Before we jump in — quick note. DeepSeek V four Pro is writing our script today. Which is appropriate, given we're about to talk about powering things through clever engineering.
I don't know, it felt like a segue. It wasn't.
Alright — so Daniel sent us this one, and it's a layered question born directly out of the war with Iran and the ceasefire that followed. He was sheltering with about a hundred people in a public shelter, concrete overhead, ballistic missiles incoming, and nobody had connectivity. The Home Front Command started issuing stricter warnings because Iran was using cluster munitions — unexploded ordnance, fragments everywhere — and the standing order became: do not leave the shelter until you get the official all-clear. But the all-clear arrives via cell signal. And concrete shelters are basically Faraday cages. So you've got this absurd situation where following the safety instructions requires receiving a message you structurally cannot receive.
Daniel's point about the mass effect is real. When a hundred people are packed into a shelter, everyone's phone is frantically pinging the same tower, and the tower's already struggling to punch through all that reinforced concrete. It's a congestion collapse on top of a physical barrier problem.
So his wife asks him whether any of the weird gear he's bought over the years would actually work. He realizes he has a cellular router that would be perfect, except it needs mains power. Two hundred thirty volts. So between rounds of conflict, he orders a mini UPS from AliExpress, opens it up, and finds it takes four 18650 batteries. And the core question is: how do four little batteries produce two hundred thirty volts? Because if you just divide two thirty by the nominal voltage of an 18650 cell, you'd need something like sixty-five cells. So what's actually happening inside that box?
This is one of those questions where the answer reveals something elegant about how we manipulate electricity. The short answer is: you don't get two hundred thirty volts by stacking batteries in series. You get it by switching.
Walk me through that.
Alright, so let's start with what an 18650 cell actually is. It's a lithium-ion cylindrical cell, eighteen millimeters in diameter, sixty-five millimeters long. Nominal voltage is about three point six or three point seven volts. Fully charged, around four point two. Four of these in series gives you roughly fourteen point eight volts nominal. Parallel them, and you still have three point seven volts but more current capacity. Either way, you're nowhere near two hundred thirty.
The battery pack itself is low-voltage DC.
And the magic component is the inverter. Specifically, a switch-mode inverter. It takes that low-voltage DC, chops it up into high-frequency pulses, runs those through a transformer to step up the voltage, and then shapes the output into something that looks like the AC sine wave your router expects.
Chop it up. Let's get specific. What's actually happening at the circuit level?
You've got your DC input — let's say fourteen volts from those four cells in series. That hits a set of switching transistors, usually MOSFETs, which turn on and off at something like twenty to a hundred kilohertz. What you get out of that is not smooth DC anymore — it's a series of pulses, still at fourteen volts peak, but now it's a high-frequency square wave.
The high frequency is the key, because transformers care about frequency.
A transformer works by magnetic coupling between two coils of wire. The ratio of turns between the primary and secondary coils determines the voltage ratio. If you have ten turns on the primary and a thousand on the secondary, you get a hundred-to-one voltage step-up. But transformers only work with changing magnetic fields. DC just saturates the core and does nothing useful. The faster the change, the smaller and lighter the transformer can be for a given power level.
By chopping the DC into a hundred-kilohertz pulse train, you can use a tiny transformer instead of the fifty-pound iron brick you'd need at fifty or sixty hertz mains frequency.
And that's the entire revolution of switch-mode power supplies. Back in the nineteen fifties, if you wanted to convert voltages, you used a linear power supply with a massive iron-core transformer running at mains frequency. Heavy, inefficient, hot. The switch-mode design lets you shrink the transformer by orders of magnitude. That's why your phone charger is a little cube instead of a brick. And it's why Daniel's mini UPS can fit in a small box.
The flow is: batteries provide low-voltage DC, MOSFETs chop it into high-frequency AC, a tiny step-up transformer multiplies the voltage, and then what? You've got high-voltage high-frequency AC, but the router wants fifty-hertz sine wave at two thirty volts.
Right, so after the transformer you've got what's called the DC link — a rectifier and capacitor that convert that stepped-up high-frequency AC back to high-voltage DC. Now you're sitting at something like three hundred twenty volts DC. Then comes the output stage, which is another set of switches that reconstruct a fifty-hertz sine wave from that DC bus.
Reconstruct a sine wave. That's where the difference between modified sine wave and pure sine wave inverters comes in, right?
A modified sine wave inverter just switches between positive DC, zero, negative DC, zero — it's a stepped square wave. It's simpler and cheaper, but some devices don't like it. Motors hum, some power supplies run hot, and certain electronics with zero-crossing detection circuits get confused. A pure sine wave inverter uses pulse-width modulation — it switches very fast, varying the duty cycle to shape a waveform that's essentially indistinguishable from what you get out of the wall. It's more complex but runs anything.
For something like a cellular router, which has its own switch-mode power supply inside that converts AC back to DC anyway?
Most routers are actually pretty tolerant of modified sine wave. Their internal power supplies are switch-mode, and the first thing they do is rectify the AC to DC and chop it again. So the waveform quality matters less than you'd think. But I wouldn't run sensitive audio equipment or medical devices on modified sine wave.
Let's tie this back to Daniel's actual question. He's looking at four 18650 cells. How much runtime does that actually give him on a cellular router?
Let's do the numbers. A typical 18650 cell from a reputable manufacturer — say a Samsung 35E or a Panasonic NCR18650B — has a capacity of about thirty-four hundred milliamp-hours. At three point seven volts nominal, that's about twelve point six watt-hours per cell. Four of them gives you roughly fifty watt-hours total.
A typical LTE router — something like a Teltonika RUT series or a MikroTik wAP LTE — draws maybe five to ten watts under normal operation. Let's call it seven watts average. Fifty watt-hours divided by seven watts gives you about seven hours of runtime. But that's theoretical. In practice, the inverter has efficiency losses — maybe ten to fifteen percent. The batteries might not be at full rated capacity, especially if they're from AliExpress. And the router's power draw can spike during data transmission.
Realistically, somewhere between four and six hours.
Which is actually pretty good for something that fits in your hand. And Daniel's right — routers are ideal candidates for this kind of backup because they're low-draw devices. The engineering challenge isn't delivering a lot of power. It's delivering the right voltage at a modest current for a useful duration.
Now there's another dimension here that Daniel touched on — the SMS versus data connection question during an emergency. He pointed out that an all-clear message is a few kilobytes of text, and yet even that wasn't getting through. What's actually happening when a hundred people are competing for signal in a concrete box?
There are really two separate problems. One is the physical barrier — reinforced concrete attenuates radio signals dramatically. The higher the frequency, the worse the penetration. Modern cellular networks in Israel operate on bands ranging from seven hundred megahertz up to two point six gigahertz for 4G. The lower bands penetrate better, but they're still severely attenuated by concrete and rebar.
The rebar essentially forms a mesh — a partial Faraday cage.
If the rebar spacing is smaller than the wavelength of the signal, you get significant reflection and attenuation. At two gigahertz, the wavelength is about fifteen centimeters. Rebar grids in structural concrete are typically spaced fifteen to thirty centimeters apart. So you're right at the threshold where the concrete structure starts behaving as a partial shield.
Even before we talk about network congestion, the signal physically reaching the phones is already marginal.
Then you add a hundred phones, all with weak signals. A phone with a weak signal has to transmit at higher power to reach the tower. That increases the noise floor for every other phone nearby. The cell tower's receiver has to work harder to distinguish each phone's signal. As more phones connect, the tower allocates less time and fewer resource blocks to each one. Eventually, the control channel — the part of the signal that handles connection setup and SMS delivery — gets congested.
This is where Daniel's point about SMS being theoretically more resilient is interesting but also incomplete. Yes, SMS uses the control channel and requires almost no bandwidth. It's a store-and-forward system that can wait for a brief connection window. But if the control channel itself is overwhelmed, even an SMS can't get through.
SMS doesn't require a persistent data connection. It doesn't need a TCP handshake. It's about as lightweight as a cellular message can be. But it still requires your phone to successfully register with the tower, authenticate, and receive the paging message that says "you have an SMS waiting." If the paging channel is saturated, or if your phone can't complete the registration handshake because the signal is too weak and the tower is too busy, the SMS just sits in the network's message center waiting for a delivery window that never opens.
Daniel mentioned Israel's emergency wireless alert system. That's Cell Broadcast, right? Different from SMS.
Cell Broadcast is a completely different mechanism, and it's actually much more robust than SMS for mass alerts. Instead of sending individual messages to each phone, the tower broadcasts a single message that any compatible phone in range can receive. It's one-to-many, not one-to-one. It doesn't require the phone to be registered or authenticated. It just needs to be listening. And Cell Broadcast messages can be assigned a high priority that forces the phone to process them immediately.
Why did that fail in the shelters Daniel was in?
Cell Broadcast still requires the phone to receive the signal. If you're behind enough concrete, the broadcast simply doesn't reach you. The tower can shout as loud as it wants, but if you're in a radio shadow, you won't hear it. And Cell Broadcast is typically sent on the paging channel, which can be congested in an emergency scenario. Plus, not all phones are configured correctly for Cell Broadcast reception. Some carriers disable it by default. It's not as universally reliable as it should be.
Daniel's core frustration — that nobody prepared the shelters for this — is actually a systems integration failure. The civil defense authority knows about Cell Broadcast. The Home Front Command knows about shelter construction standards. The cellular carriers know about capacity planning. But nobody connected the dots: if you tell people to shelter in reinforced concrete, and your communication strategy assumes cellular connectivity, you've created a contradiction.
It's not a new problem. Israel's been dealing with rocket attacks for decades. The Home Front Command has detailed specifications for shelter construction — concrete thickness, rebar spacing, blast doors, ventilation. But there's no requirement for signal repeaters or distributed antenna systems in public shelters. It's not part of the building code.
Which brings us back to Daniel's practical solution. If the shelter doesn't provide connectivity infrastructure, you bring your own. A cellular router with an external antenna that you can place outside the shelter, connected by a cable to the router inside, powered by a battery backup that can run for hours. It's a personal micro-cell-site.
The antenna connection he mentioned — SMA — that's crucial. SMA connectors are standard for cellular antennas. You can get a directional antenna, mount it on a pole, point it at the nearest tower, and run the cable through a ventilation shaft or a gap in the door. The router sits inside the shelter, broadcasting Wi-Fi to the people around you.
Now you're the one person in the shelter with connectivity, and you can share the all-clear message with everyone else.
Which is exactly the scenario Daniel was describing where one person gets the message and it turns out to be old. If you're the connectivity hub, you need to make sure you're pulling fresh information, not cached data. That's a software problem on top of the hardware problem.
Let's go back to the battery engineering, because I want to understand the limitations. Daniel's mini UPS uses four 18650 cells. We've established that the inverter steps up the voltage. But what about the current? If you're stepping up from fourteen volts to two hundred thirty volts — roughly a sixteen-to-one voltage ratio — the current on the battery side has to be at least sixteen times higher than what you're delivering on the AC side, plus efficiency losses.
And it's one of the constraints that limits how much power these small units can deliver. If your router draws point zero five amps at two hundred thirty volts — about eleven and a half watts — then on the battery side, at fourteen volts, you need to supply at least zero point eight two amps. In practice, with inverter efficiency around eighty-five percent, you're looking at closer to one amp. That's well within what an 18650 cell can handle — most are rated for continuous discharge of five to ten amps.
If you tried to run something hungrier — say a laptop charger pulling sixty-five watts — now you're drawing four to five amps from the battery pack. Still doable but getting warmer. And if you tried to run a kettle?
A kettle pulls fifteen hundred to two thousand watts. At fourteen volts, that's over a hundred amps. Your 18650 cells would sag dramatically, the protection circuit would probably trip, and if it didn't, you'd have a thermal event on your hands. These mini UPS units are designed for low-power loads. Routers, modems, small switches. They're not general-purpose backup power.
The engineering is clever, but the use case is narrow. Which is fine, because Daniel's use case fits perfectly.
It's worth noting that these AliExpress mini UPS boards often use a chip like the XL6019 or the MT3608 for the boost stage, and a dedicated inverter chip like the EG8010 for the sine wave generation. These are mass-produced, well-understood designs. They're not cutting-edge, but they're reliable enough for the price.
Let's talk about the batteries themselves for a moment. Daniel said he ordered 18650 cells. There's a whole rabbit hole there about authenticity, capacity ratings, and safety.
The 18650 market is flooded with counterfeits and exaggerated ratings. You'll see cells advertised as nine thousand nine hundred milliamp-hours — that's physically impossible in an 18650 form factor. The highest genuine capacity currently available is around thirty-five hundred milliamp-hours, from cells like the Samsung 50E or the LG M50T. Anything claiming more than that is a lie.
Then there's the chemistry question. Not all 18650s are the same.
The most common chemistries are ICR — lithium cobalt oxide — which has high capacity but is less stable and requires a protection circuit. IMR — lithium manganese oxide — which has lower internal resistance and can deliver higher current, but lower capacity. And INR — lithium nickel manganese cobalt oxide — which is a hybrid, balancing capacity and discharge rate. For a UPS application, you probably want INR or protected ICR cells. And you absolutely want cells from a known manufacturer — Samsung, LG, Panasonic, Sony, Molicel. Not the no-name cells with the rainbow wraps and the impossible capacity claims.
There's a safety dimension here that's especially relevant in a conflict zone. Lithium-ion cells can fail violently if they're overcharged, over-discharged, short-circuited, or physically damaged. In a shelter scenario, the last thing you want is a battery fire.
Which is why you want protected cells — they have a small circuit board built into the negative end that cuts off the cell if the voltage goes too high, too low, or if the current draw exceeds a safe threshold. And you want a charger that does proper CC-CV charging — constant current until the voltage reaches four point two volts, then constant voltage until the current drops to near zero. The cheap chargers often skip the CV phase or do it poorly, which degrades the cells over time.
Daniel's preparedness approach — buying the UPS between rounds of conflict, testing it, sourcing genuine batteries — is actually the right way to do it. As opposed to scrambling during an active attack.
That's exactly his point about preparedness. By the time the sirens are going off, it's far too late to be ordering batteries from AliExpress. You need to have this stuff sorted out in advance, tested, and ready to deploy.
I want to circle back to something Daniel mentioned about the Home Front Command's standing order. The ten-minute rule — wait ten minutes after the all-clear before leaving the shelter, because of unexploded ordnance and fragments. And with cluster munitions in play, that guidance got even stricter.
Cluster munitions change the risk profile significantly. You're not just dealing with a single warhead that either detonated or didn't. You're dealing with dozens or hundreds of submunitions scattered over a wide area, many of which may not have detonated on impact. They can be sensitive to disturbance — someone walking past, a door opening, vibrations. The Home Front Command's guidance to wait for an official all-clear rather than self-assessing is absolutely sound from a safety perspective. But it only works if the all-clear can actually reach people.
That's the systems failure we were talking about. The safety protocol assumes communications infrastructure that the shelter design negates. It's not that anyone was malicious or incompetent — it's that the left hand and the right hand weren't talking.
This is actually a well-known problem in emergency management called the "last mile" problem. You can have a perfect alert generation system — satellites detecting launches, trajectory models predicting impact zones, automated alert dissemination — but if the final delivery to the person in the shelter fails, none of it matters. And the last mile is always the hardest part. It's where the technology meets the physical world, with all its concrete walls and network congestion and dead batteries.
Let's broaden this out. Daniel's solution — a cellular router with an external antenna and a battery backup — is one approach. What other solutions exist for getting connectivity into a concrete shelter?
There are a few options. The simplest is a passive repeater — essentially two antennas connected by a cable, one outside the shelter and one inside. No electronics, no power required. It just captures signal outside and re-radiates it inside. The downside is that it's inefficient — you lose a lot of signal in the coupling — and it only works if there's decent signal outside to begin with.
It's broadband in the sense that it repeats everything, not just the frequency you care about. So you're also amplifying noise.
The next step up is an active repeater, also called a bi-directional amplifier or BDA. This has an outside antenna, an amplifier, and an inside antenna. It requires power, but it can provide meaningful signal gain — twenty to fifty decibels. The better ones are carrier-approved and include automatic gain control to avoid interfering with the tower. In Israel, you can get these from companies like Cel-Fi, and they're used in some public buildings. But they're not cheap, and they're not installed in most shelters.
Then there's the approach Daniel's taking — a dedicated cellular router that creates its own connection to the tower and then provides local Wi-Fi. That's more like a personal femtocell than a repeater.
The router is a full cellular device — it has its own SIM card, its own radio, its own identity on the network. It's not just amplifying the signal; it's creating a new endpoint. The advantage is that you can use a high-gain directional antenna pointed at the tower, which gives you much better signal quality than a phone's omnidirectional antenna. And then you share that connection over Wi-Fi, which works fine inside the shelter because it's short-range and the concrete walls don't matter as much when the access point is in the same room.
Wi-Fi can handle dozens of clients without breaking a sweat, so you can share the connection with everyone around you.
A modern Wi-Fi access point can easily handle fifty or a hundred connected devices. The bottleneck becomes the cellular backhaul, not the local distribution. And in an emergency scenario where everyone just needs to receive the occasional text message or check a news site, the bandwidth requirements are minimal.
There's an interesting irony here. In normal times, we're all pushing for higher bandwidth — streaming 4K video, video calls, cloud gaming. But in an emergency, the requirement collapses down to almost nothing. A few kilobytes of text. A simple "event over, safe to leave" message. And yet that minimal requirement can be harder to meet than streaming Netflix in peacetime, because the infrastructure is stressed and the environment is hostile.
That's a profound point. It's the difference between normal operation and degraded operation. Our networks are optimized for normal operation — high throughput, low latency, lots of concurrent users doing data-heavy things. But under stress — whether it's a missile attack, an earthquake, a hurricane — the network has to fall back to its most basic functions. And those basic functions, like SMS and Cell Broadcast, are often the most neglected parts of the system because they're not revenue-generating and not what customers complain about day to day.
Until the day they're the only thing standing between a population and a lethal mistake.
And that's the preparedness mindset in a nutshell. You're not preparing for Tuesday. You're preparing for the Tuesday when everything else has failed.
Let's dig into another aspect of Daniel's question — the specific hardware. He mentioned a cellular router with an SMA antenna connector that runs on mains power. What should someone look for if they're putting together a similar setup?
A few key specs. First, the router should have external antenna connectors — SMA or TS9 are the most common. Without those, you're stuck with the internal antenna, which won't do much good if the router itself is inside a concrete box. Second, you want a router that supports the cellular bands used by your carrier. In Israel, that means LTE bands three, seven, and twenty-eight primarily, with band twenty-eight at seven hundred megahertz being the best for penetration and range.
Look at the power input specs. Most of these routers run on twelve volts DC, even if they come with an AC adapter. The adapter is just converting mains AC to twelve volts DC. If you can bypass the adapter and power the router directly from a twelve-volt battery, you skip the inverter entirely and gain a lot of efficiency. Each conversion step — DC to AC, then AC back to DC — costs you ten to fifteen percent in losses. Skip the middleman.
Instead of a mini UPS that outputs two hundred thirty volts AC, you could use a simple twelve-volt battery pack with a DC barrel jack that plugs directly into the router.
That's often the smarter approach if your router supports it. And many do — the Teltonika RUT series, for example, can be powered by anything from nine to thirty volts DC. You just need the right connector and a stable voltage source. A three-cell lithium pack — eleven point one volts nominal — would run one of those routers for hours with no inverter losses.
That eliminates the whole inverter question. No switching, no sine wave generation, no transformer. Just batteries and a voltage regulator.
But Daniel's router apparently needs two hundred thirty volts AC, or at least that's what its power input expects. Some devices have internal power supplies that only accept AC. In that case, you do need the inverter. But it's worth checking — a lot of "AC only" devices actually have a bridge rectifier as the first stage of their internal power supply, which means they'll run on DC just fine if you feed them a high enough voltage. Though I wouldn't recommend experimenting with that unless you really know what you're doing.
The mini UPS approach is a universal solution — it works with anything that plugs into a wall socket — but it's less efficient than a DC-direct solution for devices that support it.
It's the difference between a general-purpose tool and a purpose-built one. Daniel's mini UPS can power his router, but it could also power a small lamp, charge a laptop, or run a medical device in a pinch. That flexibility is valuable in a preparedness context, even if you pay for it in efficiency.
Let's talk about the power station market, because Daniel mentioned those as a point of comparison. The Jackerys and EcoFlows and Bluettis of the world.
Those are essentially the same technology scaled up. A Jackery Explorer 500, for example, contains a lithium-ion battery pack — usually NMC chemistry, around five hundred eighteen watt-hours. It has a built-in pure sine wave inverter rated for five hundred watts continuous. The inverter design is fundamentally the same as Daniel's mini UPS — DC from the battery, switched at high frequency through a transformer, rectified to a DC bus, then inverted to AC. It's just bigger MOSFETs, bigger transformer, bigger heatsinks.
The price difference reflects that scaling. Daniel's AliExpress mini UPS board probably cost fifteen or twenty dollars without batteries. A Jackery 500 is four or five hundred dollars. But the Jackery can run a refrigerator for a few hours. Daniel's unit can run a router.
Right tool for the job. For a shelter connectivity kit, the mini UPS makes sense. It's small, it's cheap, it uses standard batteries you can replace. If it gets damaged or lost, you're not out five hundred dollars. And it's sized for the load — no point carrying a two-kilogram power station when a two-hundred-gram board will do.
I want to circle back to something you said earlier about modified sine wave versus pure sine wave. Daniel's router — how do we know which one it needs?
Most networking equipment uses switch-mode power supplies, which are pretty tolerant of modified sine wave. The input stage is a rectifier and a capacitor — it doesn't care about the waveform shape as long as the peak voltage is right and the RMS voltage is close. The capacitor smooths out the ripple anyway. So a modified sine wave inverter is probably fine for a router. But not guaranteed. I've seen some devices where the power supply's inrush current limiting doesn't work right with modified sine wave, or the EMI filter rings at the switching edges. It's rare, but it happens. If you want to be certain, get a pure sine wave unit. They're a bit more expensive but not dramatically so for low-power applications.
That's the kind of thing you'd want to test before you're actually in a shelter during a missile attack.
Plug the router into the UPS, pull the wall power, and verify that it stays online for the expected duration. Run a speed test. Check that the antenna connection works. Make sure the SIM card is active and has credit. Do all of this on a quiet Sunday afternoon, not when the sirens are wailing.
That's the preparedness ethos in a sentence. Do it now, not later.
Write down which cable goes where, which antenna connector is for which band, how long the battery lasts under real load. In an emergency, your cognitive capacity is diminished. You don't want to be reading tiny labels or trying to remember which SMA connector is the primary one. You want a checklist.
Daniel's probably already got a spreadsheet.
I would be disappointed if he didn't.
Alright, let's zoom out to the broader engineering principle here. The switch-mode power supply is one of those invisible technologies that shaped the modern world. Everything from phone chargers to electric vehicles to solar inverters depends on it. And yet most people have no idea how it works.
It really is an unsung hero. The basic topology — the buck converter, the boost converter, the flyback, the forward converter — these were all developed between the nineteen sixties and the nineteen eighties. But it wasn't until power MOSFETs became cheap and fast enough in the nineteen nineties that switch-mode supplies became ubiquitous. Now they're in everything. The global market for switch-mode power supplies is something like forty billion dollars a year.
The key enabling component is the MOSFET. A switch that can turn on and off millions of times per second, handle hundreds of volts and tens of amps, and do it with very little loss.
Modern power MOSFETs are incredible devices. A silicon carbide MOSFET can switch at hundreds of kilohertz with on-resistances measured in milliohms. They're what make compact high-power inverters possible. The same technology that's in Daniel's mini UPS, scaled up, is what drives the motors in a Tesla.
The physics is the same. The difference is just the size of the transistors and the capacity of the batteries.
That's the beauty of it. A boost converter that steps three point seven volts up to three hundred twenty volts works on exactly the same principle whether it's powering a five-watt router or a five-kilowatt solar inverter. The inductor might be bigger, the switching frequency might be lower, the cooling might be active instead of passive — but the core idea is identical.
The core idea, for anyone who's been following along, is: store energy in a magnetic field, release it at a different voltage. That's what a boost converter does. The inductor gets charged up when the switch is on, and when the switch turns off, the magnetic field collapses and dumps its energy into the output capacitor at whatever voltage the load demands. The ratio of on-time to off-time determines the voltage multiplication.
In a flyback converter — which is probably what's in Daniel's mini UPS — the transformer itself is the energy storage element. The primary winding gets charged during the on-time, and the energy transfers to the secondary during the off-time. It's called flyback because the secondary current "flies back" during the off cycle. It's an elegant design because the transformer provides both energy storage and voltage step-up in a single magnetic component.
When Daniel looks at that little circuit board with four batteries attached, he's looking at a flyback converter driving a small transformer at maybe fifty kilohertz, followed by a rectifier and an H-bridge inverter generating a passable sine wave at fifty hertz. All from four commodity lithium cells.
The whole thing probably cost less to manufacture than the shipping fee to get it from Shenzhen to Jerusalem.
Which is, in its own way, as remarkable as the engineering. The global electronics supply chain has made sophisticated power conversion available at prices that would have been unimaginable twenty years ago.
There's a downside to that, of course. The quality control on these ultra-cheap AliExpress boards is inconsistent at best. I've seen units where the solder joints were cold, the transformer wasn't properly varnished, the capacitors were counterfeit, and the safety clearances between high-voltage and low-voltage sides were inadequate. If you're buying one of these for emergency preparedness, you should open it up, inspect it, and test it thoroughly before relying on it.
What should someone look for when they open it up?
First, check the isolation between the high-voltage and low-voltage sides of the board. There should be a visible gap — at least six to eight millimeters — between the primary and secondary traces. Any less than that and you risk arcing, especially in humid conditions. Second, look at the capacitors — they should have recognizable brand names and voltage ratings comfortably above the operating voltage. If the caps are no-name and the voltage rating is marginal, they'll fail early. Third, check the transformer — the windings should be neat and the core should be properly secured, not rattling around.
From a safety perspective?
The enclosure should be non-conductive and properly vented. Lithium cells generate heat during discharge, and the inverter generates heat. If the whole thing is sealed up with no airflow, you're going to cook the batteries, which degrades them and creates a fire risk. Also, there should be a fuse on the battery input. If there isn't one, add one. A five-amp automotive blade fuse in an inline holder costs pennies and could save you from a short-circuit fire.
This is the kind of practical detail that doesn't make it into the AliExpress product description.
The product descriptions are often machine-translated gibberish anyway. "High quality power bank inverter pure sine wave camping emergency solar generator portable UPS." All the keywords, none of the specifications.
Alright, so we've covered the physics of voltage step-up, the engineering of inverters, the practical considerations for shelter connectivity, and the quality control caveats for cheap electronics. Is there anything we're missing from Daniel's question?
He asked about other solutions for creating AC-level voltage at low amperage. One alternative worth mentioning is a DC-DC boost converter followed by a small inverter module. You can get these as separate boards — a boost converter that takes your battery voltage up to, say, forty-eight volts DC, and then a small inverter module that converts forty-eight volts DC to two hundred thirty volts AC. This modular approach lets you mix and match components for your specific needs.
If you don't need AC at all, you can skip the inverter entirely and just use the DC-DC converter to power your DC devices directly.
Which is the most efficient approach. Every conversion step costs you energy. Battery to DC-DC converter to device — one conversion. Battery to inverter to AC to device's internal power supply back to DC — three conversions. You're losing ten to fifteen percent at each step.
The ideal preparedness power setup is: know your devices' actual DC input requirements, and supply them directly from a battery bank with appropriate voltage regulation.
That's the engineer's answer. The practical answer is: have a small inverter-based UPS for devices that need AC, and direct DC connections for devices that support them. Redundancy in both power sources and conversion methods.
Speaking of redundancy — Daniel mentioned that even the SMS-based alert system failed during the war. What's the backup to the backup?
For emergency alerts, the ultimate fallback is broadcast radio. FM and AM radio transmitters are physically separate from cellular infrastructure. They operate at much lower frequencies — FM around a hundred megahertz, AM around one megahertz — which penetrate buildings far better than cellular frequencies. A battery-powered radio can receive alerts even when every cell tower in the area is down or congested. Israel's Home Front Command distributes emergency radios to some communities, but they're not as ubiquitous as they should be.
That's a technology that's been around for a century. Sometimes the old solutions are the most robust.
There's a reason ships still carry marine VHF radios and aircraft still use analog AM for air traffic control. Simplicity is reliability. A crystal radio receiver doesn't even need a battery — it runs on the power of the received signal itself. You can build one from a diode, a coil of wire, and an earpiece. It won't stream Netflix, but it'll tell you when it's safe to leave the shelter.
I think that's the note Daniel was striking with his frustration about the "high-tech nation" not figuring this out. Sometimes high-tech is more fragile than low-tech. The sophistication that gives us gigabit speeds and ultra-low latency also gives us more points of failure.
The failure modes are often non-obvious until you're in the middle of them. Nobody tested what happens when a hundred people with smartphones crowd into a concrete box during a ballistic missile attack, because that's not a scenario you can easily simulate. But now we know. The question is whether the lessons get incorporated into the next round of preparedness planning.
Daniel's doing his part — he's building his own solution and sharing the engineering questions it raises. That's more
