Arduino Nano - voltage limits / heat / PoE

I’m powering some home automation controllers remotely using PoE from the server cabinet.

Option 1 is using active PoE (802.3af) directly from a switch and some electronics at the arduino end to negotiate power according to that standard. This is technically the best option, as 48v over long cable runs means minimal power loss. However this is a brain-ache for various reasons:

  • PoE switches are very noisy, not nice anywhere in the home
  • PoE electronics on the arduino end adds a good £15 to the total project cost for each unit (expensive for 20 units)

Option 2 is using passive PoE. I’ve seen these nice injectors that take a single power supply (I get to decide the voltage) and then I can split the cable at the arduino end and simply hook the power lines up, making use of the arduino’s regulator. This is cheaper, quieter, and more simple.

On to my question.

To achieve minimal power loss, I want to put the highest voltage over the line as I can. I know Arduino is spec’d at 20V limit (as per here) but

Q1. How hot will the arduino get if that regulator is working at 20V? Too hot to include project into a wall back box? Or is heat related more to current draw? I am not at the stage I can test this yet.

The regulator on my nano clone is labelled AMS1117 5.0 H424TE. Data sheet from that chip suggests maximum voltage is not 20V but 15V.

Is this a different regulator to the nano original?

Q2. Any other quick win to get my voltage stepped down, cheaper and more simple than using proper 802.3af? Maybe add a secondary regulator to my board?

Q3. Any other thoughts to share on this subject?

Thanks!

I like to use one of these: http://www.recom-power.com/pdf/Innoline/R-78xx-0.5.pdf in a 9V output version so I can make use of he on-board regulator without taxing it too much. This device is $7.50 from DigiKey and has a 32V input voltage limit.

Most of my projects have been automotive, so I'm not comfortable running the car's 12VDC into the Arduino for a number of reasons.

Heat in a linear regulator is proportional to current. How many mA will your Arduinos be using?

Using an external linear as a “pre-” regulator will not be any more efficient as far as your power bill goes, but it will at least spread the heat production into two chips instead of one, so each will be working less hard.

Thanks for the responses, since posting my last message I got a bit guilty for not researching myself properly... and hit Google (more specifically Youtube) hard on the subject of regulators.

For anyone reading this wanting to know the answers (at least - my answers!) here goes...

1/ I also found out the Nano's 3.3v regulator does not operate unless it's powered by USB, which scuppered my plans to use the OLED display I was using. Whatever the plan with PoE this needs addressing separately.

2/ Having spent hours and hours learning about regulators I feel like an idiot saying this here now, as everyone else is probably like "er, yes we know". BUT. There are two types of regulator: linear regulators and switching regulators. A linear regulator can only convert voltage down (as opposed to switching which can convert up or down), and it's generally a bad idea to convert down from a very high voltage like 24 or 30V using a linear regulator as this is highly inefficient and results in a lot of energy lost in heat (also probably subject to overheating in a back box). A switching regulator involves far less energy loss and heat loss, but they are more expensive.

In short I'm going to use both - a switching regulator to get 30V through passive PoE down to 5. (Actually turns out not that expensive if you hit Ebay, "buck converter" is just over a quid from China including postage.

Then a separate linear regulator taking 5 to 3.3V to cope with the fact the Arduino Nano's doesn't work when not being powered from USB. This is obviously a low dropout voltage of 1.7V which won't overheat ... not unless a shedload of current is put through it anyway. My project is a couple of hundred mA so can get away with a small one and not worry about overheating.

On the subject of linear regulators and overheating, it's very easy to work out what will cause overheating. Measure the difference in voltage between input and output, then multiple this by the number of amps (e.g. 0.2 for 200mA) to give total power lost which would be converted to heat. Something like 0.3W lost is a pittance for a mid-sized regulator but 3W would cause it to heat up a lot.

Phew, hope the above helps others looking at the same questions!

The Nano's 3.3V regulator can only supply a pitiful 50mA anyway, you'd have burned it out driving 100mAs of load.

I thought about suggesting a switch-mode buck converter, but I couldn't find any that were inexpensive and would handle up to 48V. But if you found cheap ones rated for 30V, that's awesome.

Not sure what your definition of "small" 3.3V linear regulator is, but I would suggest a TO-220 package, not TO-92..

tylernt: I thought about suggesting a switch-mode buck converter, but I couldn't find any that were inexpensive and would handle up to 48V. But if you found cheap ones rated for 30V, that's awesome.

The ones I found were £1 including shipping to UK! Can't remember what they are rated at. Either way I settled on non-standard PoE, 12V down ethernet cable as the current draw is not huge and the cable is CAT7 (better conductivity) AND the runs aren't that long. So hopefully this shouldn't be too much for the Nano's main regulator. (Current permitting!)

tylernt: Not sure what your definition of "small" 3.3V linear regulator is, but I would suggest a TO-220 package, not TO-92..

Funny you should say that, I only just read your response now, in the meantime I bought a TO-92 package regulator then found out it was only good for just over 100mA which isn't quite enough for me to feel happy to mount the thing in a backbox.

I then purchased a TO-220, much healthier as it should be fine for an amp or two I think.

rmetzner49: I like to use one of these in a 9V output version so I can make use of the on-board regulator without taxing it too much. This device is $7.50 from DigiKey and has a 32V input voltage limit.

Now those are in fact switchmode regulators. The OP complained

hazymat: - PoE switches are very noisy, not nice anywhere in the home

PoE converters are in fact switchmode converters; notably they are isolated so that neither terminal is commoned to the ground. So if they are noisy, there is little reason to believe another switchmode converter will be better except that the above is suggested to be somewhat better designed with its integral suppression.

If however you are going to use such a converter, you should use a 5V one and supply direct to the 5V system, there really is no purpose at all to double-regulating. Ethernet is of course, intrinsically isolated (at each end) and proper PoE is also. The so-called "passive" version is to use a cruder non-isolated and possibly linear regulator with a common ground to the local system which means that you do not want to have any other data connection which is not isolated or you will introduce ground loops (and noise).

@Paul__B

Thanks, interesting points about double regulating (I agree) and ground loops.

As a point of interest when I said "PoE switches are noisy", as opposed to EMI from the switch mode regulator I was referring to fan noise of PoE network switches! (I'm yet to find a good network switch with PoE throughout that doesn't sound like a jet plane taking off)