...

There´s also a nice riser card for ISA/PCI cards

When I got this board it wouldn`t start due to some damaged traces near the CPU socket that were easy to repair with thin wires

The latest BIOS - 1.07 - can still be obtained from the NCR website
http://www5.ncr.com/support/support_drivers_p … pc_library_3348

Floppy drives are NOT recognized by this mainboard when they are connected to the second connector of the floppy cable (the one behind the wire-crossing). Floppys need to be connected to the first connector instead.

The board offers 50, 60 and 66 MHz FSB and multiplicators x2 and x1.5.
Jumper JP2 switches between core voltages 3.3V and 3.45V.
An additional VRM socket could also be soldered in to provide 2.8V for Pentium MMX but the BIOS doesn´t know any MMX models and given the multiplicator limitations I don´t thik it would make any sense.

Here´s the "feature connector" that can be enabled in the BIOS menu, it´s not populated in later revisions of this board.

t.b.c.

There´s some progress here. This board has a Socket 5 CPU socket, but is prepared for the connector for an additional 2.8V VRM. I suspected it´s in fact capable of split voltage AKA Socket 7.
So I fitted the 2x15 pins for the VRM in and cut the traces between pin pair 6 and 7, put one of my VRMs on top and a Pentium MMX 233MHz in the socket:

Then I set the multiplier dip-switch to x1.5 and started the system. To my disappointment the MMX was running at 133 MHz, it didn´t respond to the 1.5 multi by interpreting it as x3.5.
Checked BF1 -> open,
checked BF2 -> open,
checked BF0 (formerly BF) -> open.
On this mainboard the dip-switch releases BF0 from ground but there´s no pull-up resistor present to lift the BF0 pin up to Vcc. So I added a 4K7 resistor and voila

The next - and bigger - problem is the COAST cache module. FIC / AT&T obviously have a non standard pinout here that prevents regular COAST modules from being used.
I tested about 25 different COAST modules, all of them are either not detected or prevent the system from starting.

majestyk wrote on 2021-02-21, 11:15:

[...]

The next - and bigger - problem is the COAST cache module. FIC / AT&T obviously have a non standard pinout here that prevents regular COAST modules from being used.
I tested about 25 different COAST modules, all of them are either not detected or prevent the system from starting.

I don't think that pinout is the problem, rather the cache type. This board contains a VLSI 82C590 PowerCore chipset, which is pre-i430FX, so almost certainly also pre-PLB cache. That means it would need an async SRAM cache module. I've not been able to find a datasheet to confirm this assumption, but given you've thrown probably about every PLB config in existence against it, it seems more than likely.

This might well be the case, or both.
With PLB modules the machine won´t start at all.

Then I have a lot of "sync cache" modules like this:

Withe these the system starts bot no cache is being detected.

I couldn´t find any datasheet of the VLSI MCH so far.

I then compared the COAST pinout of this mainboard with this one:
download/file.php?id=82702&mode=view

All Vss connections are identical. Same goes for the "Vcc5" voltage. The connections for "Vcc3" differ. On the FIC socket Vdd3 is at pins 76, 68, 60, 52, 39, 25, and 15. It´s NOT at Pin7 as it should be according to the pinout documentation. Pin 7 leads to a pin of the MCH (VL82C591FC3) instead. Pin 6 at the COAST socket is NOT "NC" but also in use.
On all COAST sticks I own, the Vdd3 - Pins 76, 68,60, 52,39, 25,15 AND 7 are bridged.
So VDD3 (3.3V) is probably connected to some Data line when a standard COAST module is inserted.