9 Voltage Regulators You Should NEVER Use in Your Product
These nine voltage regulators show up all the time in beginner designs and early prototypes.
They work fine on the bench, but in a real product they cause thermal failures, blown EMC tests, and production problems that cost thousands of dollars to fix.
Every one of them has a better, modern replacement that costs about the same, and I’ll show you exactly what to use instead.
For each regulator on this list, I’ll cover why it’s so popular, what actually goes wrong when you put it in a real product, and what to use instead.
Regulator #1 – The 7800 Series
The 7800 series is the first voltage regulator family every engineer learns about, and it shows up in every textbook, every university lab, and every beginner tutorial online.
It feels like a safe, proven choice because it’s been around for over 50 years.
But the 7800 series has a dropout voltage of around 2V, which means the input must always be at least 2V higher than the output.
That matters because with any linear regulator, all the power across that voltage difference gets wasted as heat.
So the closer you can run the input to the output, the less power you burn, and a high dropout regulator forces that gap to be bigger than it needs to be.
Say your input is only about a volt above your desired output, like running a 5V rail from a 6V battery pack.
A modern LDO with much lower dropout handles that easily and wastes very little power as heat because the voltage gap is so small.
But the 7800 series can’t even regulate in that situation because it needs a full 2V of headroom, which means you’d need at least 7V on the input just to get a stable 5V out.
That forces a much bigger voltage gap, which means a lot more power wasted as heat for no good reason.
And on top of that, the 7800 series comes in large, outdated packages that eat up board space inside your product.
For any big voltage step-down, use a buck converter instead, which steps voltage down much more efficiently than any linear regulator.
For small step-downs where a linear regulator makes sense, use a modern LDO with genuine low dropout.
The 7800 series should stay in the classroom where it belongs.
Regulator #2 – LM2596
The LM2596 is everywhere because of those one-dollar buck converter modules you can buy on Amazon.
Plug it in, get your voltage, and move on, which makes it great for quick prototyping.
But the LM2596 switches at just 150 kHz, which means it needs large inductors, large capacitors, and it throws off a ton of electromagnetic interference.
The footprint is also huge compared to modern alternatives, which eats up board space inside your product.
The module layouts were never designed for EMC compliance, so you will fail FCC and CE testing.
On top of that, counterfeit LM2596 chips are all over the market, meaning you can’t even trust the datasheet specs.
Use a modern synchronous buck converter instead, one that switches at 500 kHz or higher, uses smaller passives, and generates far less EMI.
And always design your own layout using the IC vendor’s reference design rather than dropping in a random module.
Regulator #3 – MC34063
The MC34063 can technically do buck, boost, and inverting topologies all in one chip, which makes it look like the ultimate flexible power solution on paper.
In practice, it’s a headache.
The switching waveforms are uncontrolled, so EMI is terrible.
Efficiency tops out around 80% under ideal conditions and drops fast in any real design.
The external component count is high, compensation is tricky to get right, and most designers end up with an unstable or noisy output without knowing why.
Debugging a bad MC34063 design can eat up weeks of engineering time, and I’ve seen that happen more than once.
Pick a dedicated buck or boost IC instead, because a purpose-built converter will outperform the MC34063 in every single way with a simpler, cleaner design.
Regulator #4 – XL6009, XL4015, and Other No-Name Switcher ICs
No-name switcher ICs like the XL6009 and XL4015 show up on ultra-cheap modules from AliExpress, and they seem to work fine during testing.
But the datasheets for these parts are often incomplete, inconsistent, or outright fabricated.
There’s no way to verify the electrical specs on your own, there’s no factory application support, and if you hit a design issue there’s nobody to call.
Long-term sourcing is also unreliable since these parts appear and disappear from the market without any warning.
So if you can’t buy it from a major distributor like Mouser or DigiKey, it doesn’t belong in your product.
Stick with ICs from TI, Analog Devices, ST Microelectronics, and other well known chip makers.
You’ll get verified datasheets, reference designs, application engineers you can actually talk to, and reliable distribution.
The per-unit cost difference is often under fifty cents.
Regulator #5 – AMS1117
The AMS1117 is on nearly every Arduino, ESP32, and STM32 dev board out there, so designers assume if it works on the dev board it’ll work just fine in their product.
That’s a dangerous assumption.
Despite being marketed as “low dropout,” the actual dropout voltage is over 1V at rated current, which isn’t low at all.
Counterfeit AMS1117 chips are everywhere, and the fakes perform even worse than the originals.
Transient response is poor too, so current spikes from WiFi transmissions or motor startups cause voltage sags that can reset your microcontroller.
Use a modern LDO instead, one with genuine low dropout under 250 mV and decent transient response.
And if you need high power supply rejection with low noise, take a look at the TPS793, which was actually one of the first chips I designed at TI, or the TPS799 if you need that same performance with lower quiescent current.
Regulator #6 – Cheap, Unshielded Switching Regulator Modules
There’s nothing wrong with using a well-designed switching regulator module in a product, and I actually encourage it in a lot of cases because it saves you from having to do your own switcher layout.
The problem is when people grab the cheapest module they can find on Amazon or AliExpress and drop it straight into a production design.
Those cheap modules are almost never shielded, which means they radiate EMI everywhere and you’ll have a very hard time passing FCC or CE testing.
The PCB layout on those modules is usually terrible too, which kills the switching IC’s performance and makes thermal management unpredictable.
And quality varies from batch to batch, so the power supply that worked in your first 100 units might behave differently in the next 500.
If you’re going to use a module in production, look for one from a reputable supplier that’s properly shielded, uses a quality layout, and has consistent sourcing.
That kind of module can actually be a smart choice, especially if you don’t have switching regulator layout experience on your team.
Regulator #7 – HT7333 and HT7533
The HT7333 and HT7533 cost just a few pennies per unit, which makes them look like a no-brainer for simple low-power circuits.
But the maximum rated current is only 250 mA on the HT7333 and just 100 mA on the HT7533, and both ratings drop further at elevated temperatures.
There’s also no thermal shutdown protection, so either part can overheat and damage itself without giving you any warning.
Load regulation falls apart above 100 mA, and power supply rejection drops off above 1 kHz, which means switching noise from other regulators on your board passes right through to your sensitive circuits.
That kind of noise coupling can cause all sorts of strange behavior in your microcontroller or sensors, and it’s extremely hard to track down if you don’t know the regulator is the source.
For just a few cents more per unit, you can get an LDO with thermal protection, better load regulation, and specs you can actually rely on across temperature.
Regulator #8 – LM317
The LM317 is the most common adjustable voltage regulator, and it feels like the ultimate flexible part because one chip can give you any voltage you need.
But the problems here aren’t unique to the LM317, they apply to any adjustable regulator.
That external resistor divider adds tolerance stackup, where small errors in each resistor compound together, plus thermal drift and two extra failure points to your design.
If the bottom resistor in that divider opens, the output jumps to the full unregulated input voltage and fries everything downstream.
If the top resistor opens instead, the output drops to about 1.25V, which will likely cause your circuit to malfunction in a completely different way.
The LM317 specifically also has no enable pin and a dropout over 2V.
And unlike most regulators, it has no ground pin, so it’s a floating topology where the chip’s own supply current gets dumped straight into the output.
That supply current is fairly high, which is a big part of why the LM317 requires a minimum load current of 3.5 to 10 mA just to keep the output in regulation.
If your circuit draws less than that, the output voltage rises uncontrollably, which is a nasty surprise in low-power designs.
Here’s the thing though, almost everyone using an adjustable regulator is setting it to a fixed voltage anyway.
So just use a fixed-output LDO matched to your actual voltage rail, and you’ll end up with fewer external components, better specs, and lower risk.
Regulator #9 – MCP1700
The MCP1700 has a quiescent current of just 1.6 uA, and for battery-powered products that number is almost irresistible.
But the maximum output current is only 250 mA, and the transient response is very slow.
When your microcontroller wakes from deep sleep and draws a burst of current, the MCP1700 can’t respond fast enough and the output voltage dips hard.
If your microcontroller needs 150 mA or more in short bursts, you’ll see brownout resets that are incredibly frustrating to debug because everything looks perfectly fine under steady-state testing.
People pick this part based entirely on the quiescent current spec without ever checking whether it can actually handle their real load profile.
So match your LDO to your full load requirements, not just a single headline number on the datasheet.
Look for an LDO that combines low quiescent current with fast transient response and enough current capacity for your worst-case load bursts.
A single spec on a datasheet doesn’t make a regulator right for your product, you have to look at the full picture.
