Was it the addition of the peltier that has made 504 at 2.2 possible?  The boot-up temperature is certainly lower.  With my non TEC dual-fan heat sink, the boot-up temperature of the processor would exceed the room temperature by 10 to 15°F before Rain or another cooler program kicked in.  With the peltier, room temperature is exceeded by only 1 to 2° on boot-up.  However, while stressing the processor, the high temperatures I have measured on the slug are not that much lower than using a large dual-fan heat sink.  I hope to lower the high-end temps by designing a larger heat sink for the peltier.

It is probable that the temperatures I read from the outside of the slug do not fully indicate what is actually going on inside the chip.  I would surmise that just as it takes a certain amount of time for heat to travel from the base of a heat sink to the ends of its fins, it takes time for the heat generated inside the chip to migrate to the outside of the slug where I can measure it.  This means that I am measuring an average of the temperatures reached in the chip, not its individual "heat spikes" when it's working hard.  It seems that lowering the temperature of the slug a few degrees was all that my Celeron needed to clock to 504.  I hope to be able to try this on a few more Celerons and see what happens.  Just because mine finally made the jump doesn't guarantee that this will hold true for others.  I must say that it looks promising, though.

This train of thought leads me to wonder if the difference between those Celerons that will hit 504 out of the box at default voltage and those that don't has something to do with the assembly of the chip to the slug.  Less air-space between the chip and the slug would make for quicker heat transfer.  This, in turn would make it easier to cool the circuitry with a good heat sink.  Could this be one of the variables separating the "goes" and the "no goes."?

I have also considered that, just as many people have noted, "burn-in" seems to play some part in getting the Celeron stable at a lower voltage.  Did running the Cele up to 2.6 volts and the high temperatures that the voltage produced have an effect on the final ability of the chip to become stable at a voltage it previously wouldn't even post at?  I'd have to test a few more chips before I'd sign my name to that idea.  However, it was one of the phases the chip went through on the road to reaching 504.

My last idea has nothing to do with voltage, temperature, or understanding the processor's properties.  It just might be that the god of Celerons finally smiled on me and made the magic happen.

Test System
Celeron 300A o/c'd 4.5X112
Abit BH6 HN BIOS
Hitachi PC100 SDRAM (64 MB)
Matrox Millennium G200 Video (8 MB SGRAM)
DCS S805 A3D Sound Card
Quantum Fireball SE 3.2
Quantum Fireball ST 3.2
Diamond FirePort 40 SCSI Card
Plextor PlexWriter 4-12 CD-R
Plextor UltraPlex 32 CD
D-Link DE-528 Ethernet Card
Hayes 5675 56K v.90 / Flex Modem
Inwin A500 Case w/ cooling modifications