This time i decided to take a different approach to test the converter limits.
The setup:
Repeating drum sample (44kHz 16 bit)
The volume of the drums was automated to start at 0dB, increasing slowly to -6dB at around 1:30,
then it sharply increases to +6dB, decreases shaply to -6dB again, and then decreases slowly to 0dB
The I88X ouputs were connected to a Soundcraft mixer using unbalanced cables, the preamp gain was set to 32dB (conservative)
The channel fader was set to +10
The master LR buss was set to +10
The 2 (identical) outputs from the console were fed one to each A/D converter (PCM4222, I88X)
The gain on both converters was matched to less than .1dB
The idea for this test was to drive the converters past their clipping point and well into
what I would call "smashing". The results speak for themselves! I doesn't sound too pretty soon after the peaks go past 0dBfs. One interesting note is how the
converters "recover" from transient overloads. It can clearly be seen that the I88X stays at full (digital) output even after a reaching the clipping point.
The PCM4222, on the other hand, clips and then backs down to a stable level below 0dBfs. This difference in clipping characteristics was unexpected.
I have included the files for comparison, along with some screenshots of the "Drum Smashing"...
