That's the schematic diagram I referred to. You can clearing see that the minimum impedance of the LF input is determined by the 2000 uF capacitor. That's why I said the impedance would be 2.6 to 4 ohms, because even if you ignore the impedance of the inductors and resistors, that series capacitor C19 at 30/32 ohm will give you more than 2.6 ohms. You can calculate it yourself using the linked calculator if you don't know the formula. (
https://www.allaboutcircuits.com/tools/capacitor-impedance-calculator/)
Given the information in the schematic, It is impossible for the speaker to dip to 0.8 ohm at any frequency if the crossover is included in the measurements. Please note that at the very low frequencies, the impedance of the mid/high freq branches are very high so at such frequencies the impedance of the LF inputs are practically the overall impedance of the speaker, and again, will be at least 2.6-2.7 ohms even if you ignore the impedance of the inductor/resistor branch and the voice coil. That actually happens to be consistent with the specified nominal impedance of 4-6 ohm. There are various claim of what nominal impedance is, one being (ref source: wiki) 1.15 X the minimum impedance.
So one of the following may be the reason(S) for the seemingly contradictory information:
- The schematic diagram is not accurate, or
- Whoever made the measurements and plotted that graph did not do it right or it not accurately, or
- The measurement was done with the crossover bypassed, or
- The speaker was modified, or
- Any combinations of the above.