I'm not really confident in interpreting electronic circuits and excuse me if I'll say some nonsense but it seems to me that PAH2 is totally different to PAH1 and even to A20h\A10h.
PHA1 uses different opamp and has high output impedance (150 ohm as Solderdude says) at the 3.5 out probably as an easy way to cut voltage on low impedance loads, not having a gain switch. (
http://dieselpowered.ca/2018/sabaj-pha1-teardown/ )
A20h\A10h use TPA6120, that for TI specs needs a minimum of 10 ohms output impedance for avoiding oscillations, but they use an architecture similar to L30 that permits lowering that to near 0 ohms (as measured in Amir A10h review).
PHA2 uses the same TPA6120 but seems more an implementation on TI guidelines, I suppose it can have about 10 ohms on the 6.3 out and probably similar to PHA1 at the 3.5 out.
It remind me a lot a very basic diy board i bought times ago just for toying with:
https://it.aliexpress.com/item/1005003728321132.html
It has 10 ohm resistors at the output and I measured accordingly with a multimeter applying 10 ohm load.
Just as a side note, I enjoyed assembling it in an aluminium case with a dual 12v transformer, it would even perform well if not for a great flaw in the circuit: they didn't use a servo buffer before the tpa6120 and connect the 50kohm pot directly to it, resulting in high and variable DC offset (200-400 mV) and terrible rustling when turning the pot due to dc flowing into it. As you can see in the PHA2 pictures, it has a proper servo with on opamp before the tpa6120, so at least it should not suffer for this problem...