Planar magnetics can often mess with a microphone's operation - the magnetic part of planar magnetics can potentially affect the coils in a microphone, which have their own magnetic fields. May be why you see such a raggedy FR plot for the Sundaras. Maybe.
To be precise the one concern I had was the 5 or so dB high-Q drop at around 1.2kHz that both channels exhibited, not the overall wiggliness of the response. I was expecting rather a 2dB or so drop given the available measurements of the 2020 version on ear simulators (and my version was in all likelihood manufactured in 2021 if I interpreted the labels in Chinese correctly).
@solderdude knows more about this than me, but just like him I thought of the presence of electrets / MEMS within the front volume of most ANC headphones, although most of them aren't planars.
It's a bit of a moot point anyway since the DIY probe's microphone sits outside the earcup :
And more importantly since a second sample didn't have such a sharp high-Q drop at around 1.2Hz and something more aligned with the 2 or so dB drop seen in most other measurements (I was only able to check that on the right earcup as the left one didn't work out of the box).
The high-Q wiggles are, as Solderdude wrote, something often seen in quite a few planars, Hifiman in particular.
For the DIY probe traces, some of the high and medium-Q wiggles are a by-product of my early attempts at creating a compensation file for its raw response, I've learnt over time how to improve it, but it's still a work in progress.
The latest compensation files I use tend to look more "right", ie most of the wiggles caused by the compensation process have been eliminated. For example, with the DIY prove above vs blocked ear canal entrance microphones below, HD650 in red, HE400SE in blue :
1/48th smoothing for the blocked ear canal entrance mics. For the probe's traces it's impossible to know exactly given the way I compensate for its raw response.
For both methods please remember that the absolute values are incorrect, particularly above a few hundred Hz, and that the relative differences between the two headphones are only applicable to my own samples on my own head.
I'm only really interested in relative comparisons so creating a compensation file for the DIY probe is optional, it's just a more intuitive way to look at the acquired data, that's all.