CAMB behavior at very high multipoles

[Giulio Fabbian]

Hi,

I am trying to run camb with \ell_max=16000 (k_max=32000) with lensing and I noticed that after \ell=8900 the unlensed polarization spectrum oscillates around zero and TT power spectrum seems to increase linearly in a log-log scale. Conversely the lensing potential is show no strange trend changing. Moreover, those strange trends start earlier (e.g. \ell=6200) increasing \ell_max (e.g. 20000).
On the other end, lensed spectra looks sort of ok.

I guess this is due to numerical errors but changing accuracy parameters and k_max doesn't seem to help much.

Has anybody already experienced this kind of problems? Is there any \ell_max limit?.

I'm getting familiar with the code so any suggestion/comment would be appreciated.

Thank you in advance