I got both cca c10 and kz zs5.
Using PC setup (PC->DAC->RCA->Adapter->3.5mm->IEM), why is cca c10 loudnesss is around twice of the kz zs5 (using same volume level in PC)?
C10 has 2dB/1mW higher sensitivity. And that assumes both are accurate figures or were measured the same way, ie, 108dB/1mW and 106dB/1mW at 1000hz.
Everything same setup (sound level, cable, etc2), only IEM swap.
The thing is that this volume difference doesn't exist? (might need to reconfirm) if I use a mobile phone as DAP (don't have actual DAP to test).
What "PC" ie motherboard are you using?
At 18ohms vs 32ohms, a crap motherboard can drop power output significantly at either the same way even hifi headphone amps can have less power at 16ohms than they do at 32ohms or vice versa, except you're starting off with a higher figure so in most cases it's inconsequential.
That or your motherboard has some kind of automatic impedance sensing circuit (as on some Gaming boards) which only looks at impedance without taking sensitivity into account. It's like putting a smaller engine into a smaller car on the basis of it being lighter but that body was never tested in a wind tunnel and now above 100kph the fat sedans it was toe to toe with are dusting it.
The kz zs5 actually have about the same volume level as the normal speaker setup I use (DAC->RCA->speaker setup).
Only the cca c10 is different
Here's your other problem. You're going by perception, not actual measurements at one frequency much less a full spectrum graph, or even just your ears but on one freq sinewave.
That's like testing torque output and acceleration by how it feels at the seat of your pants, when an older V6 Mustang can go from 0 to 105kph in 6.5seconds with a satisfying kick while a Honda S2000 deceptively gets to 105kph in under 6secs without that kicking sensation. Or you sharpen a knife with a cheap 400grit, claim it's all skill and not the price of the knife or the stone...and then use as basis for how sharp it is how it feels on your fingers when that's the jagged peeled off metal still sticking to the knife that will result in raw fish slices on the plate that look like they were chainsawed off the filet, losing juices between slicing and consumption.
Yes, the Mustang has more torque, but it's not faster than the S2000, much less around a track. Yes, that knife has a sharper bevel, but that's comparing it to the dull rod you started with, which says nothing of how refined the edge actually is.
In short your perception of loudness is very, very, very far from accurate. That's not to say it's totally wrong, but by how much is it louder is problematic in taking a qualitative data as quantitative data. A good deal of difference can be due to the difference in frequency response, ie, one being louder in some frequencies can make it seem louder - either because the midrange is louder, or when the treble is louder you hold off on the output sooner and get less bass, and vice versa and perceive the sound to be much smoother - which exacerbates any number of the other things previously listed that may explain the difference.
Personally, if the sound is usable to you otherwise (save for adjusting the volume), then any other problem is not big enough to be perceptible to you anyway.