OK, if the peak amplitudes are at about 45%, that matches pretty closely what the LEDs on your MiniDisc recorder were telling you (45% modulation = 6.9 dB below full scale).
Again: If your AD-20 is similar to mine, was working properly, and was properly powered, you couldn't possibly have overloaded its preamps given that its digital output was below 0 dB (full scale) at all times. With apologies to jlykos, volume alone doesn't overload a mike preamp; voltage does. The voltage at the input of a preamp is a function of volume, of course, but it depends equally on the sensitivity of your microphones.
The AD-20 has enough input headroom across its entire range of gain settings so that any signal that could possibly overload its preamps would also drive its A/D converters to full scale (and beyond if that were possible, which of course it's not). But yours was driven only to -7. So we can safely conclude that the preamps weren't being overloaded--provided that the 9V battery was really OK at the time.
Can you recreate the hookup that you had, and play something equally loud at the mikes? If so, you could try putting resistive pads (attenuators) at the inputs of the AD-20 to see whether the distortion goes away or whether it merely occurs at a level that's correspondingly lower (by the amount of the pad's attenuation). Based on what you've said so far, I would expect the latter situation to occur (e.g. if you use 20 dB pads, the same clipping would occur 20 dB lower in the recording). If on the other hand the distortion goes away or is greatly reduced, then the AD-20 would be highly suspect because it absolutely should not overload that way, as I've said "many times, many ways" by now.
--best regards