I bought super cheap DC-DC converter on Amazon, but It was FAKE.

[@demef758]

One of the problems with ALL of these modules, real or fake, is that the catch diode is only rated for 1A while the spec says that the maximum output current is 3A. The catch diode rating must be rated for the same current as the output current. Pull 3A out of these modules and you will notice the catch diode gets real hot! Replace the diode with a 3A/40V Schottky. A second issue is the use of aluminum capacitors in a switcher application. If these caps are switcher grade, they have much lower ESR (equivalent series resistance) than typical cheap aluminum capacitors. The former are rated for switching applications, the latter are rated for 120 Hz applications. Switchers create a lot of high frequency triangular current waveforms that must pass through these capacitors. If the ESR is high, then the capacitor will generate a lot of I^2*R power loss and heat up the capacitor, limiting its useful life. I seriously doubt that a cheap $1 module uses switcher-rated aluminum capacitors! The ESR of the output capacitor has a significant effect on the ripple output voltage. Again, a lower ESR filter capacitor will lower the output ripple. Better yet, tack on a decent 1uF/35V ceramic capacitor (has almost zero ESR) across both the input and output capacitors and you will see both the input and output ripple voltages drop significantly, and the switching waveforms will be much more stable! EDIT: the uber-slow switching speed is also the result of these cheap aluminum input and output capacitors. Parallel ceramic capacitors (with voltages rated for the input/output voltages) and you should then see much smaller ripple voltages as well as the switching frequency increasing. However, one thing I have seen in some of these cheap buck converters is the typical Chinese modus operandi, where they wipe clean the controller chip's markings and laser etch a different part number on it. There is another National switcher that runs at 52 kHz, I think it is, and that's what you will get on the board! It SHOULD be an LM2596, but it's actually it's this much slower chip. Hey ... what do you expect for $1?

[@ellensedge1898]

Do you think you could wire this to a usb socket and use it to charge a phone at 5.5 volts or whatever? Or would the phone try to draw like 5 amps or something like that?

[@ahmedelwan9129]

@@ellensedge1898 the phone will never draw 5.5 amp from 5 volt source

[@BromideBride]

SS34 diodes on my boards - unless someone is bothering to fake these as well. Thanks for the esr tip. These boards have through holes under the caps so perfect to tack in ceramics.

[@MattFixesStuff]

Which diode do you mean ? The diode on mine says ss36 which supports up to 3A. But it still gets super hot when drawing 3A. Maybe upgrading to a 5A like SS54 would make sense ?

[@Jimmeh_B]

@@ahmedelwan9129 If it's USBC it might. Depends on the battery capacity. 3C charge rates are not uncommon for LiOn cells.

[@jamesbrown99991]

LM2575 has 52kHz switching frequency. It's possible you have some rebadged LM2575, or rebadged fake LM2575.

[@spehropefhany]

Also LM2576. I would prefer to see a chip with the manufacturer's actual name on it, like XL. If the input voltage exceeds 15V, the inductor should be more like 100uH or more for the fixed 52kHz frequency than the 47H they used. These are probably okay for 12V in and 3-6V out.

[@n.shiina8798]

imagine faking an already faked IC... lmao

[@demef758]

Correct! They have rebranded the LM2575 as a LM2596! They probably had a massive inventory of the 2575 lying around and scratched off the labels. The backside of the ICs will reveal the difference in the shapes of the exposed metal tabs, and the different package type!

[@psykauze]

Yes, after looking at the LM2576 datasheet, it seems both have the same pinout and the same application for adjustable voltage configuration. The main difference are the internal oscillator (52kHz for LM2576) and the recommended inductor and capacitor to use (about 4 times more capacity and inductivity for LM2576). Replacing the output components will help a lot the efficiency at low output voltage.

[@user-eq2fp6jw4g]

@@spehropefhany Yep. That's what you wanna use them nothing more. Thou had one as 12v to 5v step-down on 24/7 usage for my raspberry pi. It died pretty quickly. Heat dissipation isn't that great on those. Better get some beefier models with proper heatsinks and use them if you need em 24/7

[@JoseGustavoAbreuMurta]

To identify the frequency of the internal oscillator you can scope pin 2 of the LM2596. Thank you for sharing this information. I tested 6 modules LM2596 . All are fake with 50 KHz from China!

[@ALEFILES]

Thanks a lot for the information! I will test my dc dc converters with my oscilloscope! Greetings from Argentina!

[@j.f.christ8421]

Technically they're not fake as they us a *LM2596S* chip, I guess the "S" stands for "Slow".

[@jahrkh3518]

@@j.f.christ8421 According to the datasheet from Ti, "The LM2596 is available in two packages: a 5-pin TO-220 (T) and a 5-pin surface mount TO-263 (S)." [page 33].

[@j.f.christ8421]

@@jahrkh3518 That's a funny looking TO-263 on that board.

[@jimmatrix7244]

@@j.f.christ8421 S for Stupid

[@foogod4237]

I wonder if the reason it gets so hot is because they used a circuit design intended for 150 kHz but the fake chip is only running it at 50 kHz. That would make the choice of inductor and other components not well tuned to the frequency they're being driven at, which could result in increased inefficiency and heat generation than the circuit is supposed to produce.

[@jnharton]

FWIW if they're using cheap components they may have a larger range of variance than the circuit design calls for. E.g. a 'standard' carbon film resistor could vary by as much as 20% from the marked value. You can usually purchase parts that have been verified to be with +/-10% or +/-5%. Presumably other components, like capacitors or inductors, could have similar differences.

[@alexlo7708]

Conversely , low switching tends to dissipate less heat than high switching because it has less switching loss.

[@d614gakadoug9]

Running at a third of the intended frequency might result in inductor saturation which can cause higher dissipation in the switch, depending on how fast it limits peak current. The surface mount package has a thermal resistance of about 50 °C/W if the copper area it is mounted on is just a little bigger than the package. On a crowded board there isn't much space on the top for more copper area. Lots of vias to conduct heat to foil on the back helps. I didn't notice if that is used on the boards in the video. From what I"ve seen, there isn't likely any converter board of any sort sold to the hobby market at ebay or the like that could actually deliver the claimed power for more than a brief period without forced-air cooling.

[@jdlech]

Generally speaking, the higher the switching frequency, the less efficient the circuit is, and the more heat it will produce. So if anything, 50KHz switching frequency will produce less heat than 150KHz. The tradeoff is the lower frequency ripple voltage at the output, which is harder to filter out. This applies when you compare apples to apples - the same circuit with different ICs. There are other, much more elaborate circuits that will be more or less efficient at a given switching frequency.

[@foogod4237]

@@jdlech Lower frequencies can also require larger inductors to avoid reaching saturation. If the inductor is being driven close to (or over) saturation due to the lower frequency, that would reduce the efficiency and result in excess heat generation too.

[@Bapate-rh9be]

I bought a whole bunch of these exact buck converters 1 year ago and I use them for a lot of applications. Most of these applications do not require anything spectecular (i.e. power a bunch of 12V fans after upgrading a 3D printer to 24V). They perform quite well in these applications. I remember measuring the ripple on the output and I do not recall anything quite that high. I suppose they might have changed the IC in the meantime, as back then they were also quite a bit more expensive.

[@SianaGearz]

I don't think much changed in between . Back when you bought them, it was semiconductor shortage and shipping shortage, thus more price. Do observe that you need to reach a very high current to reach that high ripple, you won't measure it with benign loads. Of course genuine circuit can stay low ripple up to the spec limit

[@p4spooky]

Excellent video. Will be very useful if you can show what happens when you install the original IC to max current/ripple and switching frequency and temperature measurements

[@monteceitomoocher]

Yes my thoughts exactly, replace the ic with a genuine one and recheck the performance.

[@spehropefhany]

@@monteceitomoocher Need to replace the inductor too. It's too low.

[@n.shiina8798]

@@spehropefhany the inductor value is good for genuine LM2596. this module usually uses re-marked LM2576 which runs at around 50kHz but the supporting components were taken off genuine 2596 datasheet

[@jayrose6312]

I would like to see this too. I could be wrong, but I would think that the operating efficiency will increase by approximately three-fold. The temperature will go down significantly as well as the ripple. I personally would still consider a 0.22uF ceramic cap to prevent HF noise if still necessary, but there should no longer be a need for a large capacitor since the ripple would be under control. I would need to watch/listen again as I don’t remember the exact usage of the diodes in this circuit because it may or may not have protection from inductive loads as that could still damage the IC. I guess if you’re not planning on using this circuit for motors or the like, it wouldn’t be necessary to include. Just a thought. The only thing that still wouldn’t hurt is a small passive aluminum heatsink to dissipate any remaining heat as it simply wouldn’t hurt to account for it even if the temperatures are not too high secondary to using a legit IC within the correct specs. Any thoughts? 🤔💭

[@kayakMike1000]

​@@spehropefhanyha! I was thinking the same

[@lohikarhu734]

this kind of very low frequency, with big ripple, can sometimes be traced back to your test setup. if you are running a fairly high load current and have insufficient input capacitance, or too much resistance, or poor regulation, in your bench power supply, the device will "motorboat", not actually reaching a stable operating point, it keeps "re-starting", as its soft start can't get it running, and it just keeps re-trying to start, and 15-50 kHz is pretty common.

[@copernicofelinis]

From my experience, markings on ICs and transistors bought from Chinese resellers on the marketplace are just... decorations. They have nothing to do with what's inside the case. Also, you need to convert from Chinese units to SI units by diving by 2. If it says it will withstand 1 Chinese ampere, it will only withstand one half SI ampere. They can still be useful, tho.

[@catskgsw6218]

/Technically:/ There is often a fatal misinterpretation about the single data items in a datasheet, benevolently proclaimed by a manufacturer's announcement. Then it is often hard to distinguish between "Absolute Maximum" and "Recommended Ratings" and to keep in mind, that even then not every and all "recommended" can be stretched at once to all edges - for a simple example: a maximum current and a maximum voltage, but limited by the maximum power dissipation, which is all limited in addition due to cooling conditions, and that these values are variable with (are a function of) the working temperature. So the given values can be right - but not in the way _we_ think they are connected and to be used. Planning for 30..50% of the given value will work. /DC-DC-Converters:/ For example *step-up* converters are often given the maximum ampere the switch mosfet can handle, eg. 3A - but If we convert a single battery cell to a useful higher voltage, e.g. by factor 2x - due to the power transfer (V_in × I_in ≈ V_out × I_out) the output will only provide 1.5A! What do you think, which value will the seller propagate? Volts by ampere, how to calculate watts, cooling? - ahhh, much tooooo complicated, do not confuse the buyer to buy!! /Chinese Markets:/ So I often see ads, where the seller proclaims anything, remember "aim high", "more is better" (ignoring knowingly or unaware or uncaring?) - there also seems to be some "bullshit bingo" phrases too (e.g. every thing is "miniature" and "LED" even with incandescent miniature lamps) - once a seller answered to me, why I am moaning, what he does would be standard, cause that's what all others do 😕

[@mikkelbreiler8916]

Yup. Chinese stuff can be useful. But I buy fireworks for fireworks, not other stuff.

[@PeterM-PeterM]

I have used this module before in a small, non critical application with no issue. I was about to start using them to power some sensitive electronics... I can't thank you enough for the demonstration. I will be looking for a different device for my sensitive equipment.

[@uwepolifka4583]

Very informative video. I bought some of these modules a few years ago. At only 1,1 A and from 38V down to 30V the coil becomes so hot that it melts.

[@AccessAccess]

Read the whole datasheet. Many of these converters are designed to skip cycles or "beats" and will do so when operating at lower current draw. National refers to this as "discontinuous mode", other manufacturers may use "burst mode" or some other terminology. If the inductor value is not properly chosen for the operating parameters of the supply (voltages, current draw, etc.), the circuit may only be capable of discontinuous mode rather than being able to transition between the two modes as necessary. So while the IC may be counterfeit, it might be performing like the real thing, only that the inductor on the module is too low resulting in the converter skipping cycles and the ripple being quite high. Or perhaps the ripple may be elevated due to their being no load, or insufficient loading (very little current draw) on the circuit, and it will drop to better levels once an actual load is applied.

[@EJEuth]

This IC is in fact the LM2576 that runs at 52KHz, per TI’s datasheet. So, it is “re-baptized ” as LM2596, but the IC doesn’t know it should now oscillate at the remarked frequency of 150KHz.

[@rc-fannl7364]

Probably even a copy of the LM2576 then, considering the inconsistencies in the chips' appearance...

[@foogod4237]

@@rc-fannl7364 Or they're old LM2576es which have been desoldered from old boards and then resurfaced and reprinted to be labelled as LM2596 instead, because those will sell for more money. (It's even possible that the board manufacturer doesn't even know they're counterfeit, they just bought them from someone else who gave them a _really_ good deal on them, etc.) This sort of practice happens a surprising amount (it's actually a big issue for people sourcing harder to find components (such as for old retro computers), because people will sometimes salvage old ICs and then relabel them as something completely different (that they know will fetch a higher price), and the buyer shells out the money and then just gets some weird random unknown IC that doesn't work at all as what it claims to be...) It would be worth rubbing some acetone on those ICs and see if the markings rub off. Genuine chip markings are engraved and won't wipe off, but lots of the reprinted ones are just done using ink and will wipe away with acetone or alcohol.

[@EJEuth]

@@rc-fannl7364 surely a clone, but at least now we would know what to expect - approximately, at least.

[@Noname-hc4hc]

This chip is a remarked lm2576. The spec matches quite well.

[@stevensteltenpool904]

Sure seems like it.

[@MirunaIordachescu]

Yeap, is the old 2576 :)) , I use it 15 years ago , nice work tho to rename it just for the marketing reasons.

[@nielsdaemen]

At least it is a genuine LM2576

[@SianaGearz]

​@@nielsdaemen who knows, thermal pad differences suggest not all are genuine. Also it's unhelpful to be paired with wrong support components.

[@jxmai7687]

For that price I feel something is not right, just bought it for play around, so far haven't use anyone at all.

[@1djbecker]

I'm surprised that this wasn't about the knock-off chips being marked as the HV variant. Rather than working up to 60V, these fail at 30v-40v with an internal short that applies full input voltage to the output. I'll use them, but not anywhere close to their claimed input voltage. I'll have to check the still-working modules I have to see what the switching frequency is.

[@ketas]

i hate dc-dc's with that failure mode. i had one doing 20v to 5v for a phone. suddenly phone lost power. soon after that, output cap popped. luckily phone, note 4, survived it. because on measuring, it showed 20v on output. also output diode desoldered itself. maybe diode went short first. then the lm* (iirc). or did the heat do it. unsure, invest external crowbar? some dc-dc's include i/o tvs'es and ptc fuses in input and tell it protects devices. no idea, maybe. i'm talking about 2/4 port usb/qc3.0/etc dc-dc from china. you find them if pcb shows tvses on them. some are without. kind of sad, but i guess you get what you pay for. dc-dc can fail in that way but fakes make it even worse

[@1djbecker]

@@ketas The Note 4 supports 9V charging, so it might have an input switch that is able to adapt to (or block) somewhat higher voltages.

[@ketas]

i bet samsung included port protection or so. they do have something there. but likely 20v is stretching qc 9v input protection. i mean it does make sense as people connect them to bullshit power sources. i thought maybe half way blowing cap took voltage down. or indeed it had some circuitry. i bet it also does reverse polarity. i mean i once tried it in nokia 5800, it took 5v 2a until i suddenly realized it's reversed. survived. i'd bet nokia of course having something there. i could look if samsung does it. it makes sense. but that doesn't protect from every broken dc dc. i actually have other circuits i think how to crowbar protect them

[@martinwhitaker5096]

I have a batch of the HV modules. The IC is certainly fake - the HV version has been discontinued for many years and the switching frequency doesn't match the data sheet. That being said - they do withstand 50v with no issues.

[@beakytwitch7905]

I use these without problem, well within the supposedmaximum specifications. To solve the output ripple I use both ferrites for high frequency switching noise, and resistive regulators for the 50 kHz ripple. The result is reliable, clean, stabilized volt factories.

[@brendajanes4139]

Hmmm, I bought lot of 10 off of Amazon a few months ago. I tested them with a load tester at 1A and barely even got warm. I will have to dig them out and try 1.5A or 3A and test them again but this time, use my old trusty Tektronix 561A with a 3A1 plugin and see what the results are! Yes, my Tektronix 561A is mostly tube type, but still works great!

[@ABehrooz]

You're using the wrong datasheet. The input limit for DIP LM2596 is 37V +/- 4%. You also killed the output cap. which contributes to the voltage deviation. the frequency is not guaranteed since you have no consumer and the output cap is broken. there are output ESR constraints that are probably not met with the dead cap. The HVS version with 63v input cap can accept up to 58 volts. and no more than 28 volts is recommended for the S version. The packaging and font is not necessarily fake because there are multiple package factories. source: I've used dozens of these. all versions.

[@maudiojunky]

It's unlikely an electrolytic cap would die from a short overvoltage of that degree. The deviation, along with the abnormally high ripple, are more likely related to not loading the output with a nominal current during testing - you can see clearly the output is open circuit during the voltage tests, and SMPS require a minimum load to operate nominally. It's also very likely the ripple is high because the caps used on these cheap boards don't have an appropriately low ESR.

[@Orcinus24x5]

Take a drink every time he says "keelo hurts." 😂

[@tookitogo]

Or if you drink every time they say singular “volt”, “amp”, etc, where it should be plural! (I totally recognize that this is narrated by a person (or text to speech software) who likely is not an electronics person at all, reading a script likely translated by someone who isn’t an expert, either, and/or isn’t a native English speaker.)

[@mondotv4216]

@@tookitogoNo it's AI voice translation.

[@tookitogo]

@@mondotv4216 I said “or text-to-speech software”. The thing people are now suddenly calling “AI voice” is simply text-to-speech software, which has existed for around 50 years. With that said, text-to-speech software has been _really_ good for years now, to the point that it typically won’t make the mistakes heard here, which is why I actually suspect it’s a real narrator.

[@nimoy007]

​​​@@tookitogo It's definitely not a real person. The intonation of the "Goodbye" message is way off. There is also a difference with the way speech is generated, depending on how the model works, including if a LLM was used. Most people mean that an LLM was involved when they say "AI." Considering how natural the narration sounds, it's not that unfair to call it "AI-generated."

[@janami-dharmam]

I don't understand the joke; you should explain in the next line.

[@jdlech]

I also bought a lot of 10 of a similar, if not the same module for about $12 on Amazon. I also found the high ripple voltage - as high as 10% ripple. But you can compensate with a larger capacitor, or a small LC filter circuit. I blew a couple of them up while experimenting with them. I also pulled a continuous 3 amps. While the module got too hot to touch, nothing failed. Adding an aluminum alligator clip to the diode helps with heat dissipation. I used 3 of these modules to install a USB charger into the dash of my car, and one as the power supply for a high gain / low current output RF amplifier. The latter required a robust LC filter circuit to cut the ripple voltage to less than 1% ripple. But the power requirements were low enough that I could put it into a small container with no regard for heat dissipation. That was about 2 1/2 years ago, and all four modules are still working. So there does not seem to be any durability issues. Of course, now that I wrote that, Murphy's law is probably going to bite me.

[@electrotsmishar]

Just want to point out, a better way to measure ripple is to exclude that ground clip and use a spring probe instead in order to avoid picking noise. I have these cheap modules and they barely measure 30mV P-P ripple. Regarding switching frequency, the datasheet mentions that the frequency is lowered if there is high current. It's better to measure ripple while some current is flowing, not open circuit

[@tttuberc]

Great video. I've always wondered why these modules are so cheap compared to cost of parts I buy from Digikey

[@noth606]

that has a simple explanation, look at the price per unit for 1000+ units at digikey then take 30-50% of that. Now you're looking at something in the ballpark of what a manufacturing company at small scale pays. Used to have an electronics prototyping etc company, and had both commercial and semipersonal ties with digikey as well as some specific component makers. The handling cost for an order of 0.1$ of components is over 2$ min, I don't remember the numbers but you're not paying much for the component itself at low numbers.

[@paulsmith9574]

I had bought a module similar to that off eBay that was rated for 60V. On my bench I applied 41 volts, as that is what my project source voltage was. The IC on the board smoked. My load was only like 25 mA. I thought that perhaps I did something wrong. I worked in electronics for more than 40 years. I have designed many products for major companies. I have another application where I wanted to replace a LM7805 that was getting hot. The circuit had 2 LM7805's in series, the first had its GND pin raised to 3 V to pre-regulate the 12V to about 8 volts. Thus the wattage (Heat) spread between the two regulators (Both on heat sinks). The 8 volts is used for another circuit in the thing, so I left that there, and took the 12 volts direct to a unit like the one in this video. There are already 1000uF caps on the load side that I replaced with decent ESR and ripple current ratings. The project seems to be functioning much cooler than with the 7805. I just hope the IC there will survive and not put 12 volts into my 5 volt PCB. These were so cheap that I bought 6 of them for like $8, I don't recall exactly. If there are real TI parts that are re-labeled, I am OK with it. Digikey for just one LM2596 with shipping is about $10. But I can't design and order blank PCBs and all those parts as cheap as this deal. I have heard stories of fake capacitors and even empty memory devices. That is, chip carriers that have no chip inside, all coming from China. The thing is, when we need parts for something, we take chances, who knows what is inside. Many components are made in China. Even parts that Digikey, Mouser, and other places sell, can be from China.

[@transientaardvark6231]

Actually, an up-stream resistor will dissipate the large drop better - and it automatically does the most work when you need it to. Also, if you do the math, reducing the ripple actually increases the power dissipation bith in the regulator and also in the upstream supply. In the regulator it is always dropping ALL the volts, whereas with lots of ripple, anout half the time it has less voltage to drop. In fact at full load the ripple should just bottom out at the minimum head the regulator needs. Also, if the ripple is low, it means a transformer feeding the capacitors through a bridge rectifier is delivering all the charge-per-cycle in a smaller and therefore bigger pulse of current, Leading to more I2R losses, probably worse magnetics and more radiated noise. The more the ripple, the earlier the rectifier turns on and the longer it conducts for. I was gobsmacked when I realised this, because I was always taught "use big capacitors and then you'll get less ripple, which is good". And it is, but only if you don't have a regulator afterwards.

[@luxmonday]

As the switching frequency goes down, the inductor size (and saturation current) needs to go up. This is probably why the efficiency is very low. If the switching frequency was 150kHz, the inductor would be cooler and the efficiency would be higher. There is a sweet spot at about 400kHz for DC-DC switchers that allows relatively small inductors while using common MOSFETs and not dissipating too much heat in the MOSFET. Once you go up into MHz switching, the MOSFETs must be more specialized (low gate and transfer capacitance). Efficiency can drop again as you go up into the MHz.

[@wtmayhew]

Thanks for stating that clearly. The module shown in the video appears to be inexcusable junk. It seems some inexperienced designer decided to substitute the cheaper (knock off) lower performing regulator which switches at about 1/3 the rate of the part for which the circuit appears to be designed. I question whether the feedback is operational because the pulse width does not seem to be modulated by variations of input voltage or load current. It looks like you could get about the same “regulation” by using a 10 Watt passive 1000 Ohm rheostat, cost no withstanding.

[@kdt718]

I bought these modules from Amazon to use in the tail light of my infiniti qx60 2017 damaged due to sloppy installation, which resulted in water flooding the leds and shorted it. I used a couple of 2.2v leds and power from this module. It has been a couple of yrs, and it is still working.

[@konradgrima2872]

As usual great educational videos, keep up the good work.

[@ovalwingnut]

Always GR8T informative videos!!! You RoCk. Your "voice over guy deserves a big nod too. Best I've heard. Cheers from So.CA.USA, 3rd House on the Right.

[@modorangeorge4991]

Yes, that scope is a siglent 1204x-e for triple the price, with lecroy logo in it. No hobbyst buys it, but plenty of people but under siglent name .

[@sullivanzheng9586]

What you see as "switching frequency" is not the real switching frequency. Something is wrong with the feedback circuit, and the output is oscillating.

[@elpolaco7654]

I have one such module from 2014. The frequency of the waveform at the output of the IC is also 50 kHz. Like others, I also hope you will replace the IC with one from a reliable source and see how much difference there is. Although supposedly in the data sheet there is a lot of information on how the output parameters should look like depending on the capacitor and inductor used.

[@LimbaZero]

I think one of our local electronics shops sells those but have warning label to not use in finished products but as only in prototyping. Edit: Was it on some 60V modules you maybe can see old prints from caps that show 50v and 63V is overprinted.

[@ProtonovaR54]

Thank for this analysis, I had recently bought one of these from Amazon. You saved me from potentially killing the rest of the circuit I was going to hook this up to. The LM2596 was indeed a fake.

[@RonanKerdian]

This is great. I was using these modules to build a ups for a router, at while the draw was just under 3 amps, I've burned my finger. Since I bought a lot of them, I used one for charging and one for powering the ethernet equipment. Still very hot. Even under 1.5-2A, but it works!

[@MicheIIePucca]

A great eye-opener again on paying low prices and getting fake chips of questionable quality (and sometimes with dangerous outcomes).

[@kinetic1259]

I have heaps of these and always thought they got a bit too hot, even melting my 3D printed casing above 20W. Thanks for explaining why! They are incredibly inefficient, you can't use them for battery powered projects. どうもありがとうございます！

[@caletorino]

I find a lot of fakes on hobbiest websites. Purchasing from a reputable supplier helps but of course now they charge a pretty penny... So it's a balancing act. The rule of thumb I keep in mind with power supply's (of all kinds) never run it more than 50% of its component ratings and for the fakes always divide the claimed ratings by 1/2 😅

[@philippepaternotte2965]

lol: if you divide by 1/2, you multiply by 2!

[@Desert-edDave]

In the PC-building world that'd get very expensive and be a perfect example of wasted money due to ignorance. By your logic / 'rule of thumb' a 1KW psu would be insufficient for many desktop scenarios. 😆

[@marcogenovesi8570]

@@Desert-edDave in PC building world the fake 1kw PSUs cost less than the real/reputable 500W PSUs (and are usually 500-ish W PSUs) so his logic isn't wrong. With PC power supplies it's much easier to find reputable sources so you don't have to do rules of thumbs.

[@jnharton]

@@marcogenovesi8570 It's still generally a good idea/rule to not depend on perfect outcomes, though. Seriously, don't even use a 22uF capacitor or 1 kilo-ohm resistor /without testing it/ if you need it to be exactly that value. Likewise, if a component is rated for 50V you should probably assume that it will only handle ~40V just to be safe. Even in an engineering context it's better to design+construct a project for a worse scenario than you expect. Doing so allows some "breathing room" for unexpected situations.

[@RestoreTechnique]

Yeah but how to know if you've got a fake one 🤔 I don't have an oscilloscope

[@thomaslindell5448]

What happens if you swap in a legit ic

[@yasotube]

I think it will work

[@acoresbrasil]

​@@yasotubeor it will work better...😁

[@GiNodrog]

Opps you beet me to it haha. Same minds think alike.

[@AutieTortie]

It works properly.

[@Sparky_D]

It would be interesting to see the temperature test with a real IC.

[@Die-Spezialisten]

I have bought a lot of these modules few years ago. The output ripple is really bad. I got it a bit down by adding a line filter behind the output. You should not take the whole power, then it works well. If you need more you need to cool it. You can stick a heatsink on the chip. I use on of them for a fan under my living room radiator to increase the efficiency. The potentiometer is attached on a long cable. Works fine over years now. Greetings from Germany.

[@PipeManPeep]

Have you checked the switching frequency under different load conditions? Often, these types of devices will have a fixed "on" time, and vary the "off" time. This will vary the frequency. Also, check the inductor rating, as that will affect the switching frequency. These nodules are usually advertised as needing additional cooling at high loads. They are available with small heatsinks. If 0.4 v ripple is unacceptable for your application, a second LC stage is normal. I'd always use a local low ESR capacitor in my circuits anyway. Plus a very local 100nF MLCC near any sensitive I.C.s. I don't understand how you've concluded that the I.C.s are fake.

[@4211Rob]

Wow! This is excellent and you solved a noise problem I had with this circuit years ago. I'm going to Mouser.

[@bondarenkodf]

That's the Chinese amps. To get the real one, you need to divide them by 2 or 3. In some cases additional airflow or\and radiator is required. It's not perfect for mission critical stuff, but I never had any issues with this type of module.

[@wizrom3046]

"Chinese Amps" are like "American Horsepower" 😎 Like when you buy an American Shop-vac and the box says "9 hp" in big letters lol, but it plugs into a standard wall outlet...

[@iblesbosuok]

If the 53 kHz switching frequency works consistently, maybe it's the LM2576. Please compare the datasheets. Thanks

[@lonndawg7554]

I like the video... And your in-depth explanation on what you found, but here's my comment on the overheating and what my life's experience has shown for current rating of transistors, SCR's, TO-3 style Devices the traditional LM1205 series voltage regulators, .. all of these devices without a heatsink attached to them, they will get hot, to the point where their lifespan could be cut short Ha ha ha, go ahead and grab one of these devices that I have mentioned and see your results, When you apply their max rated current to them and see if they meet the criteria you are explaining in this video.(even a genuine manufacturers part) I'm sure there are manufacturers graph and charts showing the heat dissipation with applied current, "possibly" with and without a heatsink. this particular type of package used for "this LM2596 mounted on this circuit board isn't properly designed" to meet the maximum current rating. Life's lessons in learning electronics has demonstrated heat dissipation is another component in creating a proper circuit. But, You did find it interesting characteristic when you started using the scope, and that is the frequency, I interpret that to mean that is a telltale sign that that part is a fake part, and that bit of knowledge may help somebody in the future. This video does show a bit of skepticism Needs to be incorporated in the usage of these very inexpensive devices that come from our friends overseas 🙂 Great video thanks for the journey down memory lane,... about how I learned about the current rating on components, and heat dissipation. And when to incorporate a heat sink into the final design of something your building.

[@MrKentaroMotoPI]

Thanks, dude 😎 Most of these modules on eBay have the fake looking IC and a 35V output cap. There are some with the genuine looking IC and 50V output cap. No surprise, they are more expensive.

[@eccosabanovic1589]

I have used these and still do, in plenty of projects. Only thing i did is, added 1000uF capacitor on output pins. Works without any issues so far on plenty of devices delivered.

[@TranTek]

does that take care of the crappy cap they use and the LM2576 chip they use to boost it up to 150khz ?

[@Yukanhayt-Mhenow]

​@@TranTek No that'll just smoothen out the voltage/ less noise ripple

[@petersnape1537]

Most of your tests appear to be made with no load. The output ripple should be measured under load. Say at 0.5A, 1A, 2A, 3A. You may also find that the switching frequency will change under load. I found once that a step up converter was so lightly loaded that it took several oscillation periods to reduce the output voltage from the top of the ripple to the low point. If we assume fixed on and off voltages from the differential amplifier, the ripple frequency is dependent on the load current and the effective output capacitance of the converter. The low and high points of the ripple seem to correspond with the effective low and high sense voltages of the differential amplifier. The load seems to be very important in the operation of a switch mode converter. Peter snape

[@YouCanHasAccount]

I have a bunch of these fake modules as well. I usually replace the capacitors with salvaged ones. It does improve performance significantly. The ones it comes with are most likely fake as well and even then a 50V output capacitor is obviously not ideal to drive a 3.3V project with the higher ESR and lower capacitance per volume.

[@ecercuit]

At 1:12, I can see six burnt spots on the negative crocodile clip. How many takes you did?

[@Домсхем]

Thank you for detailed exaening and revers engineering that you have done for this widely used DC-DC step down converter.

[@Gaston413]

The sales advertisements for the modules on Ebay often point out that they can only be used up to 2A without an additional heat sink. However, I find a converter efficiency of less than 90% to be quite bad. This must also be due to the use of a diode as a freewheel.

[@JoJo-nq9ru]

Hey ,bro 👋 love from bangladesh 🇧🇩 yor explaining idea is soo butyfu ❤

[@DoktorJeep]

I routinely use these for low power applications. 100 - 200 mA. Never had issues. Would not try this with higher current but high current projects have more financial backing for more "real" stuff.

[@mynung7008]

The problem starts with the user expectation that you can build a cheap "one size fits all" SMPS. If you want the IC (fake or not) to perform well at minimal component costs you need to narrow down the range of input and output voltages and the desired current, then select suitable capacitor and inductor values. Examples: if you need a fixed output voltage of 12V only you can use an output capacitor rated at 16V. You will get a higher capacity and thus less ripple at the same cost and physical volume. You need less current? You can increase the inductance, but will the inductor will also have a higher resistance. And so on.

[@MrAruakdo]

You clarified a problem I had with this module. It simply exploded, I believe there was not much load, because its load was on standby. But the input oscillated a lot because it received very varied voltages, between 11v and 15v coming from a rectifier bridge. 9:27

[@klaust.2769]

Thank you for the video! I think the module is to small to handle 3A @ 5V output. Even the original datasheet from IT shows an efficiency of slightly over 80% under this conditions.

[@lilmatt1914]

HAHAHA, no way!!!! i think i bought something similar. they're were so cheap, i used multiple in parallel for my raspberry pis psu. they typically work well but sometimes i would get a low voltage warning and measuring them with an oscilloscope showed something similar to what you pointed out (that being that their voltage ripple is terrible). but yea i fixed it using a capacitor in parallel with their output, i don't understand why they didn't just up the microfarads of the one supplied if theyre already using a fake ic. i mean honestly its a great deal and i would probably buy more in the future just be sure for applications that require finer voltage adding a more capacitance. maybe they just didnt want the markings from a higher uf cap to reveal them as a fake.

[@deboraalves3267]

Thx very much! You are my best Chinese youtuber now!

[@peha524]

Japanese !

[@JoeMama-nk9xk]

Thanks!

[@davidkwok5720]

Simple and easy to understanding what you say. Useful lesson❗

[@ingotruppel9344]

Are you sure you measured the switching frequency? It might also be the frequency of the voltage regulation loop. Did you measure the ripple with load or without load?

[@guyteigh3375]

I think it is this particular design which did something I thought would be impossible - either this one (or one VERY similar in appearance) is capable of working in reverse. I didn't think that would be possible as I cannot see how the controller / charge pump would work. HOWEVER. if you have one of these spare and it is working just fine, then set it up (how I found out about this affect) so that the output is charging a cell or a battery - perhaps start off with an 6v input - and let it provide 4.18v to a Li-Ion cell. Now, just disconnect the power supply, and instead, place a meter on what WAS the input side. Get a reading? Now try adjusting what WAS the output voltage. What happens? It may overheat under load, it may be inefficient like this - but there is no avoiding the fact, it DOES work in reverse as a step-up. (or if not this one, then one which looks like a clone). There is also a tiny little "stamp" sized DC DC converter which does the same - here, I had a DC input that could be from 10V to 30V (covering 12v and 24v vehicles), it had a high efficiency green LED which was placed over the input, then the little module which kept a block of 18650 cells sitting at about 4.175 V. You have probably guessed already, that the green LED came straight on - without it being plugged into a cigar lighter socket. Happily working backwards. Ended up just adding a diode AFTER the green LED so it could only be powered by the incoming power - though that of course wasted 0.6v. Still confused about how in could work - as all the control circuitry is surely referenced to the INPUT side. All I know is at least two DC-DC step downs WILL work in reverse.

[@Video-ee1dn]

Funny, i just tested one of my old (fake) LM2596s modules, wich i bought several years ago. The switching freuency of these is nearly 200khz, so also out of spec. They work pretty well, one of them is used in my bedside lamp for nearly 10 years. The only problem i have with it is, that it is sensible for the mobile phone. If the phone sends data, near the light, it starts to flicker.

[@exgenica]

With virtually all of these kinds of circuits/components the maximum current rating (e.g. the 3A in this case) on the data sheet is dependent on having a heat dissipating system in place that complies with parameters also normally specified in the same datasheet. Without any heat removal the current such components can handle at a specified voltage is usually significantly lower than the maximum spec'd.

[@dougc314]

Typically derate an electrolytic voltage rating by at least 20-25%. So I would limit a 35 V capacitor to 28V, 30V max. The 50V input cap means 38V - 40V max input.

[@d614gakadoug9]

There is no compelling reason to derate the voltage of good-quality electrolytic capacitors. In fact good ones typically have specifications for how long they can be operated at a voltage above their steady-state rating. I used to design switchers and talked to aps engineers from some of the big-name manufacturers about this issue. Additionally, I've never seen any manufacturer recommend voltage derating in any of their applications notes for aluminum electrolytic caps. One thing you do need to be concerned with is the ripple current rating of capacitors. Buck converters input ripple current is very high but if the source impedance is reasonably low the input caps aren't worked too hard. Ripple current in the output filter caps can be controlled by design. Boost converters are brutal to the output caps. If you use monolithic ceramic capacitors you must be very careful about the absolutely horrendous negative voltage coefficient of capacitance of some types.

[@transmitterguy478]

Excellent video! I have these, but I will have to check them. If they are fake I will order original ICs to put on the board.

[@firasgh871]

Thank you for video. But wahy to bay original lm 2576 and test circuit.( fake vs original)

[@retroaria]

It happened to me, I bought a very cheap amplifier board for a small project. After assembling everything, I discovered that something was wrong and searching the internet, a forum gave me the tip: be aware of very cheap PCBs, manufacturing errors sometimes occur on a large scale and then this can end up in the trash and some collector finds it and puts it on the Internet. to sell. The reduced price prevents the buyer from complaining or denouncing the seller.

[@jamesg1367]

Had you considered replacing the counterfeit device with a genuine one, and comparing the performance?

[@Homestead_Adjacent]

Was the inductor sized correctly for the actual switching frequency?

[@sobertillnoon]

2:22 the way the ai voice says datasheet is amazing.

[@andygozzo72]

according to the datasheet the output voltage can 'oscillate' appearing like high ripple under some conditions, maybe just unsuitable inductor or schottky diode used? maybe this is the frequency the scope is measuring?

[@simonwatson2399]

Scope the voltage under load. These ICs change operate PWM in discontinuous mode. They also always need a heatsink at high current.

[@epiendless1128]

This was my first thought as well. But rewatching, he says the load is 1.5A.

[@simonwatson2399]

@@epiendless1128 I missed that for the scope tests.

[@FixitEasyDIY]

Excellent brother 😊 Thanks for the great video ❤

[@disablewong]

If you can have a laser scribber to reveal the situation inside the IC package, then compare with the original IC, this video would be perfect!❤ Anyway, u did a good job for this video already

[@AMATISIG]

That is a LM2576S-ADJ from UMW / Tech public or LM2575 ?

[@3maisons]

At 3:28 the voiceover says "bipolar transistor" but the diagram shows a MOSFET -- I wonder what you said in the original language.

[@zhuwkdal]

LM2596ですが「中は単純なスイッチ」と、あとはフィードバックの回路でできています。

[@fajar_engineer2043]

In datasheet freq is not fixed it will variable 110kHz to 200 kHz due to load, maybe already cc in your module

[@mikejones-vd3fg]

I put a 48v charger on in the input (54v) and tried to step it down to charge a 36v battery (42v) but it was stuck at 39v, couldnt get to 42v. Then it died as i tried charging my battery with the 39v. Then i realized the limits of the converter and, yeah it will break. The outputs became shorted which thanfully my attached load was able to handle but your might not, be careful. Thankfully since my load was able to handle the short it was also abloe to handle the 54v chargr without stepping it down in the end. You can charge a 36v battery with a 48v charger if the 36v can handle it and i think most do , or wont turn on if you exceeded their input range. Should be safe, but dont take my word for it.

[@Miata822]

Counterfeit ICs from Ebay, Ali Express, and Bang Good are very common. I have read comments elsewhere about these cheap DC-DC converters dramatically overheating. If you use them derate them significantly and keep separate from flammable materials..

[@malectric]

Worth noting that those capacitors (surface-mount aluminium electrolytics) are *not* the low ESR types required for switchers.

[@g4z-kb7ct]

The ripple is not an issue. I built a PSU using this IC and just added 4x more output capacitors. The ripple is nearly 0. For a simple home project the LM2575 re-badged as a LM2596 is more than good enough. For the little amount of money you really can't complain ;-)

[@ALEFILES]

Yes, these dc dc modules are cheapest than others similar, BUT the seller cannot lie with false promises...

[@g4z-kb7ct]

@@ALEFILESThe seller is CLUELESS. They just list what they see. All they are doing is buying from aliexpress for 50c and re-selling it as-is for $2.... clueless re-sellers are very very common.

[@ALEFILES]

@@g4z-kb7ct Yes, I've made a mistake. The really guilty behind these fakes are the builders, that print lies in its products...Greetings from Argentina!

[@maherf768]

Looking for Help! Hi I'm a beginner armature and I live in a place where I don't have access to tools or quality components. I use this cheap converter (12V to 5V) to power a small project (nano + dht22 + 6 relays + lcd) what capacitor type and value should I use if I want to reduce (remove?) the ripple? I read comments saying 1000uF and others 1uF/35V ceramic. any help would be appreciated🙏

[@jstep4146]

Those units where ordered in huge bulk and rejected (because of the fake IC et al) The company that made them then dumped them to wholesalers. There are about 20 different modules like that in various configurations that were rejected and dumped. When you see modules that cheap, there is a reason, and it is usually not the sellers that did the scamming. Most likely the builder thought the chips were genuine and had to eat the loss by selling them for almost nothing. It is cheaper for them to build new boards with genuine chips, than to rework the rejects.

[@muctop17]

I also had problems using such modules! Now you showed me some explanation. Thank you!🙏

[@user-rs8zg8ey2b]

You should try a decade cap box and tune the compensation cap (in parallel to the pot).

[@motoizm7893]

A very useful video. Can you test the xl4016 model as well?

[@christopherleubner6633]

It sounds like they recycled defective switching IC chips and packaged them in the same part format. This would probably explain the issues. It would definitely need a higher final capacitor and more inductance to work well at 50khz.

[@alxp-zd1zj]

They now have blank ic. Have you tried ps47e01 overvoltage undervoltage module? They do not work. The light is either permanently on or off when you apply electricity to them. It does not switch on the set voltage. I also noted that the plus situated strange, Is there anything uploaded about them? They are more expensive so throwing away money to that

[@JouMxyzptlk]

Hey, I got some of those modules! Now I know why it did not work everywhere as expected: Way too low efficiency and that ripple.

[@asm_nop]

For me, the first indication of a counterfeit part is that the part number is laser-engraved with a "Serif" style font, which is significantly harder to engrave, and gives worse visual quality. For some reason the Chinese are obsessed with using Serif fonts on everything. Any cheap knockoff PCB, chip, device, etc. from China almost always has a serif font on it somewhere, and it is more of a red flag to me than Comic Sans.

[@tookitogo]

Yeah, 100% agree. For some reason, the fonts for Chinese and Japanese seem to most commonly have serif characters for the Latin alphabet. So when they design things using one of their fonts (even if they don’t need any Chinese or Japanese characters), and then just enter Latin text, we get the serifs.

[@louisbellanca9293]

I just notied this issue when I tried to use this design for a 16V to 5V usb power. My devices were acting strange and I scoped the output and noticed the same noise you show. The 500mv or so noise was exceedng the USB range of 4.75 to 5.5V. I had to add extra output capacitor to reduce the ripple to stay within the USB range.

[@EdgarLTShadow]

The efficiency can be increased by changing the inductor. In the datasheet there is a equations and examples, the problem is the frequency is used to calculate the inductor. The hot because it is not a synchronous source and according to the diagram hte ic use bjts for switching and not mosfets. It is not the best sdc dc buck but for 1 or 2 amp projects it is good and cheaper.

[@stanleydsouza6723]

Appreciate Your valuable information greatly, so to save scores of people who struggling to solve a problem in line with repairs..fault findings of various Electronics Service

[@rajanne2947]

Interesting! I bought 4 of these! I used one for charging my car battery! At the input of this module, I have a 220v-18v AC transformer followed by a 3amp bridge rectifier and 2x470 mfd capacitors. I run it at 2amps because of the overheating! I adjust it's output to 13.5 v needed by my car battery. I think it gives me peace of mind because of short circuit protection! And the fact it will stop charging automatically at peak car battery voltage!

[@batteryman2852]

the 50% amp rating is a trend i'v seen on many boost converters i have, i was thinking the rating was a peak rating, as its not meant for long run, but fake IC's could indeed be a reason, as i just discovered a few weeks ago when i wanted to salvage some Mosfets from a board from AliExpress, and i had troubles, when i ordered genuine Mosfets, i notice the diode reading on the Mosfets from the salvaged board was 500+ , but the real one whas near zero.

[@ElektrischInkorrekt]

To be honest, yes the font CAN change between different lots. Also the packaging and the way and layout the marking is printed on can change printed. I have some original charges, where this is applicable to. (Also different manufacturers, e.g. TI, NS, etc.) Especially if the line is produced over a longer timespan, you can find these differences...

[@effedrien]

Power components are typically produced with older equipment, everything is relatively big and doesn't require high end equipment anyway. As a result there are bigger differences in the calibration of different production lines, resulting in slightly different character sizes and length width ratios, and even the marking depth can vary a bit. A completely different font, that is a bit unusual, but it can happen, like some ancient machine only uses special files that no one knows how to create new fonts ;)

[@effedrien]

And even within 1 batch coming from 1 lasermarker, the whole mark often shifts over relatively large distance. That is because such components lay in trays during the marking, and the components can move a bit in the trays, and each column in the tray is located under a bit different angle from the laser to make it more difficult

[@ElektrischInkorrekt]

@@effedrien One of the device-groups, I regularely see it, are e.g. parts from the 74xx-Logic family (as I have many of them from different decades and multiple manufacturers, some even dating back to 1969 or older). These IC's are e.g. from HFO (not existent anymore), TI, NS, Signetics(also dead), Siemens (They tried something different... SN7400N is the equivalent to their FLH101)

[@nisalsawinda8804]

Hey buddy I want to know about MT3608 buck converter... Because I plan to use it for some projects ... pls review

[@lsdlsdk4845]

Do you mean that Amozon had delivered a duplicate dc to dc module OR the Manufacturer has done the fraud of selling a Duplicate DC to DC module?

[@BrokenCircuitRanch]

Nothing like debunking a dollar buck converter using 12,000.00 USD in test equipment ;)

[@vishva8kumara]

For that lower frequency an output capacitor of greater capacity (x 3 ~ 5) would reduce the ripples.

[@Selyan_]

incredible videos all over your channel, very informative, thank you

[@ewertontsk]

I bought one step down type and gave me the same performance. the transformer got quite hot, but stopped getting hot after removing and installing again. Thanks