Strange pad problem

I just got a classic 2000 that needed repair for real cheap, so I figured I'd give it a shot. I got everything working on it except there's a weird problem triggering sounds. It seems to be linked to velocity of the pad strike... check out what's happening:

Take the first row for example (pads 1, 2, 3, & 4)

Tapping gently... really gently on pads 1, 2, and 4, you will get the correct sound from the program. Any harder than that and you will get whatever pad you're hitting and pad 3. This happens for all four rows of pads the same way but it's pad 7 in row 2, pad 11 in row 3, and pad 15 in row 4. I have tried another known good sensor matrix and the same thing happens.

Does anybody know what might be going on here? I have searched but almost every double triggering post is someone experiencing the normal finger bouncing phenomenon. I might have a better chance of figuring out what the problem is with a schematic. I can't seem to find that either. Please help if you can... thanks

ive had the same problem except the opposite. when i would lightly touch the pad it would trigger other samples, but then i put it on full velocity and everything has been fine, i really don't have any need for velocity anyways, i just use the 16 levels.

At first I thought this sounded like a short, but if you've tried another good sensor matrix then it can't be that. Unless the velocity of your hits is causing a short in the junction where the matrix goes onto the mainboard. Have you tried cleaning it with a bit of compressed air?
Also - have you checked the pad assignments in the program to make sure that Mode:NORMAL and not VEL SW?

There is a service manual for the MPC2k, but it's $44
http://www.rangeraudio.com/manuals1.htm