Been fighting this one on and off for months and I'm about ready to call it. Looking for a sanity check or any last ideas before it goes on the shelf for good or I take a bat to it so I stop temping myself to go back and wasting days at a time.

Board: Shuttle HOT-403H, OPTi 486SXWB chipset (82C493 / 82C392 / 82C206), AMIBIOS, 486DX-33. Classic leaky-battery damage concentrated under the BIOS EPROM and 8042 keyboard controller area.

What I've already done:

CPU is known good, works in other boards
Eprom can be read with a tl-866
Replaced the 82C493 (old one was dead — spurious ~500kHz ATCLK with no clock input). New one generates proper ATCLK.
Original AV9107C-05 clock generator died (my fault, let the smoke out). Running a 66MHz oscillator module instead. Clock rebuild was a saga but it's now a solid, stable 33MHz at the CPU (verified — had a flaky crystal socket joint fooling me for weeks).
Turbo/FLUSH# config correct, TURBO pulled high.
Found and bodged a few broken BIOS data line. The bios sits next to the leaked battery. A few data lines needed to be ran back to the 🤣 bus
Confirmed all 8 EPROM data lines reach the chipset. Address lines traced through two series 74LS245 address buffers, all appear correct. I pulled the 245 for the lower 8 bits of the 🤣 bus to eprom and tested it in the tl-866, it's fine.
Reset clean at the CPU, no bus holds (AHOLD/HOLD/BOFF# all correct), power/ground solid across all chips (0.5-0.7 ohm).
Stitched several vias in the damage zone, reflowed and filled the corroded 8042 through-holes.

Current state: ADS# is dead at the CPU, no sustained bus cycles, EPROM address/data lines static. BUT it intermittently flashes brief signs of life (ADS# pulsing) with no reproducible trigger. Early on it favored being held vertical; now extreme flexing, localized heat, and freeze spray produce nothing consistent. EPROM reads fine in a programmer (sign-on message intact).

You power it up and everything that can be stuck high, it stuck high. All the 🤣 bus data lines, all the address lines to the eprom, the same across the ISA bus, and every address or data line on the 493/392. Occasionally it'll spring to life and begin pulsing for a brief moment then everything gets stuck high again.

Anyone revived a board this far gone, or is this the point where it's just not worth it? Happy to post scope shots or the specific pinout tracing if useful.