Quite the difference...
I have been in belief (on FPGA 259, DSP 111 -firmware), that ADI-2 Pro´s EQ scales its analog output peak levels according user set maximum emphasis -level to 0 dBFS level. I.e. if user has set EQ of PH 3/4 to +10 dB emphasis, say @ 50 Hz, and turned volume to 0 dB(r), THAT 50 Hz area of the output passband is that +10 dB -peak part of the "graph", which volume indicates to tickle that 0 dBFS maximum at PH 3/4. And the rest of the output "graph" stays <10 dB lower.
So in hardest case scenario (0 dBFS signal in from AD/DA), ADI´s PH 3/4 could be OVR:ed only >0 dB(r) -volume settings (+2.5 on Main Out 1/2). Below 0 dB(r) volume setting it could not be OVR:ed from output side in any circumstances.
So what part I haven´t understand, doesn´t it nevertheless work that way? How is that possible in your picture to get PH 3/4 to OVR @ -10 dB(r) volume setting on "KaiS HD650" -EQs? Is there differences in firmware versions how EQ and volume behave regarding analog output levels? I don´t say this aloud but could active loudness -function cause that... ?
Last winter I had those mysterious, still unsolved, clipping problems on other firmwares than 259/111, if you remember (propably don´t). Clipping occurred suspiciously pretty EQ -boost amount much earlier than with 259/111 (meters didn´t show OVR). Final consensus here was I was fucking things up myself, overloading analog input and/or my headphones or using crap level external gear with blasphemy (headphone output of AVR). Which wasn´t the case. I gave up since then though, and been just using 259/111.