfilling, even if the CRT now works succesfully.

Last time I posted, I was having the issue that the horizontal sync didn't want to lock in; the trimmer for the horizontal lock settings seemed to not have any place where it would lock on, and a range that was way too wide. I traced it back to around the horizontal sync chip. I initially said it wasn't anything in the splash zone of the capacitor glue, but desoldered the resistors there anyway, and replaced the one that was 1K: I happen to have 1K through-hole resistors and I seemed to measure an off value for that resistor once (and only once). Faffed around a bit more as that didn't seem to fix it, measured voltages etc, until I touched the horizontal lock potentiometer again... now it did have the proper range and I managed to make it lock in.

Super-unfulfilling as I have no idea what changed it: could be the 1K, could be me resoldering joints, could be some sort of short in some component burning through or a cap reforming, no clue.

2025-12-03 05:31:10

Sprite_tm

sprite_tm@social.spritesmods.com

After that, the monitor worked great, and I played around with it a bunch... until I noticed that the color somehow changed without me noticing: the red channel was gone, and somehow green now also controlled the red (making everything green be yellow instead). Me faultfinding again... First of all, reflowed all the usual spots, to no avail. VGA pre-amp seemed OK (and the fault stayed with red if I swapped the red and blue pre-amp outputs), so I reverse engineered the cascode after that that drives the actual CRT. Couldn't find an obvious fault there, so I decided to swap around drive transistors, with the idea being that if I swapped a transistor and the issue swapped as well, that must've been the part. And after swapping the main drive transistors... the problems just went away? I needed to re-calibrate the outputs a fair bit, so perhaps the red transistors just went so much out of spec that the added error made the thing stopped working and swapping them around generated a combo that did work OK? No clue; I've marked down the types of transistors for if this thing ever goes wonky again.

2025-12-03 05:36:59

Sprite_tm

sprite_tm@social.spritesmods.com

Anyway, the thing is together and works pretty well. No sign of burn-in, color accuracy seems OK (although I could probably calibrate it a bit better), the tube doesn't seem to be too tired. CRT seems to be happy to do 1024x786, although the image is a tad fuzzy at that resolution. I'm probably gonna use it for DOS games anyway, so that doesn't really matter that much.

2025-11-26 08:58:22

Sprite_tm

sprite_tm@social.spritesmods.com

Let's trace the HSync signal as it comes in. The VGA connector terminates at the neckboard, but a second set of wires takes the syncs down to the main PCB. First it goes into a WT8041 chip; it has no info but in an old post on Twitter, @Foone🏳️‍⚧️ mentioned it's likely similar to the WT8043. This is an interesting chip: it detects the video mode so other bits of hardware can adjust voltages etc to match it, and it also conditions the sync signals before sending them on. This chip seems OK; the HSync/VSync that comes out looks as good as the incoming signals.

2025-11-26 09:08:33

Sprite_tm

sprite_tm@social.spritesmods.com

The output is taken via a transistor as a buffer, and that brings us back to the LA7851chip. It goes into some weird analog stuff before being applied to the IC, lemme reverse engineer that first to see if it matches the schematic.

2025-11-26 09:24:33

Sprite_tm

sprite_tm@social.spritesmods.com

It's deja vu all over again! The trail leads to something that is also connected to a few components in the splash zone of the capacitor glue. I bet one of them is broken. I don't really want to have to desolder and check them, they're not easily reachable, but if I have to...

2025-11-26 10:01:57

Sprite_tm

sprite_tm@social.spritesmods.com

And nope, it was not the resistors in the splash zone, the all measure what the datasheet says. Of note: there is a H-hold trimmer, but surprisingly it was dialed all the way to one side, and dialing it away from that makes the horizontal go away entirely. Wondering if that's a design thing or if that has something to do with my issue...