Switching Regulator Success / Failure stories?

Gateways
1

Anyone have any specific successful outcomes or failures with particular switching regulator solutions for powering LoRa concentrator cards in gateways?

Had a past experience with a 433 MHz AM (on-off keying) product where noise from the switching regulator limited receive range, so in interest of time we de-populated that in favor of the linear one included in the board design as an emergency backup. LoRa should be somewhat more immune to AM-type noise, and higher frequency helps a tiny bit, but it seems like there could well still be an issue.

Currently have several RAK833 mPCIE concentrator cards for experiments, which does mean the possibility of doing A/B testing with various power solutions feeding two side-by-side systems listening for the same distant node (and of course then swapping the cards or supplies to rule out any variation in the RF paths)

It does look like the RAK833 has an on board switching regulator for a lower logic core supply, plus empty pads for a second - I wonder if they considered letting the module accept 5v in like the RAK831 does, although that would not be mPCIE compliant. It is nice that the same module can operate both in USB mode in a PC-derived router chassis, and in SPI mode with various ARM-based embedded platforms (tested by soldering rework wires to the SPI pins on a modified mPCIE-to-external-USB adapter). However the tiny switchers on those adapters quickly overheat when running a LoRa concentrator rather than the mobile network modems they were designed for, so coming up with a replacement power solution is the next step in a modular proof-of-concept.

Any of the other mPCIE or similar cards looking to have a positive future, especially in supporting SPI rather than only USB?

2

I have had issues with switching regulators in the past when some USB powerbanks were used (these typically have switchers insides) for portable receivers. They had a fairly drastic affect on reception.

Whilst you might think that LoRa would be more immune to switching type noise, do remember that at the limits its receiving signals that are between -25dB and -30dB weaker than a typical FSK type receiver.

My preferance would be for a linear supply, or at least take the time to measure the performance differance between linear and switched.

3

Yes, I can’t really imagine doing (an initial) adapter plate board that didn’t include a footprint for a fallback linear regulator option.

But I do think that power and thermal budget necessitates at least giving a switching regulator a try. And the fact that the concentrator cards in hand have a switching regulator on them for logic is a positive sign that this is possible (even if the the critical RF parts may not be running on that, it is there in close proximity)

So the practical question is more “which switcher to try”?

4

No idea, some might be OK, some might not.

Thats why I avoid them.