L2 filter problem

[samezrp]

I would like to rise another problem.
I'm preparing my implementation of your PowerMeter and I can see there is a problem with channel connected to: ADC1-CH7 (GPIO35).
I have HW with all 3 current lines connected to SCT013-030 installed on the same wire + 3 ZMPT101B fed with the same voltage. So all 3 voltage and 3 current reading should be the same.
And when you look at live view all is the same, but when you take a closer look at L2 and current, it is seen that this value is different and is not stable as others.
Take a look at this video:
https://youtu.be/L46QpZTydEg
on oscilloscope readings for L1 and L2 are the same. But when you look at spectrum and filtered value you can see difference. On spectrum you can see some pumping, L2 value jumps.
All channels are set the same, my measure.json file below.
@sharandac would you mind helping me with that?
measure.json
I checked different ESP32 boards, I played with sensors, channels and look like problem is attached to: ADC1-CH7 (GPIO35) directly.

edit:
After some research
-it is not good idea to change ADC channels for predefined L1-3 current and voltage, so follow default:

L1 I ADC1-CH5 GPIO33
L1 V ADC1-CH0 GPIO36 SVP
L2 I ADC1-CH6 GPIO34
L2 V ADC1-CH3 GPIO39 SVN
L3 I ADC1-CH7 GPIO35
L3 V ADC1-CH4 GPIO32

if you change this any calculations defined in virtual channels such a power will be broken;
-in default measure.json microcodes for power and reactive power are mixed, so please be aware of that.

BTW why "all power" as a sum of all powers is not a sum?

0,042+0,041+0,043 != 0,101