Voltage booster works with a resistor load but does not work with a real load

[midix]

The short story.

2.4V -> booster -> 8.7V -> the target load that consumes around 25mA = fail, voltage on the booster drops down to 5.6V

2.4V -> booster -> 8.7V -> artificial load (resistor) that consumes around 40mA = success, no voltage drop.

What's going on? Why does the booster fail to feed the load with less current while capable of feeding a load with a higher current just fine?

Here's the long story.

I know I shouldn't expect much from cheap DC-DC voltage boosters from AliExpress. But this time it's more about curiosity than solving a problem. I'm puzzled and would appreciate some ideas or at least theories that would explain the behavior.

I have two old MT3608 modules. I've been using them for boosting my NiMh battery pack with 2 batteries up to 3.3V or 5V, typical for my Arduino experiments.

It has been working flawlessly, handling even nrf24l01 modules that are known to be picky about their power source.

Today I connected an old Yamaha WX11 MIDI controller to one of these MT3608 adjusted to supply around 8.7V. WX11 has its own internal buck converter down to 5V and it is known to work well with input voltages between 8-9V.

Everything worked well while I was powering my experimental breadboard setup from USB of my computer. Then I decided to switch to 2.4V of my battery pack. And a strange thing happened. MT3608 output voltage dropped from 8.7V to 5.6V. Also, I started hearing coil whine from MT3608 (very quiet, as a mosquito). I tried my other MT3608 - the same behavior.

I thought, maybe 2.5V is not enough for boosting to 8.7V with a load. That would not be strange at all, right? I disconnected the load - the voltage was immediately restored to 8.7V.
Then I measured the current normally consumed by the MIDI controller. It was about 25mA. When I power my Arduino Pro Mini from the same 8.7V (on RAW pin), it also consumes about 20mA and works fine with MT3608 and 2.4V input. No voltage drop, no coil whine.

Then I did an experiment. I took a 220Ohm resistor and used that as a load on MT3608. It consumed about 40mA. And... it worked just fine! No voltage drop! That got me confused.

Do you have any ideas? What in the WX11 could mess with MT3608 output so badly that it fails to keep the voltage stable when it obviously has enough juice to deliver much more current?

Here's what the power input of the WX11 looks like (sorry for the right-to-left current flow, that's how it was drawn in the manual):



I suspect the input coil might do something bad with MT3608 output. The coil looks like a small toroid, but still, it is pretty large, almost the largest component on the circuit board.

Can you confirm my suspicions and explain what exactly is going on there? What's the purpose of that coil?

[DavidAlfa]

The difference with a simple resistor is that your device will cause a current surge when connecting it, all the internal capacitors get charged, you're very close to the 2V limit, if the voltage drops just a little when the converter does the switching it will turn off, then recover, turn off again...
The device will probably not work at all below 6-7V, so the surge will keep happening everytime the converter voltage rises.

Try adding some 100...470uF caps at the input and output pins of the module.