Superb. Good work so far woppy!
You may begin to reach the limits of the motherboard in terms of max Base HTT so if you experience instability when overclocking further, I would start looking into lowering the Base HTT and increasing the CPU Multi. Do you remember around which speeds you had to start increasing voltages from it's default 1.325V? I would suspect that around 3.8GHz will be your Phenom II 940's sweetspot in terms of clockspeed vs voltage required if you're already up to around 1.5V indicated.
You've got two options really. Keep ramping up the clockspeeds either by means of CPU HTT and/or CPU Multi and see where it takes you.
Or, work towards verifying stability at around 3.8GHz and then tweaking the existing settings while maintaining that CPU frequency to attain better performance.
I would try the first but not to spend too much time or dropping too much voltage through the CPU because in my experience with Phenom II, the second is probably your best bet for 24/7 every day usage.
Things to tweak
CPU Northbridge Frequency - This is the frequency at which the internal memory controller is operating act. Increasing this rakes in some nice performance increases, such that it'll decrease the "clock for clock" gap between Phenom II and Core 2 Quad gap quite considerably. There's two ways of increasing this and it's either by means of raising the base CPU HTT, which you have been doing or increasing the Northbridge Multi (This may be called something else in your BIOS).
At the moment, this is how it works by default.
CPU Northbridge Frequency = Base HTT * 8x
Base HTT is 200MHz out of the box so by default, Northbridge frequency is
1600MHz.
Right now, it's
255MHz * 8x =
2040MHz indicated in CPU-Z's Cache tab. I would start by raising that existing 8x Multi to 9x and then 10x and see how performance improves. SuperPi can be good to observe that although the improvements seen within said application probably won't necessarily be the same in other applications. I think 10x or 11x with your existing CPU HTT of 255 is your best bet so long as it's stable.
DDR2 Frequency and Timings - This is an interesting one. The K10 platform reacts to both Frequency and Timings (but often it likes tighter timings more) so what you need to do is have a play with the memory dividers and test higher ddr2 frequencies with loose timings vs lower ddr2 frequencies vs tighter timings and see what gives you the best performance.
Before you can reliably make not of performance improvements from given tweaks, I think it's paramount that you have a dual channel kit running in your system. An overclock that is stable with RAM in single channel mode may not necessarily be stable in dual channel.
Hope that helps.