23

I'm planning on adding multiple IoT devices in my house, and I'm looking for an elegant solution to power them all. They'll all need 5 V, a small number of them (cameras) would draw up to 2 amperes, while most of them would draw less than 500 mA.

I'm thinking of running a separate 5 V line, through the house, with 10 AWG cable (to reduce the voltage drop) and a 120 V - 5 V, 20 ampere transformer on each floor. I would add some kind of connectors, each a couple of feet, in order to be able to connect to the line easily in the future.

The main purpose of all that is to keep my electrical outlets free, save all the single transformers I'll need, and allow me to extend my IoT network in the future, without using any extra outlets / unnecessary hanging cables. Is that a viable solution, or is there a more efficient way of accomplishing my goal?

I'm contemplating whether it would be better to use a 120 V - 12 V transformer, in order to compensate any voltage drop and add buck converters / linear regulators on each device, but I would like to save the extra hardware if it's viable.

Peter Mortensen
  • 1,676
  • 3
  • 17
  • 23
Velimir Tchatchevsky
  • 333
  • 2
  • 6
  • Are any of these IOT devices small enough that a local battery would run them for 6 months? I understand zigbee devices can run for a year on a CR2032. – Criggie Oct 03 '19 at 07:16
  • 6
    These days I can buy socket faceplates with integrated USB 5V outputs. Another option you should consider is power over ethernet (PoE) – pjc50 Oct 03 '19 at 09:56
  • As all these devices are network-connected, power over ethernet (PoE) is the most sensible solution. Yes, you will have to use adapters to step down the voltage at each device, but the fact is, you're not going to get away from needing a higher distribution voltage than 5v. Besides, most professional equipment deployed in this way has built-in PoE power supplies. And it sounds like standalone PoE=>5v power supplies are relatively easy to obtain. – Roy Tinker Oct 04 '19 at 17:34

4 Answers4

40

I've used power-over-ethernet successfully in various devices deployed around my house.

Advantages include:

  • Can use any existing Cat-5 wiring you may have
  • Any wiring you put in can be repurposed later if not needed.
  • Off-the-shelf adapters for supplying and receiving power are readily available (usually sold for CCTV systems).
  • You can use for wired network connections, which are generally more reliable than Wi-Fi

I'm using "passive" PoE (4 wires for network, 4 wires for DC +/-) at 24V, with local 5V stepdown at the receiving end. The PoE 'injector' runs off my UPS, so the gadgets stay powered over a power cut.

Ian H
  • 401
  • 3
  • 3
  • This could work. There are PoE switches (pricy) that can also be used. But, again, the current has to be stepped down to 5v. This will require an adapter. And since it is 5v (I bet it is for USB), you have to get an adapter for it as well. [Aparently, such things exist](https://www.howtogeek.com/294628/how-to-power-a-usb-powered-device-over-ethernet/) and shouldn't be too hard to use. Since the cables are ethernet cables, you could use an already existing cable installation, which cuts down in setup time. – Ismael Miguel Oct 02 '19 at 14:51
  • 2
    I find this answer much more useful than the other. The OP is not specific about what "IoT" devices he is using, but many such devices come PoE-ready. So PoE is the perfect solution for them. For other devices, if they require 5v power, it won't matter what power delivery design is used, because you're going to want to use higher voltage than 5v in any case, and so will need to step the voltage back down at the device anyway. Again, might as well do this with existing PoE-compatible hardware instead of trying to implement yourself – Peter Duniho Oct 02 '19 at 18:14
  • 1
    @PeterDuniho I wrote "the other answer". No one solution will suit everyone. PoE is essentilly a version of one of my 3 options and is fine enough if cost is no great object. Converting the output to various requirements remotely either adss cost of extra equipment or PoE with that capability - eg USB charger, 12V for external disk drive (of which I have many), +/- DC or AC for some coms gear , ... . || The OP specifically asked about 5V loads. My answer and this answer digressed into more general loads. PoE does have the advantage of mains fail backup - but then needs a UPS. – Russell McMahon Oct 03 '19 at 00:45
  • @Ian H, does PoE require a separate cable for each device really? From what I read it's possible to run multiple devices on the same cable, but it's a non standard solution, that wouldn't work with the more powerful PIDs that require a handshake (or am I wrong)? – Velimir Tchatchevsky Oct 03 '19 at 02:50
  • 3
    This is the answer - I've run all sorts of things from POE, either natively like phones, cameras, APs, or using a splitter at the device end. Something like https://www.adafruit.com/product/3785 works great for a pre-POE raspberry Pi, and they have variants that output 5V or 12V. Also you have reliable wired networking not wireless, and one central point to put on a UPS to protect them all from power outage. – Criggie Oct 03 '19 at 05:36
  • 5
    +1 This was the solution to the problem that industry settled on and they had a LOT of money riding on the whole deal. The telcos used 48 volts to power the old POTS phones and PoE is using the same sort of voltages for the same sort of reasons. Trade off between safety/insulation and wire gauge. – KalleMP Oct 03 '19 at 07:20
  • 1
    @IsmaelMiguel Perhaps you should handle the stepdown to 5v on the IoT device themselves... you already need to handle circuit protection on the device, lest a surge or some faulty converter on the network takes them all out. This would make things a little more simple for you, network-wise. – SnakeDoc Oct 03 '19 at 19:15
  • 1
    @SnakeDoc that could be an answer all of its own - referencing to the design of 100V line speakers for Malls which do exactly that, locally drop the 100V down to whatever that speaker needs, – Criggie Oct 03 '19 at 19:36
  • @SnakeDoc Why? The PoE adapter does it. Besides, an IoT device can be a raspberrypi zero. And that lil thing doesn't have any room for anything. – Ismael Miguel Oct 03 '19 at 19:57
  • @IsmaelMiguel The PoE adapter makes sure it's safe for PoE devices... which you RPi isn't (unless you use a PoE Hat). If you're running straight current into your RPi's... you'll kill them before too long. They don't have circuit protection on the GPIO's, and on the regular inputs they just have fuses that'll pop. Probably better to make a small proto-board with some basic protections on it to feed your RPi. – SnakeDoc Oct 03 '19 at 20:10
  • @SnakeDoc check the link in my comment from yesterday. Its perfect for the Pi application. – Criggie Oct 05 '19 at 02:40
36
  1. A 5VDC network is liable to be "a bad idea".
  2. A somewhat higher voltage network with local regulators should be slightly better but adds electronic complexity.
  3. Local 5V psu's with a shortrange distribution system is liable to be better overall.

12 gauge cable has a resistance of ~= 1.6 Ohms per 1000 feet or per 500 "loop feet".
So you get 1.6V drop per amp for 500 loop-feet or about 3 mV per loop-foot per amp.

Option 1:

A 5V run of say 100 feet at 10 A will drop 3 mV x 100 x 10 = 3V.
That's obviously too much at 5V.
You MAY deciede that 5A max is OK - or that the mean current is half the max if ecenly distributed, and a max length of 50 feet may be acceptable.
But , making compromises will still give you drops in the 0.5 - 1.5 V range.
A few cameras at the far end are liable to 'spoil your day'.

Your network needs to either be "dumb" and able to supply max power at 5 VDC or intelligent and only supply power by negotiation. At 10A that's 5V x 10A = 50 W and probably double that to allow for drop and fusing and ... . 50 - 100W is enough to start a fire with a little ingenuity. Murphy has lots of ingenuity. It's not fatally flawed but care is needed. This is lower current and lower voltage than mains AC circuits provide - so if you take as much care with your DC network as your AC mains one it may be OK - noting the need for switches able to handle DC well at rated current (DC being substantially more difficult than AC).

Option 2:

Distribution at say 8VDC or higher (10-12VDC better) (or you can use AC) will allow freedom from voltage dropout issues. If you use linear local regulators then you lose efficincy with increasing feed voltage. If 8VDC is "just enough" you are losing (8-5)/8 ~= 40% of you power. And more again with more voltage. And you have added complexity and cost and a nonstandard system.

If you use switchmode local converters efficiency is relatively constant with feed voltage, and wiring losses drop as voltage increases due to resistive losses being proportional to current squared. But the converters add complexity and cost. POE (power over ethernet) is one version of this - and the cost per converter is liable to be substantial for off the shelf equipment, and compared to using mains powered commercial supplies.

Option 3:

A single power socket per location allows you to operate either a local 5v PSU OR a mains AC plugboard with a collection of plugpacks.
At the desk where I am typing this I have 18 x 12V powered USB hard drives on a shelf. I've considered the use of a 20A plus 12V supply, custom leads (std socket and cable from HDD to a connection system) but, so far, the 18 x 12V 1A powerpacks plus requisite mains plugboards has won through. Voltage drop is not an issue, I have excellent power supply redundancy either by providing say 2 spare 12V psus or by borrowing a supply from a less critical supply if needed.

My 5V "USB" supply needs are met by the several PCs with USB sockets, plus some of the HDDs have powered 5V outputs. And 5V "USB" psus are very very available, well priced (especially as used ones are usually reliable) and the connector is universal.

And mains AC (230 V 50 Hz in my case) is available throughout the house, or in any home or business I visit.

For me, and probably for you, option 3 makes most sense.

Russell McMahon
  • 147,325
  • 18
  • 210
  • 386
  • what are the drawbacks of using a 12v transformer (w. standard protections) vs. using a dedicated psu on each device – Velimir Tchatchevsky Oct 02 '19 at 05:48
  • 2
    @VelimirTchatchevsky 12V transformer - or more likely a 12V DC psu, is almost as good as separate supplies supply wise but needs DC plugs and leads suited to each device (maybe all 2.5mm power plugs). There is slightly more protection against failure recocvery. Shorting a 12V 10A supply MAY be more exciting than shorting a 12V 1A one. – Russell McMahon Oct 02 '19 at 05:59
  • @VelimirTchatchevsky I edited my Option 2 to make it more general - but my conclusions are much the same. I MAY yet change my many 12V HDD plugpacks to a few larger 12V supplies, but probably not. – Russell McMahon Oct 03 '19 at 00:47
  • I have a 24V DC home network which works just fine (using 4mm^2 / AWG11 cable between floors). Simple (and cheap) local DC-DC converters for 12V and 5V wherever needed work great. 24V DC power supplies (e.g. for DIN rail) are readily available on eBay (e.g. used Phoenix Contact from automation industry) and work like a charm 24/7. – Ned64 Oct 04 '19 at 20:21
  • @Ned64 Yes. If you can and are willing to seek out suitable equipment it can work for you. For interest can you provide a link to the converters that you mention. Comparison of prices with mains to 12V and 5V would be interesting. In the case of my 20 or so USB HDDs that require 12V at about 1A (on average far lower presumably) it would be interesting to see how the costs compared. Most consumer equipment comes with an appropriate mains supply (which is hard to beat on cost :-) ) but comparison at retail prices would be interesting. | My mains works Ok about 8763/8765 of the time - YMMV. – Russell McMahon Oct 05 '19 at 00:19
  • DC-DC converters are abut $1 for XL4015 modules capable of 2A continuous and 5A peak. (from Aliexpress) – Jasen Слава Україні Oct 05 '19 at 07:35
  • @Jasen 1.True (almost). I have some. But, then you'd have to use them :-) . 2. Not a direct comparison with what we started with, but can be made so. Add a case and/or mounting. Add output terminal and/or cord. And input means. If I was using these for 'mission critical" stuff - or just things I didn't want to emit magic smoke, I'd add an output crowbar. As these are variable voltage to suit task crowbar needs to be too. Needs not be complex. TL431, pot, R or few, fuse, crowbar means, mounting. Pricing one's labour at stupidly-little, as one does, a maybe $5 supply ensues. Quite reasonable. – Russell McMahon Oct 05 '19 at 12:34
  • @Jasen These are the types of converters I mean: https://www.ebay.com/itm/173254382932 or https://www.ebay.com/itm/143210607609 (both under $1 inc. worldwide shipping) or https://www.ebay.com/itm/163239615562 with more power. Sealed versions like https://www.ebay.com/itm/302578784673 exist but you need to check them for the usual eBay/Aliexpress/Amazon Marketplace "fake capacity effect" - they may be rated for much more power than they deliver. Test them well upon receipt and use your judgement (heatsink size etc.). – Ned64 Oct 06 '19 at 12:28
10

Voltage drop is the killer

The problem is, voltage drop is extremely acute at very low voltages, because currents are inherently higher, and voltage drop is a function of currents (E=IR). It's a square function against voltage. That means that distributing 5 V instead of 12 V is worse by a factor of 5.76. As a quick example, for a 100' (30 m) run with 3.6% voltage drop:

5 volts @ 12 amperes (60 W) needs 212 kcmil (4/0 AWG)
12 volts @ 5 amperes (60 W) needs 26 kcmil (6 AWG)

Don't "bomp the wah"; actually do the voltage drop calculation

One time I looked at a solar project. 15 A @12-19 V, 14 AWG wire would be fine. Except he was going 200' (60 m). So he gave it a wire size bump, and an extra one or good measure, clear up to 10 AWG wire. Of course, the system was a complete failure and didn't recharge his batteries as calculated. He couldn't understand. "Bat ah bomped the wah!"†

† Translation: "But I bumped up (increased the size of) the wire"

And you armwaved that too. You gave it a perfunctory bump to #10 and went "oughta be enough!" You actually have to go into the voltage-drop calculator and check. This should be based on actual current drawn by the loads on the wires, not breaker or supply rating. Rating a long branch for 20 A when it only needs to carry 2 A is a waste.

I've done bump calculations where it turned out I didn't need the bump.

Comply with Code ... But it's not too bad

You still need to follow the Electrical Codes for installation that is part of the building. Fortunately, the electrical codes are very relaxed for low voltage installations, and further relaxed for <55 W installations.

Neighbor site diy.stackexchange.com is a good one for Code questions.

12 V is so much better that you ought to do it.

As you saw from the 12/5 voltage drop calculation above, 12 V works worlds better. It has voltage drop considerations, still; but they are vastly reduced. Meanwhile, it is highly versatile, and the system can even be extended to be a whole-home backup system.

  • LED 12 V lighting has huge variety and is more dimmable and hackable than mains lighting
  • Most Internet routers and modems can run on it
  • Some TVs can run on it
  • Batteries are readily available for it cheap or in a number of chemistries
  • Most solar charge controllers can supply it
  • which means in a power outage, you have lights and Netflix
  • Some smart doorbells can even run on it.
  • That with a gas-only furnace like an Empire, gas water heater and stove can make a house perfectly livable with the grid down or absent

What's more, getting from 12 V to 5 V is really really not a problem. It is possibly the most commonly sold power converter anywhere!

Peter Mortensen
  • 1,676
  • 3
  • 17
  • 23
Harper - Reinstate Monica
  • 21,639
  • 27
  • 68
  • 13
    Could I suggest changing **"Bat ah bomped the wah"** to whatever you really meant? I googled it with no luck, either its localised jargon, or regional phonetics for "something bump the wire (to the next size)" meaning "round up to the next one and then one further step" – Criggie Oct 03 '19 at 07:15
  • The 12V compromise is a good one as there is so much support equipment freely available and safety is no issue. – KalleMP Oct 03 '19 at 07:21
  • Fixed that @Criggie (In review *8') – Mark Booth Oct 03 '19 at 13:26
  • 1
    Please provide a link to the original source of the images in your answer. We need to be sure that we give proper credit to the creator of those photos. On the other hand, the images don't really add anything to the answer so if you want to avoid "advertising" you could just delete them. – Elliot Alderson Oct 03 '19 at 14:57
  • 2
    @Criggie I approved it myself, well done Mark Booth... – Harper - Reinstate Monica Oct 03 '19 at 15:07
  • @ElliotAlderson changed it to a link. – Harper - Reinstate Monica Oct 03 '19 at 15:12
  • I would think 24v might be better because of the losses and the fact that some electronics require it. Perhaps a system similar to home AC wiring where you have a "Middle" line with 0-12v and 12-24v so you could get 2 12v circuits if you wanted to. – Bill K Oct 04 '19 at 20:02
  • Yah, sure, that's what Edison did. Split-phase actually originates with Edison's DC systems. The only issue is if you have a battery system, balancing parts of the pack may be an issue. LED lighting is readily available at 24V, but devices like routers and TVs may not be able to take it directly. – Harper - Reinstate Monica Oct 04 '19 at 20:15
0

It depends on the distance as others mentioned. We actually ran a RS485 network with a master powering slaves 20 feet apart in a star configuration.

We could easily run with a 5v 3A on the master, since the slaves were needing only 3.3v which was managed using a simple voltage regulator.

And we used cheap twisted electrical wires to reduce cost for a multi-site implementation.

Arundale Ramanathan
  • 131
  • 7