Daniel sent us this one — he's been building that LoRa parking occupancy system, and he already wrestled with the five volt versus twelve volt question. Now he's asking about the next rung up: twenty-four volts and forty-eight volts DC. What hardware actually demands these voltages, how do you distribute power from AC mains or batteries at those levels, what specific components should you know about, what are the safety concerns, and what other DC rail standards exist that we haven't covered yet. It's basically a request for a field guide to middle-voltage DC.
It's a great question because these two voltages sit in this weird no-man's-land between the hobbyist bench supply world and actual mains wiring. Most people building Arduino projects never touch them. But the moment you start doing anything industrial, anything with long cable runs, or anything that draws real power, twenty-four and forty-eight volts are suddenly everywhere.
The parking project is the perfect entry point for this. He's got twelve volt backup beepers and warning sensors. But what if he wanted to add a proper industrial proximity sensor fifty meters out at the entrance? Twelve volts over that distance just evaporates.
Let's run the numbers. A twelve volt system pulling five amps over fifty meters of eighteen AWG wire — that's a voltage drop of over eight volts. You're losing two-thirds of your voltage before it even reaches the sensor. The thing won't even turn on. But a forty-eight volt system carrying the same power — now you're at one and a quarter amps, and the drop is only about two volts. Four percent loss instead of sixty-seven percent. That's not a minor improvement, that's the difference between a working system and a paperweight.
This is really about solving two problems at once: longer cable runs and higher power delivery, while staying under a regulatory boundary that most people don't even know exists.
Right, and that boundary is the key to why these specific voltages exist in the first place. Under IEC six two three six eight dash one and UL six zero nine five zero dash one, forty-eight volts DC is the highest voltage classified as extra-low voltage — ELV. Stay at or below forty-eight volts, and you're exempt from a whole raft of regulatory requirements that kick in for mains-voltage equipment. You don't need the same insulation, the same clearance distances, the same safety certifications. It's the sweet spot where you can deliver serious power without entering the regulatory regime of line voltage.
Forty-eight volts is basically the maximum voltage you can use while still being able to tell the inspector "don't worry about it.
That's the one-sentence summary, yes. And twenty-four volts is the industrial workhorse that sits comfortably below that. So let's map out what actually lives at each voltage. Twenty-four volts DC is the standard for industrial control per IEC six one one three one dash two. It's what programmable logic controllers expect on their digital inputs, it's what proximity sensors and pressure transmitters and flow meters run on, it's what solenoid valves and motor contactors and LED machine lighting use. Walk onto any factory floor and look inside the control cabinets — you'll see rows of twenty-four volt DIN-rail power supplies feeding terminal blocks with blue and white-with-blue-stripe wiring.
Why did twenty-four volts become the industrial standard specifically?
It's a Goldilocks voltage. High enough to drive relays and small motors and overcome noise in electrically harsh environments — factories are full of variable frequency drives throwing off electromagnetic interference — but low enough to be touch-safe under SELV, which is Separated Extra-Low Voltage. You can put your fingers across twenty-four volts with dry skin and feel nothing. It's also high enough that a ten to thirty meter cable run from the control cabinet to a sensor doesn't cause problematic voltage drop. And it divides nicely from common battery configurations: two twelve-volt lead-acid batteries in series gives you twenty-four volts for backup.
What specific hardware are we talking about? Give me the shopping list.
The most common thing you'll encounter is the DIN-rail power supply. The Mean Well LRS-one-fifty-twenty-four is practically the default choice — a hundred and fifty watts, twenty-four volts output, about the size of a paperback book, snaps onto a DIN rail. Mean Well sells over a million units a year across the LRS series. You wire AC mains into the top, and you get twenty-four volts DC out the bottom into terminal blocks. From there you daisy-chain to sensors, PLC input cards, relay coils, whatever. Then you've got the sensors themselves: inductive proximity sensors — the little threaded barrel things that detect metal — those are almost universally twenty-four volts. Photoelectric sensors, ultrasonic distance sensors, pressure transmitters with four-to-twenty-milliamp loops, they all expect twenty-four volt supply.
The four-to-twenty-milliamp thing is worth flagging because that's how industrial sensors communicate — they don't output a voltage, they vary the current in the loop. A lot of hobbyists have never encountered that.
In the industrial world, current loops are preferred because they're immune to voltage drop over long cables. A four-to-twenty-milliamp signal is the same at both ends regardless of wire resistance. But the sensor itself still needs a supply voltage to power its internal electronics, and that's almost always twenty-four volts DC. You also see twenty-four volts in LED strip lighting for commercial applications — the longer runs you can achieve versus twelve volt strips before you get visible dimming at the far end. And solenoid valves for pneumatics and hydraulics, motor drivers for small DC motors, even some 3D printer heated beds are moving to twenty-four volts because they can deliver more power without melting connectors.
That's the twenty-four volt world. What about forty-eight?
Forty-eight volts owns two domains: telecom and Power over Ethernet. In telecom, the standard is actually negative forty-eight volts DC with positive ground. And that negative sign isn't a typo.
This is one of those historical decisions that's still with us a century later. Telephone exchanges have used negative forty-eight volts since the early days of the Bell System. The positive ground configuration — where the positive terminal of the battery bank is connected to earth — was chosen to reduce electrolytic corrosion on copper conductors in underground cables. If you have a negative voltage on the line relative to ground, any moisture that gets into the cable causes metal ions to migrate from the ground rod to the copper conductor, rather than the other way around. It's sacrificial protection — your ground rods corrode instead of your expensive buried copper.
That is genuinely clever. A century-old corrosion hack baked into the voltage standard.
It persists today in every central office, every cell tower base station, every telecom rectifier shelf. If you walk into a telecom facility, you'll see racks of forty-eight volt rectifiers converting AC mains to negative forty-eight volts DC, feeding massive battery banks for backup. The batteries are typically twenty-four lead-acid cells in series — each cell is nominally two volts, so twenty-four cells gives you forty-eight volts. During a power outage, the batteries carry the load seamlessly because the equipment is already running on DC.
Then there's PoE.
Power over Ethernet is where most people encounter forty-eight volts without realizing it. The IEEE eight zero two dot three bt standard from twenty eighteen — that's PoE plus plus — delivers up to ninety watts at nominally forty-eight volts over four twisted pairs of Ethernet cable. Some injectors push to fifty-six volts to compensate for cable drop, but the nominal system voltage is forty-eight. This is what powers pan-tilt-zoom security cameras, video phones, thin clients, LED lighting fixtures, and increasingly building automation sensors. A hundred meters of Cat6 cable carrying forty-eight volts at over an amp — that's real power delivery over data cabling, and it works because the higher voltage keeps the current manageable.
Let me do the math on that. A PoE plus plus camera drawing sixty watts at forty-eight volts is pulling one and a quarter amps. If you tried to deliver sixty watts at twelve volts, you'd need five amps, and over a hundred meters of twenty-four AWG, the voltage drop would be catastrophic.
More than eight volts lost. The camera would see less than four volts. It's not even a question of efficiency at that point — the thing simply wouldn't power up. This is why PoE settled on forty-eight volts. It's the highest voltage that stays under the sixty-volt DC ELV boundary, which means standard RJ45 connectors and unshielded twisted pair can be used without the safety requirements that would apply at higher voltages.
We've got twenty-four volts for the factory floor and forty-eight volts for the server closet and the cell tower. What about actually getting these voltages from a wall outlet or a battery?
For twenty-four volts, the most common approach in industrial settings is the DIN-rail power supply I mentioned. You bring AC mains into the panel, through a circuit breaker, into the supply, and you get twenty-four volts DC out. From there you distribute via terminal blocks. For battery backup, you'd typically use two twelve-volt batteries in series — either lead-acid or LiFePO4 — with a charger that maintains float voltage and an automatic transfer switch or diode-OR arrangement so the batteries take over when AC fails.
The centralized versus distributed debate?
This is a real design decision. One big twenty-four volt supply in a central cabinet feeding everything means you only have one AC connection to worry about, one point of failure, and simpler wiring. But your cable runs can get long, and a short circuit anywhere can take down the whole system. Multiple smaller supplies distributed near the loads means shorter cable runs, better fault isolation, but more AC wiring to route and more points of maintenance. For most projects under a few hundred watts, centralized is simpler. For a factory with dozens of machines, distributed wins.
For forty-eight volts?
The telecom approach is a forty-eight volt rectifier shelf — these are purpose-built units that convert AC to negative forty-eight volts DC with battery charging capability. Companies like Eltek and Delta make these. For PoE, you use a PoE injector or a PoE switch — the switch handles the negotiation with the powered device to determine how much power to deliver. For off-grid solar, you'd use a solar charge controller configured for a forty-eight volt battery bank — typically four twelve-volt batteries in series, or a single forty-eight volt LiFePO4 rack battery. These are becoming very common in the off-grid and backup power world because forty-eight volts at high capacity means lower current, thinner cables, and less heat.
The Mean Well forty-eight volt equivalent to that LRS series — what's the comparable unit?
The RPS-two-hundred-forty-eight — two hundred watts, forty-eight volts out, similar form factor, about thirty percent more expensive. But the interesting thing is that the forty-eight volt DIN-rail supply market is much smaller than twenty-four volts, because forty-eight volts in industrial settings usually comes from PoE infrastructure or dedicated telecom rectifiers rather than general-purpose DIN supplies.
If you're speccing a project, the ecosystem around each voltage is different. Twenty-four volts has this enormous industrial components catalog — sensors, relays, terminal blocks, supplies. Forty-eight volts has the networking and telecom ecosystem.
And that should drive your choice more than any abstract voltage preference. What hardware do you actually need to connect to?
Let's talk about safety, because I think there's a dangerous assumption that anything below mains voltage is harmless.
This is where people get complacent. Twenty-four volts DC with dry skin — you'll feel nothing. The body's resistance is high enough that the current flowing is below the perception threshold. But get your hands wet, or have a cut on your finger, and suddenly the skin resistance drops dramatically. A high-current twenty-four volt supply — and some industrial units can deliver over a hundred amps — can absolutely deliver a dangerous shock under those conditions. It's not the voltage that kills you, it's the current through your heart, and wet skin can drop your body resistance from hundreds of thousands of ohms to a few thousand.
Forty-eight volts is closer to the edge.
Forty-eight volts is the voltage where you start to feel it even with dry skin for some people — a tingle, maybe a mild zap. With wet hands, it can be a painful shock. There's a reason it's the ELV ceiling. The standards bodies determined that below sixty volts DC, the risk of fibrillation under normal conditions is acceptably low, but it's not zero. And with high-capacitance sources — think a large battery bank with thousands of amp-hours behind it — the available fault current is enormous.
Which brings us to fusing.
This is a gotcha that trips up people coming from the AC world. AC-rated fuses and circuit breakers may not safely interrupt a DC fault, especially at forty-eight volts. The problem is arc extinction. With AC, the current crosses zero a hundred or a hundred and twenty times per second, and the arc naturally extinguishes at the zero crossing. With DC, the arc is continuous. A forty-eight volt DC arc is harder to quench than a twenty-four volt one, and an AC-rated breaker might just sit there arcing while your wiring melts. You need DC-rated overcurrent protection — fuses or breakers specifically tested and labeled for the DC voltage you're using.
I've seen photos of what happens when someone uses an AC breaker on a DC battery bank. It's not subtle.
The breaker welds itself closed, or the arc sustains until something else fails catastrophically. For a forty-eight volt battery bank, you want fuses rated for at least the full battery voltage — and ideally with a high interrupt rating, ten thousand amps or more, because lithium batteries can deliver staggering short-circuit currents.
What about arc flash? That's usually a mains-voltage concern.
At forty-eight volts, arc flash risk is low but not zero. If you have a large battery bank with low internal resistance — lithium iron phosphate packs can have internal resistance in the single-digit milliohms — and you accidentally short the terminals with a tool, you can get an arc that vaporizes metal. It won't be the explosive arc flash you'd see at four hundred eighty volts, but it can still cause burns and eye damage. Treat forty-eight volt battery banks with respect.
The safety summary: twenty-four volts is benign under most conditions but not all, forty-eight volts demands real caution, and both require DC-rated protection.
That's it. And one more thing — grounding. Twenty-four volt industrial systems almost universally use negative ground. The negative terminal of the supply is bonded to earth. Forty-eight volt telecom uses positive ground for the corrosion reason we discussed. If you're mixing equipment from both worlds, you need to be aware of this. Connecting a negative-grounded twenty-four volt device to a positive-grounded forty-eight volt system without isolation will create a dead short through the ground path.
That's the kind of mistake you make exactly once.
Then you spend a week debugging why everything is fried.
Let's zoom out. Daniel also asked about other DC rail standards we haven't covered. We've talked five volts, twelve volts, twenty-four, forty-eight. What else is out there that someone building electronics should know about?
Let me do a rapid tour. Three point three volts — this is the logic level for most modern microcontrollers, FPGAs, and SDRAM. It replaced five volts as the dominant digital logic voltage because smaller transistor geometries can't handle five volts. Below that, you've got one point eight volts, one point two volts, and even zero point nine volts — these are core voltages for modern CPUs and GPUs. The voltage is tiny, but the current is enormous — a high-end GPU can draw hundreds of amps at less than one volt. The regulation has to be incredibly tight, within a few percent, because at those voltages even a small drop causes instability.
The opposite end of the spectrum from what we've been discussing — ultra-low voltage, ultra-high current.
Then you've got plus and minus fifteen volts — this is the classic analog op-amp supply. Precision instrumentation, audio equipment, analog signal processing. The split rail gives you a signal swing that can go both positive and negative relative to ground, which is essential for AC signals. Five volts is still everywhere — USB legacy, logic in older systems, single-board computers like the Raspberry Pi. Twelve volts is automotive, PC power supplies, and Daniel's parking project. Fifty-six volts shows up in some PoE injectors that boost the voltage to compensate for cable drop — it's still within the sixty-volt ELV ceiling but gives more headroom.
Then there's the one that surprised me when I first heard about it: three hundred eighty volts DC.
This is the emerging standard for data center power distribution under IEC six two zero four zero dash five dash one. The idea is that instead of distributing forty-eight volts to server racks and converting down at each rack, you distribute three hundred eighty volts DC — which is roughly the rectified equivalent of two hundred seventy-seven volts AC — and convert down at the server power supply. You save two to three percent in conversion losses compared to forty-eight volt distribution, which doesn't sound like much until you realize a hyperscale data center might draw a hundred megawatts. Three percent of a hundred megawatts is three megawatts — that's real money and real cooling load.
It's DC end to end, no AC rectification stage at each rack.
The servers still need their internal DC-DC converters to step down to the various voltages they use, but you eliminate the AC-to-DC rectification stage that would otherwise be in every power supply. The tradeoff is that three hundred eighty volts DC requires completely different safety practices — it's well above ELV, so you need arc-flash protection, insulated tools, different connector standards. It's not something a hobbyist is going to touch, but it's where the industry is heading for high-density computing.
We've got this whole landscape from zero point nine volts to three hundred eighty volts DC. But for someone like Daniel building a real project, the decision usually comes down to twelve, twenty-four, or forty-eight.
I'd argue there's a decision framework here. First, look at your cable runs. If anything is over thirty meters, forty-eight volts is the clear winner — the voltage drop math is unforgiving. A forty-eight volt system carrying five amps over fifty meters of eighteen AWG drops about two volts, or four point two percent. A twenty-four volt system at ten amps — same power — over the same cable drops about four volts, or sixteen point seven percent. That's the difference between a sensor that works and one that's flaky.
The second factor?
Above about a hundred watts, forty-eight volts starts making more sense because the lower current means less I-squared-R loss, thinner cables, and less heat at connectors. Below fifty watts, twenty-four volts is perfectly fine and gives you access to that huge industrial sensor ecosystem. Third factor: what hardware do you actually need? If you're using industrial proximity sensors and PLCs, you're in the twenty-four volt world whether you like it or not. If you're powering PoE cameras and networking gear, you're at forty-eight volts.
Twenty-four volts maps nicely to two twelve-volt batteries in series — automotive, marine, common lead-acid. Forty-eight volts maps to four twelve-volt batteries or dedicated forty-eight volt lithium racks, which are becoming standard in solar and telecom. If you're doing an off-grid solar system, forty-eight volts is increasingly the default because the higher voltage means lower current from the panels to the charge controller.
One thing you mentioned earlier — the efficiency crossover. Higher voltage means lower I-squared-R losses but higher switching losses in DC-DC converters. Where's the sweet spot?
For most DC-DC converter topologies, the crossover is around forty-eight volts. Below that, the conduction losses dominate. Above that, the switching losses — the energy lost every time the MOSFET turns on and off — start to eat into your gains. This is one reason forty-eight volts became the telecom standard: it's roughly the point where the efficiency curve peaks for the converter technology of the era, and it's stayed there because the physics hasn't changed.
Forty-eight volts isn't just a regulatory sweet spot — it's also an efficiency sweet spot.
It's one of those rare cases where regulation and physics align. The ELV ceiling at sixty volts DC, the efficiency optimum around forty-eight volts, and the corrosion chemistry that favors positive ground — they all point to the same number. It's almost elegant.
Until you accidentally connect a negative-ground device to a positive-ground bus and let the smoke out.
Engineering elegance always has a trapdoor somewhere.
Let's bring this back to Daniel's parking project. He's got twelve volts for his backup beepers and warning sensors. If he wanted to add twenty-four volt industrial sensors — say, a proper inductive loop detector for vehicle presence — what's the cleanest way to do that?
He has two good options. One is a dual-output supply like the Mean Well RD-sixty-five-A, which gives you both five and twelve volts from a single AC input. But for twelve and twenty-four, he'd want something like the Mean Well RD dash eighty-five — that's eighty-five watts with dual outputs. Or he could use separate DIN supplies, one twelve volt, one twenty-four volt, fed from the same AC circuit. The dual-rail approach is cleaner than trying to step twelve up to twenty-four or twenty-four down to twelve for everything, because every conversion stage costs you efficiency and adds failure points.
If he wants to run a PoE camera at the far end of the parking lot?
That's where forty-eight volt PoE shines. A single Cat6 cable carrying both data and power, a hundred meters no problem. He'd need a PoE switch or injector, and a PoE splitter at the camera end if the camera doesn't natively support PoE. The splitter takes forty-eight volts in and gives you the voltage the camera actually needs — often twelve volts or five volts — plus Ethernet data.
His system could end up being a hybrid: twelve volts for the short-range beepers and sensors, twenty-four volts for industrial detection, forty-eight volt PoE for the far-end camera. All from one AC feed, with appropriate supplies for each rail.
That's actually a very common real-world architecture. Nobody says you have to pick one voltage. The key is knowing what each rail is good at and matching the hardware to the voltage that makes sense for its location and power requirement.
What about connectors? That's a practical detail that bites people.
Twenty-four volt industrial uses M12 circular connectors a lot — they're rugged, IP-rated, and standardized for sensor connections. Terminal blocks with screw or spring clamps are also universal. For forty-eight volt PoE, it's RJ45 with integrated magnetics — the Ethernet connector handles both data and power. For forty-eight volt DC distribution outside of PoE, you'll see Anderson Powerpole connectors, which are genderless and rated for high current, or terminal blocks with higher voltage ratings. The key thing: don't use audio connectors or barrel jacks for forty-eight volts. They're not rated for it, and the arc when you unplug under load can damage the connector.
The barrel jack arcing problem is something I've seen people discover the hard way.
Unplugging a forty-eight volt DC load from a barrel jack draws an arc between the pin and the sleeve as they separate. With AC, the arc extinguishes at the zero crossing. With DC, it just keeps going until the gap is wide enough. Over time, this pits the contacts, increases resistance, and eventually the connector fails — sometimes melting in the process. DC-rated connectors have features to manage this, like faster separation or arc-quenching designs.
The connector choice is part of the safety design, not just a convenience.
And that ties back to the DC fuse issue we mentioned. A system is only as safe as its weakest component. You can have the best power supply in the world, but if you fuse it with an AC-rated breaker and use barrel jacks not rated for DC, you've built a fire hazard.
Let's hit a few misconceptions before we wrap up. One I hear a lot: twenty-four and forty-eight volts are interchangeable, just use a different power supply.
The connectors are different, the fusing requirements are different, the grounding conventions are different — remember positive ground in telecom — and the hardware ecosystem is different. You can't just swap a forty-eight volt supply into a twenty-four volt system and expect it to work. The sensors won't accept the voltage, the fuses won't be rated correctly, and if the system was designed for negative ground and you connect a positive-ground telecom rectifier, you've got a short.
Another one: forty-eight volts is completely safe to touch.
Dry skin, casual contact — you'll probably feel nothing or a mild tingle. But wet conditions, broken skin, or a large contact area change the equation. And from a high-capacity battery bank, the available fault current is terrifying. Treat forty-eight volts with the same respect you'd give mains voltage in terms of not working on it live, using insulated tools, and making sure overcurrent protection is properly sized.
The third one: higher voltage always means higher efficiency.
True for distribution losses — I-squared-R goes down as voltage goes up for the same power. But false at the system level because converter switching losses increase with voltage. The crossover is around forty-eight volts for most topologies. Above that, the gains from lower current are offset by the increased switching losses. This is why three hundred eighty volt DC data center distribution is only worth it at very high power levels where the distribution losses dominate.
All of that theory is great, but let's bring it back to something someone can use next week. What are the actionable takeaways?
First: calculate your voltage drop budget before you choose a voltage. The formula is straightforward — voltage drop equals two times the cable length times the current times the resistance per meter of your wire gauge. If the drop exceeds five percent of your nominal voltage, step up to the next standard voltage. Twelve to twenty-four, twenty-four to forty-eight. This one calculation at the start of a project can save you from a system that works on the bench and fails in the field.
There are online calculators that make this trivial.
Plenty of them. Second takeaway: always use DC-rated fuses and breakers for twenty-four and forty-eight volt circuits. Look for the DC voltage rating on the component — it'll usually say something like "one hundred twenty-five volts AC / forty-eight volts DC" or similar. If it only has an AC rating, don't use it on DC. The extra cost of a proper DC-rated fuse is nothing compared to the cost of a fire.
The third one — this is specific to Daniel's project but applies generally — consider a dual-rail approach rather than forcing everything to one voltage.
If you've got twelve volt beepers and twenty-four volt sensors, run both rails from a dual-output supply or separate supplies on the same AC circuit. Don't try to step twelve up to twenty-four with a boost converter for high-power loads — it's inefficient and adds a failure point. And don't try to run everything at twelve volts if you've got long cable runs — the voltage drop will bite you. Match the voltage to the load and the distance.
For most new projects, the default is: twenty-four volts if you need industrial sensors and your runs are under thirty meters. Forty-eight volts if you need high power or long cables. Twelve volts if you're in the automotive or hobbyist ecosystem and everything is within a few meters.
That's the heuristic. And five volts if you're powering microcontrollers and logic, but you already knew that.
One open question before we close. USB-C PD now goes up to forty-eight volts at two hundred forty watts. And we've got three hundred eighty volt DC creeping into data centers. Do twenty-four and forty-eight volts remain distinct standards, or do they converge over the next decade?
I think forty-eight volts becomes the universal high-power DC standard, and twenty-four volts remains the industrial sensor voltage. The reason is ecosystem lock-in. There are millions of twenty-four volt sensors and PLCs installed worldwide, and nobody is going to retrofit them. But for new designs where higher power is needed — mild hybrid vehicles, data center racks, off-grid power systems — forty-eight volts is becoming the default. The automotive industry's move to forty-eight volt mild hybrid systems is a huge driver. And USB-C PD at forty-eight volts means consumer electronics will increasingly speak that voltage too.
Twenty-four volts is the past and present of industrial automation. Forty-eight volts is the future of everything else.
I'd put it that way. And the smart move for anyone building systems today is to be comfortable with both.
Now: Hilbert's daily fun fact.
Hilbert: In the seventeen eighties, Roman provincial road maintenance in Britannia was funded by a mandatory labor tax called the corvée, but the actual work was so consistently shirked that local governors began paying citizens to not show up — a de facto bribe to avoid the administrative nightmare of forcing them — effectively inverting the tax into a welfare program for the idle.
The Romans invented paying people to do nothing. Truly ahead of their time.
a lot to sit with.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If you enjoyed this, do us a favor and leave a review wherever you get your podcasts — it helps. We'll be back next week with whatever Daniel throws at us.
