Reliability issues when powering a project with two AA batteries through a 2108a DC booster

I'd be especially interested to hear from people who have used 2108a DC booster. Is it working reliably for you?

I have a project with Mini Pro 3.3V/8MHz, some sensors and nRF24L01+. Average consumption seems to be about 16mA.

The requirement is to power the project with 2 AA batteries (using Ni-MH now). It will be a handheld device, so I don't want to risk dealing with Lithium complexities.

Currently, I'm using a MT3608 DC DC step-up boost converter to get 3.3V out of 2.4V. It works reliably, no issues whatsoever.

Then I thought I'd like to use a lower power converter because I don't need 2A and also there's no use for that potentiometer if I need fixed 3.3V.

So I bought some 2108a voltage regulators. They are tiny and are told to support 160mA-400mA at 3.3V. Should be enough, I thought.

Still, when I connect my project to the 2108a, it behaves unreliably. The voltage is ok-ish at 3.24V and Mini worked fine, no resets etc., but there is obviously something wrong with SPI communication or nRF24L01+ because there's insanely high packet loss. Unfortunately, I don't have an oscilloscope to see how stable the power of 2108a is.

I have a 10uF cap soldered on nRF24L01+ VCC and GND, but it doesn't help.

What could be wrong with those 2108a? Could they be unable to provide enough juice for nRF24L01+ when it's transmitting? Is their output too noisy?

Are there more reliable small fixed-voltage 3.3V DC boosters?

You mught need thise 2 Amps when transmitting.

You might be right. However, I'm using the RF24 at low power mode. And also, the weird thing is that all the packets seem to reach the target - it shows received counter increasing. It's the acknowledgement receiving part that works flaky with 2108a. It's as if 2108a fails to deliver enough power for receiving ACK packets, and it doesn't make sense.

Can you show a schematic ?
And a photo ?

When you ask something, you have to provide the information.
Please give links to the chip. Is it a ADP2108 ? Why do you think it is a ADP2108A ?
So it is made by Analog Devices ? with this page ? ADP2108 Datasheet and Product Info | Analog Devices and this datasheet ?

I think the 600mA is enough. But its voltage output drops already with a small current and the output is not very clean. This might not be the right converter for you.
However, according to the datasheet, the voltage should be higher than 3.24V. I think there is something wrong with your circuit. What kind of inductor do you use ? Can you show a photo of it ?

How about the capacitors that are mentioned in the datasheet. Do you know that certain capacitors are only for a certain frequency range ? https://www.youtube.com/watch?v=BcJ6UdDx1vg

It is not important how much current the converter can supply. The efficiency and its own current usage is important. If a converter is efficient and small and cheap and can deliver 100A, then there is no reason not to use it.

I suggest to use a modern step-up only with a better regulated voltage output. Pololu has a nice selection of modules, for example: https://www.pololu.com/product/2561 using the TPS6120X chip.

Is that a real or a fake nRF24L01+ ?

The nRF24L01+, most likely, is a fake.

The 2108A looks like this:

They are tiny, just as wide as 3 pins.

Currently, everything is connected to a breadboard with wires.

You are right, I'll try a module from a more reputable brand.

As it often happens, I'm paying twice when trying to go too cheap (and small :smiley: ) Still, I might reuse those 2108A in some other, simpler project.

The ADP2108A switches at a high frequency, 3MHz, this is probably an EMI issue with RF getting into the wireless module and deafening it. Some more supply filtering/decoupling and completely screening the converter might help.

A single LiFePO4 cell is a great way to power 3.3V circuits BTW, as LiFePO4 chemistry is 3.2V nominal. But yes you need the bespoke charger.

2 Likes

Ok, the mystery is solved.
All it needed was a large electrolytic capacitor. It seems that the 10uF cap I soldered directly on the RF24 pins is not enough, at least not for this DC booster.
While experimenting, I found that 470uF is the smallest value that ensures reliable functioning without any packets having to be resent.

1 Like

That suggests the boost converter chip can't handle rapid changes of load gracefully. This can be the case when switching from very light load to heavy load - some converters idle at a much lower switching speed, and get caught off-guard if the load jumps a lot.