MP3431 bad regulation at high/3A load (beginners question)

Hi, I’m a beginner (hope, I don’t bother you) and am testing my circuit around a MP3431(schematic and layout as pic). Output voltage is set to 6.47V and my load for testing purpose is 3Ohms and 2Ohms. Input from a battery is 4V. With 3Ohms load, the regulation seems reasonable, but with 2Ohms I have problems, like seen in the scope pictures. ch1 is Vout, ch2 is Vsw, ch3 is inputcurrent over 5mOhm and ch4 is Vin.I oscilloscoped Vfb too, but it looked good, no overly noise. I dont understand why it is switching with ca. 1.6Mhz and than stop switching and battery voltage is raising and what else is going on. I appreciate your help!

best regards, Robert

for scope, layout and schematic click:

At a guess the thing is drawing too much current from the battery. The battery voltage goes down until the UVLO undervoltage lockout is hit. Chip stops, battery voltage recovers and the cycle begins again. What kind of battery are you using. I would try first with a strong power supply before progressing to battery operation

Thanks, I have tested it. If UVLO is hit, output voltage will go to zero for a short time. But as can be seen in the scope picture, this is not the case. The blue curve is the input voltage and around 3.6V at it’s lowest value.

UPDATE!
unfortunetly I can not update the title. I have tested the same setup but with 1Ohm load and the regulation seems well again (outputvoltage dropped to 5.9V, because the max Inputcurrent got reached(about 10A)). I still don’t now what’s going on at 2Ohms. Is the system oscillating? but why does the switching, what it does (way more than 600kHz)? I will use it later to drive a led, in case that plays a role.

Have you tried the demo board?

I have to say no. It wasn’t avaiable at Mouser, too.

I have soldered a 100nF cap parallel to R32, this helps to prevent the feedbackloop oscillating and the regulation kept still stable. I thought the coupling shold worsen the regulation, maybe R31 would be the better choice. I’m still wondering what leaded to the fast 1.5MHz switching…