MC yaw leads to large altitude changes and instability - Asymmetrical Frame

When I yaw I see huge increases in altitude and overall instability. I believe this has to be a tuning issue since this is an asymmetrical frame but I did make my own model and toml file and the correct geometry which represents everything well but not sure why it’s still so unstable. I started tuning this several times but I still am not understanding the large increase in altitude. Attached are some logs. A few are just flying inside and only moving around a little but there is still the yaw alt error then the longer ones (and ones with Alt Hold) I went outside (albeit we had 15 mph winds) but its the same issue. Keeping it under control was much easier in ALT mode but still it’s not anywhere close to a treat to fly. I have one really long log but I can’t get it uploaded to Review (log is 128mb), it just gives a gateway timeout NGINX after its done uploading.

Frame is something I hacked together for fun. Asymmetrical x wide in the front (like a deadcat). About 1.5m motor to motor (front left to rear right). All up weight is 2.3kg with 6kg of thrust (spinning 1200kv motors so nothing too slow).

https://review.px4.io/plot_app?log=013b21e0-73aa-43c7-a5c9-0b0e7680c8d3
https://review.px4.io/plot_app?log=6a5ba963-2334-49f5-ba86-8cfbcc2681d5
https://review.px4.io/plot_app?log=0dc1ddaf-0da4-4fb9-89c5-25918c14511a
https://review.px4.io/plot_app?log=42319b90-007b-4d71-ac58-8d49f9d6c431
https://review.px4.io/plot_app?log=bbad021c-2c01-48d3-ac09-8706c74f07cc
https://review.px4.io/plot_app?log=101808e5-7a92-4e45-b01a-24c14e9726f7

Would love some feedback.