It hangs after step 04
Source: http://mrbios.com/techsupport/award/postcodes.htm

I found out what killed the 286. A floating protective earth. Same issue as in the kitchen

Lovely

Next thing on the bench: another 286 board.

I've already replaced a bunch of 74F244 buffers and 74F245 bi-directional bus transcievers because they had wonky looking outputs or even inputs

After replacing the 244, its input pins looked good again.

TTL logic levels apply, so <= 0.8V for low, >= 2V for high. In-between is considered undefined.

I love having a DSO. Triggering on these faults is so easy

The situation as of now is that it complains about missing RAM, if that is removed, and initializes the gfx card, but to this.
Post codes are, in reverse order: 30 25 24 23 22 21 20 1a 19 18 17 16 15 12 11 10 0c 0b 0a 09 08 07 06 05 04 14 03 02 01
It hangs after 30. BIOS is an early AMI from minuszerodegrees meant for the 5170.

After 30, Address line 20 goes low and never high again.

The fact that the A20 line is low and that it hangs around protected mode tests makes me suspect something in relation to that.

All 245 and 244 are suspect due to those runts. However, if they are semi-defective (the few I pulled check out fine in the minipro), then it should randomly glitch and not consistently hang at the same spot. I assume that a very specific control or address line is broken somewhere, or a gate in that line is completely dead.

This gate I probed is connected on the A side to the MD9 pin of the 88C215, which is a data and address buffer between the CPU and peripheral address bus. It is part of the chipset. As well as directly to the memory. The B side goes to the Data 9 pin of the ISA bus in the 16bit extension. A fault on this data pin could produce a misinitialized graphics card. However, the glitchy image persists if I move the graphics card over to an 8bit slot (afaik it is capable of that)

The chipset is branded 'SOLUTIONS', 88C212, 88C215, 88C211 and a Siemens branded 82C206. I think it is identical to the NEAT chipset, that seems to have been cloned/licensed by a number of different manufacturers, including Siemens and OPTi. Original developer is afaik Chips and Technologies.

Wikipedia says:
By design, the 286 could not revert from protected mode to the basic 8086-compatible real address mode ("real mode") without a hardware-initiated reset. In the PC/AT introduced in 1984, IBM added external circuitry, as well as specialized code in the ROM BIOS and the 8042 peripheral microcontroller to enable software to cause the reset, allowing real-mode reentry while retaining active memory and returning control to the program that initiated the reset. (The BIOS is necessarily involved because it obtains control directly whenever the CPU resets.) Though it worked correctly, the method imposed a huge performance penalty.

So naturally, I'm having a look at the reset lines now. Reset on the 286 never goes high again after the initial start. So if it does enter protected mode, it never leaves it.