Removing peaks (noise) in the output of Switch Mode Power Supply

Question

^{simulate this circuit – Schematic created using CircuitLab}

I have a 220/24V SMPS, which I use to power 2 buck convertors in chain. First one 24-12V, and the second one 12-5V. With this setup I have two problems - first one is that one of my arduino nano board resets quite often. The second one is that I have LM358 opamp as a voltage follower and its output is lower than the input by a few hundreds milivolts when the voltage is close to 5V (opamp is powered with 12V). This second problem is dissapearing when I touch the ground with bare hands. So for both problems I blame the noise in the power supply. Here I have taken a few pictures.

First one is the 24V power supply without the rest of the circuit attached. It seems that is working near 50kHz

Here is how the peaks looks at higher resolution. These peaks are bigger than 1V sometimes.

Here is the output of the 12V buck convertor, when powered by HY3005 benchtop power supply. This convertor works at 180kHz

Here is the output of the 15V buck convertor, when powered by HY3005 benchtop power supply through the 12V buck convertor. This convertor works at 180kHz too, it is the same as the other one.

Here is the 12V buck output, when powered by the 24V SMPS, you can see that the peaks from the 24V go into the output of the 12V buck

And here is the 5V buck output, powered by the 24V SMPTP through the 12V buck. Here the peaks from the 24V are visible too.

So, I blame these peaks for the strange behaviour of the circuit. I tried to get rid of them by adding different capacitors (220uF, 2200uF, 1uF, 100nF, 470pF, 10pF ) in paralel of the 24V, 12V, 5V... but they are still there. I also tried 1uH and 100uH in series to the 24 and 12V, but either didn't help, or in the case of 100uH it created a ringing. So, I'm out if ideas how to get rid of these peaks.

Edit: Some findings:

1. The peaks from the first picture are introduced by the Input of the SZBK07 DC-DC buck convertor. They are so narrow, because the duty cycle is very small at the moment of measurement. When I increased it, they got wider.
2. When SZBK07 DC-DC buck convertor is not connected, the opamp follower works fine and the output voltage is the same as the input. It is the same no matter if the circuit is powered by the 24V SMPS or by bench supply.
3. The opamp output is lower than input only when SZBK07 DC-DC AND the 12V DC-DC (thus the opamp itself) is powered by the 24V SMPS. When it is powered by bench supply it doesn't appear.
4. The opamp output voltage gets the same as the input when the ground is touched by bare hand.
5. The 5V DC-DC has nothing to do with all of this, because everything in the above points is tested when the 5V part is powered by phone charger.
6. 5V DC-DC convertor somehow resets one of my two arduino boards. No matter if it is powered by the 24SMPS -> 12V DC DC, or by bench supply - 12V DC DC or directly bench supply - 5V DC DC.I cannot see anything different on the scope when the reset appears.
7. Earth grounding the input side of 24V SMPS doesn't change anything.

Also I cannot see anything changing on the scope when I touch the ground with hand.

Edit 2: This is the layout of the boards. There are two modules missing - RTC and Bluetooth, which don't affect anything. Small board on the bottom is current sensor which I don't use yet. Top buck on the left is the 12V, middle is 5V, down on the left is SZBK07

Here are the connections - down on the left red diagonal wire is supplying 5V DC-DC with 12V and is currently unsoldered, as I provide 5V through the usb port of the arduino. The blue trimmer on the top replaces the digital potentiometer for test purpose, because it is easier to set the voltage manually.

This is the back side of the SZBK07. The output (the cable on the right) goes to a small car lamp for test purpose. On the input you can see temporary soldered 1uH and 10pF. I have to mention that there is no difference where this board is placed - it could be here, or half meter away, the problem is always present. Earth grounding the input side of 24V SMPS doesn't change anything.

Answer 1

The problem was solved in a quite trivial way. I connected the earth ground to the output ground of the SMPS (thus to the ground input of the DC_DC convertor) and the opamp output DC offset disappeared. Moreover, I tried something that wasn't working before - to control the DC-DC current directly through the digital potentiometer. And it worked too! The problem was the same, because the current control is actually a variable gain Opamp. Resets was fixed by adding 2200uF + 10nF near the microcontroller board.