I absolutely had the same problem with the combined ENS160 and AHT21 chips. I posted about it in the hass discord, but the short version is that the issue was to do with the power to the ENS160 falling below requirements at some points, and the fix was soldering on a supplemental capacitor to prevent the dropout and provide steady power to the ENS160 sensor.
I fear those messages are lost to time on the discord now, but I’ve attached a picture. It’s a 47uF capacitor from the positive side of the ENS160 supply (it was like 1.7v or something i think) to ground. Note that these boards differ sometimes… in one case i had to go from the output of the fet and in the other i could solder to the smd capacitor… they’re almost identical, but not quite.
That said, even after doing this, i found I was getting wildly variable results from several otherwise identical sensors in identical circuits in the same room.
I decided to use separate ENS160 and AHT21 boards, and unfortunately even those aren’t really as consistent as I’d like, and a few may have the same issue of the persistent 0 value issue you report above. I haven’t tried to fix them yet, so we’ll see. I’m not sure how good these really are ultimately, or at least, the ones you can get from AliExpress. Caveat Emptor i guess.