Buck converter (MAX5035D) regulation problem

Question

I didn´t expect to be having issues with a buck converter, but here I am. My design goal is a simple 15-60 V to 10 V buck converter. I settled with the following schmematic:

I designed a 2-layer board with copper pours on both layers with stitching vias:

(Ignore the terrible soldering)

During testing, I made temporary changes to component values:
Input capacitance: 11 μF,
Output capacitance: 10 μF,
Feedback resistors: 9.1 kΩ and 1.3 kΩ

I tested the converter with an input voltage of 25 V from a fairly noisy bench supply. I loaded the 10 V rail with 120 Ω. Here are the results:

"DC" results:
Vout = 5.05 V instead of ~10 V, Vfb = 0.63 V instead of ~1.22 V according to the datasheet.

"AC" results (with bad oscilloscope probing):

Data measured after changing inductor from 22 μF to 100 μF:

Weird switch node voltage:

I know that the layout and component selection is not optimal, but I think this should only explain high ripple, low efficiency, etc. but not such a massive regulation issue.

Datasheets: Diode MAX5035

Answer 1

Here's my ideas on the matter:

The inductor value is too small. The equation calls for 44uH at least

vin=[15 25 31]; vout = 10 ((vin-vout).vout./vin)/(0.32*125e3)
ans =
4.4444e-05 8.0000e-05 9.0323e-05

Counter intuitively a lower current needs a higher minimum inductor, so you might be able to put a higher load and get better results.

Not having the feedback at the right point has been a problem for me in the past. One problem is the FB point will see a smoothed version of the inductor switching and may not switch at the right time.

But before you take the effort to do switch the wire, make a differential measurement (with a diff probe at pin2 of the inductor and pin 1 of the C10 pad) if you see a difference more than a few mV than that could be the problem.