The dynamic range of an image depends completely on the sensor. Every sensor has limitations on how much light it can take before the place go completely white on the picture and also it is limited by a minimal amount of light that has to be present so the place is not totaly black. You can see huge differences between camera models varying in the range that there is between completely black places and completely white places.
HDR means composing an image where both the light and dark places are well-exposed. Here, it is important to understand that human eye is an incredible instrument and no camera is even near that. Your eye takes (without you realizing it) images in many different exposure settings quickly after each other and in your brain, those images are composed to one.
Using a camera, you have to do exactly the same (but you have plenty of time taking the pictures) - take at least one picture with good exposure on light places (where most probably other parts of the image are in the blacks) and one with exposure set to uncover the dark places (where the lights are already burned over and totaly white) and one in the middle. Then what HDR applications (like Photomattix) do, is taking the well exposed parts form all the pictures and putting them in one well exposed image.
So back to your question - no, lenses are not going to do it. With the camera you have, it was enough for me in some cases to take one picture that was exposed normally and I could create +2EV and -2EV images in post production. That is of course not the best way, but in most cases your camera model has enough dynamic range to be able to fake it this way (actually the chip is so good in this model that in RAW, you can easily go from -3EV to +3EV in one picture and it looks fine using a low ISO). So taking three different pictures with 2EV steps, as you say, must give briliant results.
Possible reason of not getting what you need:
Do NEVER use JPG for HDR resources as RAW saves a lot lot more information and you need as many of it as possible. Always save pictures in RAW format for that purpose.
Some scenes are just not fitted to become HDR. Simply because there is not enough dynamic range in the scene - and cloudy weather is typical for that. That is why cities in the night are so common HDR scenes.
just taking three images and "shrinking" their dynamic ranges may technically produce HDR image, but what you see as wallpapers is a lot more - you need to decide which part of the range will be shrinked how to produce visually attractive output. This usually is not done by automated processors like Photomattix, but hard manual work including Photoshop, many layers and some sleepless nights.
Good luck :)
Jakub