I got a Conroe-L ES 2.8GHz 1MB a couple months ago and what I noticed that it clocks very well one some boards and on other boards not good at all. I have no idea if thats a general problem for the Conroe Celerons or if the ES is troublesome.

On a Asus P5B-E Plus (same as P5B Deluxe) I hit a FSB wall at exaclty 266MHz on all multipliers from 6 to 14. Furthermore, voltage was wrong all the time. Setting something like 1.6V in bios resulted in 1.38V measured. On a Abit Quad GT I could not boot using that CPU for whatever reason.
Then I tried a Gigabyte 680i board and its behavior was completely different. FSB wall at 410Mhz and completely stable vcore. It was rockstable for running WCG (folding app) at 1.4375V set in bios with 3550MHz. Max I tested on air was 1m stable at 3910Mhz with 1.575V set in bios.
Unfortunately I only had the 680i board for a few days. After that I wanted to try to test the CPU on watercooling on a Asus P5W DH Deluxe and guess what, it was making trouble again. Vcore was very very unstable. Setting 1.45V in bios resulted in a Vcore between 1.3-1.5V. With other CPUs on the same board, its stable within +-0.01V. It overclocked like crap, which I think is because of the fluctuating Vcore. I will test the CPU on P35 as soon as I have a board.

Maybe Intel fixed some of those issues, but could not complete fix them, which might explain different behavior on different chipsets with the Celerons.