As part of my effort to get a socket 5 working with an AMD K5-PR200, Motherboard for socket 5 build , I discovered that the 2x jumper of an Asus PCI/I-P54TP4 430FX board won't set an AMD K5 to 1.75 or 2.0x. The 2x jumper on this board is for CPU's which only have the BF signal. If the CPU has, both, BF0 and BF1 signals, the combination of values for the multiplier is different.

I thought I would make a post on how I modified the Asus PCI/I-P54TP4 so that the multipliers jumpers were still flexible.

For older P54 CPUs, e.g. P75-P100 and model 0 AMD K5 CPUs the table used is:
MUL BF connetion
1.5x 1 (float)
2.0 0 (GND)

For newer P54 CPUs, the table used is:
MUL BF1 BF0
2.0x 1 0
2.5x 0 0
3.0x 0 1
3.5x 1 1
whereby 0 connects directly to GND and 1 is left floating.

For model 1 & 2 AMD K5 CPUs, the table used is:
MUL BF1 BF0
1.50x 1 0
1.75x 0 0
2.0x 0 1
1.5x 1 1
whereby 0 connects directly to GND and 1 is left floating.

While most people probably don't have a K5 PR200, the modification works with PR166 at 1.75x. Without the modification, the PR166 runs at 100 MHz rather than 117 MHz.

To make the modification easily reversible, I found an unused header via which had one of the pads going to GND. This empty jumper pad was JP24. I picked a new jumper block from my bin and shaved the corner edges of the pin to be used for BF1. I then took some wire cladding which fits just snug into the motherboard's via and put it on the shaved jumper header pin. Then put the 2-pin jumper header into the via and soldered the GND to the motherboard's via and the cladded pin to BF2 of the CPU's socket pin using 30 AWG wire. Careful not to hold the iron on the cladded pin too long as the cladding will melt and contact between the via may occur. You'll want to check this with a multi-meter when finished soldering.

I tested the modification with an AMD K5-PR166 and PR200. Cyrix 6x86 runs at 2x doesn't need this modification. I haven't tested the Cyrix 6x86 at 1x on this board.

Photos of the process.