Arduino Nano33 IoT voltage regulator performance

Hi all,

Long story short: I’m trying to power an Arduino Nano33 IoT from a 19V supply, but its 3V3 regulator gets so noisy at voltages over approx. 8V that the setup becomes extremely unstable.

The only other thing apart from the Nano itself that’s powered by the onboard regulator is a LumenRadio TimoTwo radio transceiver. Its current draw is specified to 150mA typ., 250mA max. I’m only using it as a receiver, so I wouldn’t expect it to be near the max spec, and in any case, according to the datasheet for the Nano’s regulator, this shouldn’t be a problematic load. I’m currently not using the Nano’s WiFi, BLE or IMU sensor.

There’s a SEPIC DC-DC converter on the same PCB that the Nano is attached to, and I was worried that it might be causing some kind of noise on the supply lines that could disturb the Nano, but I’ve checked the 19V supply with a scope and I can’t see any extra noise at all when the DC-DC converter is connected. I believe it’s well filtered too – I used TI WEBENCH Power Designer to verify the design, so it’s all within mfgr specs. I don’t think it’s a problem with RF noise from the SEPIC converter either, in that case it shouldn’t have worked when even when the Arduino was USB powered instead, right?

I’ve tried throwing in some extra caps between 3V3 and GND, but the difference I’m seeing is absolutely minimal, not enough to make any practical difference at all. I only have 100nF and 330nF on hand to test though.

The Nano’s regulator does not get warm at all – I checked with a thermal imager, it’s barely above room temperature.

I’ve attached a couple of scope shots of the 3V3 line with 5V and 19V as supply voltages. You can probably see which is which …

Are there any ways to solve this, apart from just cobbling in a better 3V3 regulator? I’d prefer not to do that, as this is a very space constrained application, and the PCBs are already made. They’re a tight squeeze as is.

Without a schematic and not knowing what the high frequency noise is I can only take a guess. I have seen similar things on several regulator circuits. Typically it is an indication of improper decoupling and or layout. Look at the regulator specifications and take a look at what they used for high frequency suppression. Even more important is the max lead length, probably less then 2" or 1" per lead. You also need a bulk capacitor at the regulator, the one in the remote power supply does not cut it. Part of this can be a layout problem, I am only guessing. Try soldering a 10nf (maybe 100 will work) to the input and ground lead then do the same with the output. If this is a through hold design just solder the SMD caps on the regulator leads where they come through your board. While you are thinking about this order another regulator from a different manufacturer, they are not all the same.

I would disconnect items in turn to identify the noise source, replacing with the equivalent resistive load if needed .