The 7D would give you more external controls and a better viewfinder compared to your 5D. It's also got a larger buffer.
Your 5D can actually write to media faster though. The 5D can buffer 5 photos shooting in raw, and can write to media at 1 raw frame per second after the buffer is full, using a fast card like a Sandisk Extreme III.
You can find some performance tests for both the 5D and 7D here:
But, the AF speed shouldn't be any different. AFAIK, the AF sensor assembly is the same.
According to tests performed by PopPhoto, the KM 5D's AF system was able to focus faster and in lower light (focuses in -1 EV light), compared to the other entry level DSLR models they tested (comparing the KM 5D, Canon Rebel XT, Olympus Evolt E-500 and Nikon D50).
They said this about it:
The 9-zone AF system is the fastest of this group and works down to EV –1, very dim light indeed.
They clocked it as follows in various light levels. It's my understanding that a 50mm f/1.4 is the lens most commonly used by them to test DSLR AF performance.
Konica Minolta Maxxum 5D AF Speed: EV 12: 0.32 sec; EV 10: 0.33 sec; EV 8: 0.41 sec; EV 6: 0.57 sec; EV 4: 0.59 sec; EV 2: 0.86 sec; EV 1: 0.87 sec; EV 0: 0.93 sec; EV -1: 1.85 sec.
The closest competitor in the Autofocus Area was the Nikon D50. The other models were left a bit behind, especially in very low light (even the Rebel XT couldn't lock focus in light lower than 0 EV, and it took over 2 seconds to lock focus there). Your 5D focuses faster in only half that much light (EV 0 is twice as bright as EV -1).
There can be a significant difference in AF speed, depending on the lens, lighting and subject. If you use a dim lens, the camera's AF sensors won't be able to see as well to focus. For example, a lens like the Konica Minolta 18-70mm DT kit lens is down to f/5.6 by the time you're at 35mm with it. In contrast, a lens like the KM 28-75mm f/2.8 can maintain a constant f/2.8 aperture throughout it's focal range.
Since an AF lens always focuses at the largest available aperture, the AF sensors would get 4 times as much light using a lens like the 28-75mm f/2.8 versus the 18-70mm DT kit lens by the time you zoom in muich (f/2.8 is 4 times as bright as f/5.6).
In addition to optical quality (which can impact AF) and brightness (which will impact AF), you also see difference in gearing between lenses. Some lenses may require more revolutions to make changes in focus distance compared to others. This impacts AF speed. For example, most macro lenses focus slower, because they're geared towards finer focus adjustments (as well as being able to focus closer, which means they are designed for a larger range of close to far).
If you're getting soft focus, it could be your lenses, technique or camera body (out of calibration). Out of calibration bodies are not that common with the 5D. But, it does happen.
A common technique for checking AF accuracy is to use a good prime (for example 50mm f/1.7), and place some DVD cases or books staggered closely together at different distances from the camera. Then, focus on the one in the middle and make sure it's sharpest. If the one immediately behind it is sharper, you've got backfocus. If the one immediately in front of it is sharper, you've got front focus. Manufacturers can calibrate a body that's not focusing correctly. You can also find some focus calibration test targets along with instructions for checking AF.
Lenses can also have a problem (not reporting correct information to the camera body that it needs to make judgements on how far to turn the focus motor for final focus adjustment). This is relatively uncommon with Minolta lenses, but you see some 3rd party lenses with compatibility issues from time to time.
But, chances are, it's something else. For example, leaning after locking focus, subject moving after you've locked focus, focusing on wrong part of subject when you have a shallow DOF (for example, a person's body versus their eyes), trying to use center focus and reframing which can cause backfocus in some conditions, using a lens that's soft at wider apertures (most lenses are sharper 2 or 3 stops down from wide open and a soft lens can make it harder for the AF sensors to see), etc.
Here is an article explaining why recomposing can cause focus errors:
You may want to post some samples of problem photos in the KM DSLR forum, and see if members can comment on what could be going wrong.
As a general rule, I always select the focus point closest to my subject's eyes to minimize reframing, and I've gotten excellent results this way with decent lenses, even in ridiculously low light.
In one of the local restaurants here with live music, light is so low I need to underexpose 1/3 stop just to get shutter speeds up to 1/8 second shooting with my Minolta 100mm f/2 wide open at f/2 and ISO 3200, and I can still lock focus using an outside focus point (albeit with some difficulty).
If you're bound and determined to go another direction and want good AF and shooting performance, the Canon EOS-30D would probably be your best bet out of the other models you're looking at. It's got a good AF system, decent size buffer, fast frame rate, etc. But, your lens selection will be just as important (if not more important) for any model you consider, especially in less than optimum lighting.