In my opinion, videos like this are some of the most powerful for all the DIY hobbyists. Like me, most of us have no formal training in electronics, but this video gives us very relevant reasons to try and wrap our heads around complex and somewhat arbitrary concepts. Having a broad understanding of what might cause noise or ripple on a power supply / buck-boost converter is in every DIY hobbyist's best interest. Please make more content like this!
@@greatscottlab Didnt check it, but probably an edit of the previous video would be quite useful too. Not everyone in the future is going to watch both
Ich habe das gleiche Teil in Verwendung und mich extrem über das Ergebnis gewundert. Ich habe das Netzteil analysiert, da wir Industrie Netzteile suchen, die einen kleinen Formfaktor haben, programmierbar sind und ca 500W bei ca 50V können. Das Netzteil kam in die engere Auswahl. Das Ding ist wirklich gut, da können andere echt einpacken. Das einzige, es hat kein DIN Rail.
That was both an honourable and brave move, much appreciated! I was thinking to get one of those PSUs, but decided not to when you mentioned the output noise.
I was hoping you would do a video like this. Taking cheap aliexpress or ebay electronics that are ALMOST good enough and then cheaply modifying them to fix or improve their performance. You should make more of these.
You typically do not put large MLCCs on the output of a power supply because when they fail, they fail with a burn so you use small ceramic caps paired with low ESR caps
I am curious about your answer, so I get deep into it, there is an article from TDK, called"Guide to replacing an electrolytic capacitor with a MLCC", it recommends to replace with MLCC due to their low ESR, long life and small package, but one should be careful about their capacitance will change with temparature and voltage applied to them, also some ICs are intended for electrolytic capacitor, so it might cause some unwanted feedback with MLCC. Also there is a youtube video discuss about fail of MLCC called "EEVblog #1037 - Solving Ceramic Capacitor Cracking ". Large value MLCC can be cracked physically easily, and it might get shorted when that happened, I guess that's why modern hign-end PC motherboard still using the method you said, rather than pure MLCC method.
@@tingoyeh4903 I had soldered a MLCC underneath an electrolytic in an ATX PSU which had undesirable high ripple output. Eventually while I was at the computer, there was a pop sound. The MLCC probably became hot and failed short which burnt down the electrolytic releasing gas.
Ah the classic current loop antenna, that’s why you were picking up all that noise. Any loop will act as a current antenna. And any length of copper, will act as a monopole if adjacent to a ground plane or bipole if connect to a center point. Good luck.
That's why you should always use your devices inside a hermetically sealed bunker surrounded by 10cm copper plate, 1.2m lead plate, and 3m of grounded-steel-reinforced concrete. Everything should be made with pre-nuclear-test materials to reduce self-radiation. The bunker should also be carved into the base of a mountain, or built into the sea-bed with at least 200m of water overhead if you want to block those pesky cosmic rays. You still need to account for all the Neutrinos that can still easily get through, but luckily those usually don't interact with your electronics if they can get through all your other defenses.
1. To save money 2. The only people that are going to care/notice are people like us. 3. The people like us know how to fix the problem. My guess. Keep up the good work bud.
Care? No. The only people who are likely to KNOW why their thing fails, and also choose not to buy from the same supplier, are us. This is simply capitalism at work. It is more profitable to sell more of a thing that is cheap-but-bad than it is to sell something slightly more expensive-but-better, even accounting for the returns. And since you change the company name/registration for every product, you don't care about repeat business. It's a cynical take, but this really is built-in to the business plan nowadays.
Cost is its own spec. If you want higher performance for your use case, be prepared to pay more. If the performance of the cheap part is acceptable, you can pay less for something that satisfies you. ETA: And no, it's not "capitalism," it's scarcity. When resources are limited, it doesn't make sense to spend them on performance you don't need. You need Star Trek replicators to escape this, not a replacement political economy.
@@jdotoz Managing scarcity is the PRIMARY job of any economic system. It is WHY we have economies. (Ultra)Capitalism is a BAD economic system because the way we are using it now is tuned almost exclusively to "make money number go up" instead of ACTUALLY managing resource scarcity and distribution. So yes, the problem IS capitalism, because the SYSTEM punishes you for making things well if it costs less to make them worse. None of the economic systems we use now can handle short-term products that break prematurely. They can't self-regulate when there is too much turbulence.
It REALLY grinds my gears when a product gets ruined by bean counter measures. "Oh we saved a few cents on this device!" "But its complete crap now?" "Who cares? We get money!"
Man, it's crazy how much you can accomplish pretty cost-effectively if you understand electronics. It's something I'm trying to learn, but it's not easy for me to understand electricity and how everything relates to each other. Thanks for the videos, they're helpful in slowly getting more understanding of things.
This is true for anything, from auto mechanics, plumbers, carpenters, yada yada. Outside of basic volts, amps, Ohm's Law, etc, electronics are going to require a fairly deep amount of study/work before you're going to be able to figure much out not too mention fixing a borked design. I've decided to spend the extra 25% or so for something thats generally been shown to be good vs trying to eke out the cheapest component/parts. If I'm trying to cobble something together, the last thing I want to do is be wondering if some pre-made component is even doing what I bought it too do.
Don't worry about some of these things. The higher the frequency the more electronics starts to look like black magic. That's even to people who have degrees in it or similar fields.
@@arthurmoore9488lmao, for real. I work in automation and the amount of troubleshooting I have to do with highly sensitive equipment near robots, or other high frequency loads is aggravating. Sometimes the shield isn’t grounded well enough, sometimes the shield is too small, sometimes it’s grounded too well and creates a ground loop that effectively acts like an antenna that picks up RF noise. Part of me loves it, part of me hates it. Haha
Man, I've been an electronics hobbyist since the late 1980's, and I'm always surprised at how little I know. Just keep tinkering. And when you get a chance to use an oscilloscope, seize the opportunity
Honestly, if it's a $0.50 component, that's why they didn't include it. They're shaving pennies off of the price of these things and producing them in huge numbers with tiny profits to stay in business. A device that is 90% effective is good enough from that viewpoint.
The tip at 9:56 about measuring temperatures to determine if reactive components have appropriate parameters was really helpful. That point alone made this video valuable.
That ringing on the switch node of the inductor can be suppressed with a snubber network and is easily missed, this I know from first hand experience! Also sometimes adding a beefy cap can make things worse, Mach1Designs did an EMC video on the caps and is very helpful in locating noise and killing it.
I never knew to check temperature to check out components in a design to see if good or bad. Only to look for a broken component(s) when fixing. Maybe they stole an incomplete design from someone else? Lazy? Possibly just stupid? 🎉 I am still learning. I bought an old radio/cassette player and a radio clock alarm that both work to study on. I will use your temperature trick to maybe improve them. Thank you for giving us a great channel with interesting projects.
Your devices were probably made in the 90s or earlier, so I wouldn't count on them having the same trashy engineering as modern devices from AliExpress. Good luck with your endeavors though.
High frequencies tend to cause caps & mosfets to heat, for different reasons, but the solution is pretty much the same: lower the resistance. Low ESR caps have much lower resistance, and 'bigger' mosfets have lower RDSon, which is essentially the resistance that causes most of their heat. It's always funny to see a low-tech troubleshooting technique on high frequency IC's. Elegance in it's simplicity.
Seconded. The diode temp probe is a precise method of spotting an overheating device. Slower but many times better than a finger touch. I'm impressed by your thoroughness.
I used a pile of old 1ohm 50W resistors to make a grid on an aluminum plate. You can jumper quite a few different loads and can dissipate lots of heat.
For the UPS I simply replaced the 2x220 uF capacitors with a 22 uF tantalum low ESR and a 470uF electrolytic as recommended by the chip data sheet. That cleaned up the noise. I used 1x probe and attenuation.
My guess for why they didn't go the slightly more expensive route is sourcing. This is all guesswork mind you, but if they produce in house, it's only costing them pennies per. One thing I've noticed about Ali is many vendors offer the same product at different prices, so that would explain why they want to keep the price low...sort of a lowest bidder thing. Lastly, the 50 cent price you quoted is the retail price, so it would slightly increase their profit margin as they'd be paying wholesale. 😊
For a lot of stuff on Ali/AE, there are one or two companies making gajillions of these as OEM-Contract. Some resellers want different color silkscreen, some spec differrent values for some components to get a lower BOM, etc. Chinese business looks at cost cutting as the primary goal bar none. Doesn't matter if it impact the specs of these design, unless/until they are called on it and IF they can be forced to pay some sort of remuneration. It is always worth it to keep value engineering a design/product which leads to higher profit, until/unless they are forced to remunerate. There is not a lot of care/f*cks given for 'Reputation' as you see from Japanese, EU, Noram companies. Talk to some engineers from China, and they will all admit it. They don't like seeing their work cut down to out of spec anymore than anyone else.
9:47 your forgetting that at the volume they're producing these at the increase in BOM cost is even less at under 0.19 EUR / piece on orders of 1000. Probably also a slight price decrease at 10k and 25k as well. Therefore at volume this is likely no more than 0.15 - 0.17 EUR in parts, and the lower quality parts they're using probably cost 0.08-0.11 EUR, so we're talking about them pinching literal pennies to not have a better product.
@@greatscottlab yeah unfortunately so. Though I know BigClive in his teardowns of electronics like this from china on aliexpress or otherwise mentions that they tend to have resistors and capacitors overdriven for their intended purpose for whatever reason.
@@pappaflammyboi5799 Dont forget about what the product itself cost spending 0,08€ on a 1€ board is a huge increase in price. For most stuff you dont need that clean power, so it is good practice to not build them in.
One major contributing factor to noisy outputs on many cheap/Asian Aliexpress-alike regulators is poor layout. Take e.g. 8:50 - they use polygon pours to reduce impedance (good) and provide more thermal mass/lower thermal resistance (good), but all the components are still awfully, widely spaced, leaving still considerable inductances, which is especially bad on the switch node, as it has high current ripple. As V = L * di/dt and your L (trace inductance) is high, your di is high and your dt is small, this creates ripple noise. And as layout is something inherent to your PCB (it IS your PCB basically), you cannot easily change the noise created from bad layout. Even the best components can deliver embarrasing performance if you do not do the layout right. Remeber, electronics is not so much about charges moving, but about controlling and containing your fields. But there are far more talented people out there, which can explain that in much better detail and accuracy.
A lot of noise I deal with in electronics is 60 hertz noise because all the power supplies use a iron core transformer instead of a ferrite core transformer and the reason why that makes a difference is because all iron core transformers run at 50 or 60 hertz well ferrite transformers run at higher frequencies. And 60 hertz noise is very noticeable if you're using inductors and highly sensitive electronics
That's a wonderful example of what a difference good capacitors make. I'm amazed that the short probe ground wire makes such a difference. Have you tried measuring at 20MHz with the ground wire?
I too have burned a few MT3608's...they do seem to be unreliable. At 11:09 you say add a 2.2 microfarad ceramic capacitor but the links in the description are for different ones so I am unsure which is correct.
I think it is because of cost. Like the MT3608 boost converter, here in Indonesia it listed around 5200 to 6500 IDR (0.32€ to 0.37€). By adding the better ceramic capacitor priced 0.5€, imagine how much it affecting the price. For me, the initial design is more than enough, reminding the price is dirt cheap. Unless for specific reason I need to modify, like the one you tell about creating negative voltage, I'm set my expectation according to the price I paid 😅
Well, a 20% 22uF MLCC costs 500 IDR (0.03 EUR) on Indonesia's Tokoijo, and the manufacturer would have been able to source it for cheaper. So it's not an excuse.
One of the 1st things our teachers told us at LTI is the difference between an expensive VCR and a cheap VCR is a 5 cent component. Like they will use a 1/4 watt resistor to dissipate a 1/4 watt of power rather than using a half watt resistor. Everything today is built down to a price, its the race to the bottom which got us lots of cheap junk.
The side ground pin in the oscilloscope probe is a very nice solution. A spring loaded measuring pin would still improve measuring a shivering hand (a so called test pin "needle" for production beds).
I always see keysight sponsoring many electronics chanbels but it would be nice if you could make a video how it looks inside one of their mesuring devices (maybe you could get a broken one or something from keysight from their rma or something like that )
50 cents can be a lot on a large scale production. Aluminum caps are easily replaceble and do less damage the board than the ceramic ones when they fail.
I have some experience in real market production and can take an educated guess to why such things occur: Case A - The Manufacturer is not aware of the issue due to one of the following: 1 - prototype worked well but somewhere in production someone deviated from the design due to the lack of some component 2 - a partner provided a batch of bad components 3 - lazy (minimum testing) or non-existent QC 4 - following (copying) a design blindly not knowing how slight changes might affect the result 5 - Damaged components due to bad quality / bad storage Case B - The Manufacturer is aware of the issue: 1 - Greed 2 - discovering the issue in late stages of production (happens a lot) so they lower the price to not throw away the stock and to avoid fixing the issue which is sometimes very hard to near impossible / costly 3 - they just don't care 4 - indented for uses with high tolerance to error
This was a very interesting and also useful video! I found especially interesting the bandwidth limit on the oscilloscope. Can you make a video to explain why it's limited during official measurements, the implications about having strong peaks beyond that bandwidth limit and how to absorb them if so desired (I know this part: ferrites rings/beads around the wire). Thanks!
At high frequencies such as in these switching power supplies, standard electrolytic capacitors have too high of an “equivalent series resistance” (ESR) and should not be used in this type of application; that ESR leads to excessive heating. Low ESR electrolytic capacitors are designed for these applications. The small ceramic capacitor heating up is a bit surprising since they are intended for use with high frequencies; perhaps that was a cheaply made part, and that is why it showed a higher than expected temperature.
Oftentimes, these boards are based around one key component, i.e. the regulator IC. If you were to reverse engineer the board and compare the resulting schematics to the IC's reference design, you'll find probably little to no differences. And very often, the designers doesn't really pay much attention to the textual parts of the datasheet that explains the values and different usage cases, or just use one calculation for all their product runs (i.e. the same capacitors for fixed 3.3V, 5V, 12V and variable variants of the board). Also, the reason they won't add it is cost and availability. They already most likely have tens of thousands of cheap small mlcc caps around, adding one to the output won't change the BOM cost by much, but also doesn't have to affect the output in the desirable way, or even hurt it. Thank you for showing us this
I bought the UPS circuit in 5V and 12V versions, but I've received 9V instead of 12V, so from this video I've easily recognised the resistors to swap for the voltage divider, but buying the exact components is not feasible due to the shipping costs, so I went to the harvest route. The problem is that my multimeter is not very good at measuring resistance, so I harvested some similar resistor from another circuit that had similar resistance without really knowing the exact number. Well the good part is that I managed to obtain 7V, 10V, 13,5V and 20V and left it at 13,5 that is the most similar (also should allow more power and staying close to 12V during high loads). It will be useful to know the exact resistance, maybe you could pick it up again and make a couple of final upgrades: a potentiometer to regulate the voltage and an interruptor to disconnect the batteries when not needed. Maybe design a cool box 🙃. Having a clean output is not always critical, especially if I know such regulation is done at the receiving end.
Just received mine, a 5V and a 12V. Will be using 1 for a solar project. Just be careful with the voltages! Capacitors seem rated at 16V max and I would stay at least 10 to 20% below that. Checked my 12V version and that smd resistor R7 has a marking 9092 which is 90.9 KOhm or 90.900 Ohm. Also the board has been changed as his has R7 and R7* . The R7* has turned into R12. I would always opt for a clean output. Spikes can produce noise on radios if you use longer wires and in some cases cause strange effects. I always also place a 100nF cap on the input and output.
Actually ended up using one of the smaller ones for an amp, and had added a larger capacitor to the output to support the bass. good to know it's probably helping with noise as well.
I also want to know. I bought 2 of each, 5V, 9V and 12V, for alarm/cameras and just in case. One seller has it for 4$ each and there was also 1$ from 5$ coupon for playing (and losing) in Go go match game, but this promo ended. So they were really cheap.
A lot of people don't understand ceramic capacitor dielectrics. Y5V ceramic capacitors are cheap but can lose 90%+ of their capacitance at elevated temperatures and also generally have poor ESR/ESL. X5R and X7R etc are much better but cost more. I never bother with Y5V and stick with X5R/X7R/X8R/C0G depending on the application. Standard aluminium electros usually also are only effective up to maybe 200kHz. Above that you really need to use polymer and/or tantalum caps or paralleled ceramics.
Replacing the output capacitor with a low ESR one reduces noise, but also the power supply ripple. Often that ripple is needed for stability of the voltage regulator. I would always recommend to stick with replacements to similar components and not deviate too much from the original implementation, unless one can measure the stability afterwards. Bad stability is often much worse than noise.
BTW, good on ya, for showing the error in probing technique! It is *such* a common issue that it's kind of like "is the power plugged in" question for oscilloscope AEs !
Well... I've got some of those mini boost converters and i already knew that something wasn't right. Only a small SMD capacitor on the output. On my projects when i used those boost converter modules, i always added some extra capacitors on the output (Both electrolythic and MLCC). DC-DC converters are INCREDIBLY useful in basically everything. I wanted to ask if you can make a video where you analize a 'special kind' of DC-DC converter, the one used in ALL car audio amplifiers: A single rail to double supply rail DC-DC converter. I could not find any appropriate schematics anywhere. This DC-DC converter is really useful but i just cant find a way to make a proper DIY version. A lot of people is probably searching a way to make this DIY... I think that a video about this topic will explode on views! (And help a lot of people too). A DIY version is really versatile, because you can have full control over both the output voltage and the power that the circuit can handle. It would also be a nice learning experience. With this circuit, a battery can run ANY audio amplifier!!
I ordered the parts from Mouser, so now the two UPS devices going to get the upgrade. I even have 5 of those boost converters to that I will do the same with. Thanks for this great fix video. Great stuff as always. Now I just need to find a cheap-ass oscilloscope, so I can test other electronics I have lying around.
Yeah I've been trying to measure a gas sensor, and I've found that every USB power supply I plug my NodeMCU board into produces a different result on the gas sensor. An Apple iPad charger makes the analog measurement noisy. A Samsung charger makes the measurement get cut in half. It really makes a huge difference in my circuit.
Two 100K ohm resistors, a 2.5 V reference diode, and an OpAmp and you can also get a very stable 5V with current up to the opamp’s output current rating. Also, thank you for admitting a mistake, many wouldn’t be willing to say that, very cool!
Great video again! 👍 When dealing with digital circuits, I rarely find myself having to deal with noise problems. Noise only becomes a factor when I'm working with analogue circuits. I often use a linear voltage regulator in combination with a low pass filter(LC or RC with low-value resistor) and of course decent decoupling for every chip. And it rarely stops there, one must consider whether circuit isolation is necessary to prevent current loops, as well as determine if guard traces should be implemented(for example an opamp circuit with high impedance load), etc.
I bought some of the small MT3608 boost converters and i also measured some pretty bad noise at higher output currents. Adding a 1,5uF ceramic capacitor and the problem is fixed. The ceramic caps I used literally cost 6 cents per piece
Technically, they didn't need to raise the price by 50 cents, the other 2 capacitors won't be needed anymore therefore the cost would cover these new 2. The reality is that this board was designed by someone that either didn't take noise into consideration by accident or didn't care about it at all.
100% about cost. Sure the part may only be 3 cents for them at bulk rates, but they already have the tooling setup for these PCBs and the components on them, and they're making truck loads of them per day, those 3 pennies per unit add up.
There is a reason why I always use a high quality 100nF capacitors over Voltage rails and most of the times in parallel over bigger caps. You can clearly see a lot of garbage, aka noise, when not doing so. For reducing RF that these switched things can cause also use ferrit to stop interference. Some of these Chinese pcb's are really well designed and then you order another one and measurements are different. Bad capacitors, change in pcb layout to make it cheaper. A bit of a lottery. That last pcb, released some smoke on those. They will go into self destruct like you said, but I still like them a lot.
I was confused when you mentioned the very high pp noise value in the prev video cz I have this power supply and it works according to specs and the manufacturer is very keen on improving their products. I also assumed noise pickup. Assumed adding a very small resistive load would fix it. Thanks for revisiting the test!
There are plenty of off-the-shelf USB Battery Banks that can replicate the 5Vdc "UPS" function. They'll output 5Vdc even when charging, or not charging. You don't need to go out of your way to buy one; you probably already have several.
They might just do that now. We are constantly reinventing the wheel and working towards making something as efficient as possible. Thanks for the videos you put out man it has significantly added to my progression of circuit design.
You can probably make a fair bit of money by designing neat little circuits like these that actually work right, and sell them on your store. I'd happily pay reasonable markup for something I know was designed properly, characterized, and tested.
I used to do component level troubleshooting for a manufacturer of cable tv monitoring equipment. Here is the answer to the question of why don't they use better components, at least for them. Their engineering team used 1% tolerance parts when designing and prototyping. They would hand off the list of components needed never specifying what tolerance (if any) of parts should be used. Naturally, procurement bought the cheapest stuff they could find. I practically begged management to let me talk to the engineering team. They finally brought one guy down. I sat him down, showed him the log I had been keeping, and the exact parts that needed tighter tolerances. In this case it turned out to be the inductors in a frequency generation circuit. The board pass/fail rate jumped to over 70% (it was in the upper 50's). My big reward for figuring that out was I got to be the end of the line guy. If I couldn't fix it, it went into the trash. I didn't last long there
Very clearly presented information, even I have understood this with my limited knowledge. Love electronic and what can be done with it. Thanks so much.
My guess for why a manufacturer wouldn't include the fixing product is just the cost of the better capacitors most likely being higher than the ones they used, and most of their customers probably don't check the outputs noise, but honestly I couldn't tell you if that is truly the reason.
06:22 Additional to further reduce output noise, add capacitor parallel to R1. Start with say 0.1uF. This will inject output noise in feedback path & can help reduce output noise further.
that makes for a PID controller, you are injecting a part of the noise in the PWM regulation which could create instability... If the dynamic regulation is OK, you should not touch it but improve output filtering IMHO
bois, i m new here with this type of technology and soldering etc, i'll buy soldering iron from aliexpress costs 75 SARS, but im not sure if its working perfectly for me but i'l buy it and testing and Learn how to solder like a pro!, thank you great scott for a knowloge of things like this and ofc it will take me a long run to a level pros
They don't do it because they don't need to -- 99%+ people will not be analyzing the signal, and a sloppy 5V is fine enough for the things people buy these for; either the device isn't sensitive, or has its own built-in regulation. Nonetheless, good info. Next time I need to use a cheapo boost/buck converter, I'll check the signal and maybe throw a post-regulator on there.
8:14 i love how you put the images of components next to the schematic components -- for those of us trying to take self learning electronics seriously, it massively helps our brains put to mind the components as if they were legos (but the various values of each 'lego' is like the lego color). Next after looking at this for a while as a reference during test building on our own, the schematic components begin to fit like a catalog in our brains before we even realize it. I hope i not sounding too nuts but this really works for my brain very exceleratredly, ty for this idea and i hope you and others do that much more often -- even if just once before you do the red pen work on the circuit would be a big help ^_^ p.s. i have loved your work for last many years, ty for keep doing your channel ! 😀
I'd guess that they don't include those capacitors because they're aiming for minimal viable product, minimum cost, and maximum profit. If they add $1 of parts and need to bump the price $1, a competitor might not do that and sell more because it's cheaper... Especially since these probably sell like hotcakes and are acceptable for most things people use them for like lights or things with their own regulation/filtering
So you achieved that by replacing 220uf electrolytic with 22uf multilayer ceramic SMD cap(huge one). But ceramic capacitors are capable of extreme current spikes, much larger then electrolytic ones, though since it is used in place of 220uf electrolytic total capacitance load may not be as bad
Tantalum / MLCC Caps seems to solve most noise problem which is why a lot of audiophiles just spam them a lot around their board like Onkyo or Denon when they designed their board.
This is not a complaint about this fantastic video on what can be done to achieve better results from inexpensive switching power supplies, just a reminder that buying what appears to be inexpensive component parts can be very expensive when you come to pay for them to be shipped to you! 50c for a component is a great fix, but let us remember that no component supplier is going to ship you a 50c component for free, most suppliers have a minimum order price + shipping, which can make small projects VERY expensive to build. This is even more frustrating here in the UK where buying let us say GBP 10.00 in components can result in you paying GBP 10.00 in shipping, so when you want a GBP 0.50 regulator you have to either buy 100 off of them or pay the minimum order charge plus their shipping. Sorry for the whine, but a lot of TH-cam videos on building small electronic items forget these hidden costs.
Well done. Fix the problem rather than ridiculing it and recommending something that costs 5 times the price... or worse not recommending an alternative at all!
Thank you very much for the video. How could you make the UPS not charge the battery at such a high voltage? For example, charge it only at 4.1v to increase battery life. Thank you so much
Having been exposed in the power supply side of things for so long, the main failure had always been due to some cheap capacitor. And this applies to even big brands. It's very simple, cost! Just because it costs 50c more on a component level, it will raise the final sale price by $5 or more. Also, don't forget products are designed to fail after warranty!
Thanks, I use those and don't have the knowledge or time to dive into this like you do. That was extremely helpful, since the fix doesn't require more than basic skills :)
Was just getting ready to send some Gerber files to the PCB place, now I have to make a few capacitor changes first. AAAARGH!...more work but I'm thankful for the knowledge! I get to Deutschland again I owe you a bier!...Prost my friend!
Oja, mit dem MT3608-Board (das letzte vorgestellte) habe ich auch so meine Erfahrungen gesammelt. Mein Use-Case: Reste eines 24V-LED-Bandes an einer etwas größeren 12V-LiFePO4-Batterie betreiben. Im Test an schwächeren Batterien und mit Netzteil kein Problem, am großen Akku ist mir eins nach dem anderen abgeraucht. Nach etwas Messerei mit dem Oszilloskop: Der Akku kann aus Sicht des kleinen Schaltkreises quasi unendlich viel Strom liefern, so dass der Ausgangskondensator beim Einschalten einen starken Impuls durch die Induktivität erzeugt hat. Dabei kam es, nachdem der Kondensator voll war und der Stromfluss abgebrochen ist, zu einer Spannungsspitze von über 40V. Eventuell hat da der DC-Bias der Billigstkondensatoren auch eine Rolle gespielt. Hübsch zu viel für den bis zu 30V spezifizierten Chip. Lösung: ein klassischer Elko parallel, außerdem den Regler am Ende des nicht ganz so dicken Anschlusskabels platziert statt am Anfang. Jetzt funktionierts...
I see a lot of what I call "hopeful" or "expectant" engineering especially in the cheap but plentiful switching power supplies online. I expect noise value to be absurd, and often wind up modifying supplies to remove noise. Generally, I try to design my stuff as though the cleanest DC power it will see is half-wave rectified AC. I'm an electrician and hobbyist. Cool discussion of the incredibly low cost of doing it right, especially the first time.
Have you tried asking the Chinese seller why they don't use the other capacitor? In my experience the Chinese have quite good customer support. The answer may be difficult to understand due to machine translation, but they usually do answer. Your request may even prompt the Chinese to change their PCB to use the other capacitor in the future.
I would imagine the capacitor on the MT3608 boost converter is not included as it would reduce efficiency slightly and maybe in most cases isn't functionally needed? I could be wrong.
![[#2024MAMA] ROSÉ (로제), Bruno Mars - APT. | Mnet 241122 방송](http://i.ytimg.com/vi/dgGqD28J6aQ/mqdefault.jpg)
In my opinion, videos like this are some of the most powerful for all the DIY hobbyists. Like me, most of us have no formal training in electronics, but this video gives us very relevant reasons to try and wrap our heads around complex and somewhat arbitrary concepts. Having a broad understanding of what might cause noise or ripple on a power supply / buck-boost converter is in every DIY hobbyist's best interest. Please make more content like this!
Mad props for clearing up the mistake you made in the other video.
I am always open to admitting mistakes :-)
@@greatscottlab Didnt check it, but probably an edit of the previous video would be quite useful too. Not everyone in the future is going to watch both
Ich habe das gleiche Teil in Verwendung und mich extrem über das Ergebnis gewundert. Ich habe das Netzteil analysiert, da wir Industrie Netzteile suchen, die einen kleinen Formfaktor haben, programmierbar sind und ca 500W bei ca 50V können. Das Netzteil kam in die engere Auswahl. Das Ding ist wirklich gut, da können andere echt einpacken. Das einzige, es hat kein DIN Rail.
That was both an honourable and brave move, much appreciated! I was thinking to get one of those PSUs, but decided not to when you mentioned the output noise.
I've got this in my cart at Ali so now i can buy without doubts
I was hoping you would do a video like this. Taking cheap aliexpress or ebay electronics that are ALMOST good enough and then cheaply modifying them to fix or improve their performance. You should make more of these.
YOH - Chinese labor is getting expensive, between their demographic loss and "OUR" re-shoring the supply chain - enjoy while u got it. GOT DAT ??
You typically do not put large MLCCs on the output of a power supply because when they fail, they fail with a burn
so you use small ceramic caps paired with low ESR caps
I am curious about your answer, so I get deep into it, there is an article from TDK, called"Guide to replacing an electrolytic capacitor with a MLCC", it recommends to replace with MLCC due to their low ESR, long life and small package, but one should be careful about their capacitance will change with temparature and voltage applied to them, also some ICs are intended for electrolytic capacitor, so it might cause some unwanted feedback with MLCC.
Also there is a youtube video discuss about fail of MLCC called "EEVblog #1037 - Solving Ceramic Capacitor Cracking
". Large value MLCC can be cracked physically easily, and it might get shorted when that happened, I guess that's why modern hign-end PC motherboard still using the method you said, rather than pure MLCC method.
@@tingoyeh4903 I had soldered a MLCC underneath an electrolytic in an ATX PSU which had undesirable high ripple output. Eventually while I was at the computer, there was a pop sound. The MLCC probably became hot and failed short which burnt down the electrolytic releasing gas.
Ah the classic current loop antenna, that’s why you were picking up all that noise. Any loop will act as a current antenna. And any length of copper, will act as a monopole if adjacent to a ground plane or bipole if connect to a center point. Good luck.
People who liked this also liked working with RF designs, where furrowing your brow rotates you on the smith chart.
Though the change he saw was huge and probably this is also telling for the EMC of those little PCBs.
That's why you should always use your devices inside a hermetically sealed bunker surrounded by 10cm copper plate, 1.2m lead plate, and 3m of grounded-steel-reinforced concrete. Everything should be made with pre-nuclear-test materials to reduce self-radiation.
The bunker should also be carved into the base of a mountain, or built into the sea-bed with at least 200m of water overhead if you want to block those pesky cosmic rays.
You still need to account for all the Neutrinos that can still easily get through, but luckily those usually don't interact with your electronics if they can get through all your other defenses.
@@Prophes0r Just build your swimming pool over your bunker.... ppfff...
@@hugegamer5988 nothing like carving your own ground leads to get the ground loop as short as possible
1. To save money
2. The only people that are going to care/notice are people like us.
3. The people like us know how to fix the problem.
My guess.
Keep up the good work bud.
Care? No.
The only people who are likely to KNOW why their thing fails, and also choose not to buy from the same supplier, are us.
This is simply capitalism at work. It is more profitable to sell more of a thing that is cheap-but-bad than it is to sell something slightly more expensive-but-better, even accounting for the returns. And since you change the company name/registration for every product, you don't care about repeat business.
It's a cynical take, but this really is built-in to the business plan nowadays.
@@Prophes0r no need to bring politics into this
@@cities_aviv A. It's economics.
B. Politics is in everything. It is inescapable, and the desire to escape it is part of the problem.
Cost is its own spec. If you want higher performance for your use case, be prepared to pay more. If the performance of the cheap part is acceptable, you can pay less for something that satisfies you.
ETA: And no, it's not "capitalism," it's scarcity. When resources are limited, it doesn't make sense to spend them on performance you don't need. You need Star Trek replicators to escape this, not a replacement political economy.
@@jdotoz Managing scarcity is the PRIMARY job of any economic system. It is WHY we have economies.
(Ultra)Capitalism is a BAD economic system because the way we are using it now is tuned almost exclusively to "make money number go up" instead of ACTUALLY managing resource scarcity and distribution.
So yes, the problem IS capitalism, because the SYSTEM punishes you for making things well if it costs less to make them worse.
None of the economic systems we use now can handle short-term products that break prematurely.
They can't self-regulate when there is too much turbulence.
It REALLY grinds my gears when a product gets ruined by bean counter measures. "Oh we saved a few cents on this device!" "But its complete crap now?" "Who cares? We get money!"
That's typical Chinese mentality...
@@haydenc2742 Not exclusive to chinese stuff.
This is consumers mentality. People don't care. People want low prices.
"Who cares? We get money!"
Welcome to Ultra Capitalism...
@@Prophes0r Applies to communists as well. Who cares, we all get paid the same amount of money!
Man, it's crazy how much you can accomplish pretty cost-effectively if you understand electronics. It's something I'm trying to learn, but it's not easy for me to understand electricity and how everything relates to each other. Thanks for the videos, they're helpful in slowly getting more understanding of things.
This is true for anything, from auto mechanics, plumbers, carpenters, yada yada. Outside of basic volts, amps, Ohm's Law, etc, electronics are going to require a fairly deep amount of study/work before you're going to be able to figure much out not too mention fixing a borked design. I've decided to spend the extra 25% or so for something thats generally been shown to be good vs trying to eke out the cheapest component/parts. If I'm trying to cobble something together, the last thing I want to do is be wondering if some pre-made component is even doing what I bought it too do.
Don't worry about some of these things. The higher the frequency the more electronics starts to look like black magic. That's even to people who have degrees in it or similar fields.
@@arthurmoore9488lmao, for real. I work in automation and the amount of troubleshooting I have to do with highly sensitive equipment near robots, or other high frequency loads is aggravating. Sometimes the shield isn’t grounded well enough, sometimes the shield is too small, sometimes it’s grounded too well and creates a ground loop that effectively acts like an antenna that picks up RF noise. Part of me loves it, part of me hates it. Haha
Man, I've been an electronics hobbyist since the late 1980's, and I'm always surprised at how little I know. Just keep tinkering. And when you get a chance to use an oscilloscope, seize the opportunity
You answered it yourself- to save 50c
Honestly, if it's a $0.50 component, that's why they didn't include it. They're shaving pennies off of the price of these things and producing them in huge numbers with tiny profits to stay in business. A device that is 90% effective is good enough from that viewpoint.
The tip at 9:56 about measuring temperatures to determine if reactive components have appropriate parameters was really helpful. That point alone made this video valuable.
i love people who owned up their mistakes and correct them!
That ringing on the switch node of the inductor can be suppressed with a snubber network and is easily missed, this I know from first hand experience! Also sometimes adding a beefy cap can make things worse, Mach1Designs did an EMC video on the caps and is very helpful in locating noise and killing it.
You are the example of someone who not only passes his exams but also completely understands what he has studied. Good work 👏👏👏
I never knew to check temperature to check out components in a design to see if good or bad. Only to look for a broken component(s) when fixing. Maybe they stole an incomplete design from someone else? Lazy? Possibly just stupid? 🎉
I am still learning. I bought an old radio/cassette player and a radio clock alarm that both work to study on. I will use your temperature trick to maybe improve them. Thank you for giving us a great channel with interesting projects.
Thanks for the feedback :-) And you are welcome 🙂 I love to make this show.
Your devices were probably made in the 90s or earlier, so I wouldn't count on them having the same trashy engineering as modern devices from AliExpress. Good luck with your endeavors though.
High frequencies tend to cause caps & mosfets to heat, for different reasons, but the solution is pretty much the same: lower the resistance. Low ESR caps have much lower resistance, and 'bigger' mosfets have lower RDSon, which is essentially the resistance that causes most of their heat.
It's always funny to see a low-tech troubleshooting technique on high frequency IC's. Elegance in it's simplicity.
Seconded. The diode temp probe is a precise method of spotting an overheating device. Slower but many times better than a finger touch. I'm impressed by your thoroughness.
The electronic load might also introduce some noise, would be better to measure with a pure resistive load.
Yes this. Might be noisy as hell.
I used a pile of old 1ohm 50W resistors to make a grid on an aluminum plate. You can jumper quite a few different loads and can dissipate lots of heat.
How long until the video where he adds a component to the electronic load?
For the UPS I simply replaced the 2x220 uF capacitors with a 22 uF tantalum low ESR and a 470uF electrolytic as recommended by the chip data sheet. That cleaned up the noise. I used 1x probe and attenuation.
Can i use 2x 4700uF 16v on the output instead? I only have these at hand or the 10v 1000uF ones. 😅
My guess for why they didn't go the slightly more expensive route is sourcing. This is all guesswork mind you, but if they produce in house, it's only costing them pennies per. One thing I've noticed about Ali is many vendors offer the same product at different prices, so that would explain why they want to keep the price low...sort of a lowest bidder thing. Lastly, the 50 cent price you quoted is the retail price, so it would slightly increase their profit margin as they'd be paying wholesale. 😊
For a lot of stuff on Ali/AE, there are one or two companies making gajillions of these as OEM-Contract. Some resellers want different color silkscreen, some spec differrent values for some components to get a lower BOM, etc. Chinese business looks at cost cutting as the primary goal bar none. Doesn't matter if it impact the specs of these design, unless/until they are called on it and IF they can be forced to pay some sort of remuneration. It is always worth it to keep value engineering a design/product which leads to higher profit, until/unless they are forced to remunerate. There is not a lot of care/f*cks given for 'Reputation' as you see from Japanese, EU, Noram companies.
Talk to some engineers from China, and they will all admit it. They don't like seeing their work cut down to out of spec anymore than anyone else.
9:47 your forgetting that at the volume they're producing these at the increase in BOM cost is even less at under 0.19 EUR / piece on orders of 1000. Probably also a slight price decrease at 10k and 25k as well. Therefore at volume this is likely no more than 0.15 - 0.17 EUR in parts, and the lower quality parts they're using probably cost 0.08-0.11 EUR, so we're talking about them pinching literal pennies to not have a better product.
True :-) That would make this story even sadder though.
@@greatscottlab yeah unfortunately so. Though I know BigClive in his teardowns of electronics like this from china on aliexpress or otherwise mentions that they tend to have resistors and capacitors overdriven for their intended purpose for whatever reason.
Like literally around 0.08€ savings. That's really sad. Someone should contact the manufacturer and ask them to reconsider their decision.
@@pappaflammyboi5799 Dont forget about what the product itself cost spending 0,08€ on a 1€ board is a huge increase in price.
For most stuff you dont need that clean power, so it is good practice to not build them in.
@@jonasstahl9826 8% is a huge increase in price? Ya, not really.
One major contributing factor to noisy outputs on many cheap/Asian Aliexpress-alike regulators is poor layout. Take e.g. 8:50 - they use polygon pours to reduce impedance (good) and provide more thermal mass/lower thermal resistance (good), but all the components are still awfully, widely spaced, leaving still considerable inductances, which is especially bad on the switch node, as it has high current ripple. As V = L * di/dt and your L (trace inductance) is high, your di is high and your dt is small, this creates ripple noise. And as layout is something inherent to your PCB (it IS your PCB basically), you cannot easily change the noise created from bad layout.
Even the best components can deliver embarrasing performance if you do not do the layout right. Remeber, electronics is not so much about charges moving, but about controlling and containing your fields. But there are far more talented people out there, which can explain that in much better detail and accuracy.
A lot of noise I deal with in electronics is 60 hertz noise because all the power supplies use a iron core transformer instead of a ferrite core transformer and the reason why that makes a difference is because all iron core transformers run at 50 or 60 hertz well ferrite transformers run at higher frequencies. And 60 hertz noise is very noticeable if you're using inductors and highly sensitive electronics
That's a wonderful example of what a difference good capacitors make. I'm amazed that the short probe ground wire makes such a difference. Have you tried measuring at 20MHz with the ground wire?
It might be partially because it was so close to the unshielded switching inductor.
I too have burned a few MT3608's...they do seem to be unreliable.
At 11:09 you say add a 2.2 microfarad ceramic capacitor but the links in the description are for different ones so I am unsure which is correct.
I think it is because of cost.
Like the MT3608 boost converter, here in Indonesia it listed around 5200 to 6500 IDR (0.32€ to 0.37€). By adding the better ceramic capacitor priced 0.5€, imagine how much it affecting the price.
For me, the initial design is more than enough, reminding the price is dirt cheap. Unless for specific reason I need to modify, like the one you tell about creating negative voltage, I'm set my expectation according to the price I paid 😅
Well, a 20% 22uF MLCC costs 500 IDR (0.03 EUR) on Indonesia's Tokoijo, and the manufacturer would have been able to source it for cheaper. So it's not an excuse.
One of the 1st things our teachers told us at LTI is the difference between an expensive VCR and a cheap VCR is a 5 cent component. Like they will use a 1/4 watt resistor to dissipate a 1/4 watt of power rather than using a half watt resistor. Everything today is built down to a price, its the race to the bottom which got us lots of cheap junk.
To reduce heat dissipation of lineal regulators consider using an LDO that has only 0.1 to 0.5V drop, and use a lower input voltage :)
The side ground pin in the oscilloscope probe is a very nice solution. A spring loaded measuring pin would still improve measuring a shivering hand (a so called test pin "needle" for production beds).
Thanks for the feedback :-)
I always see keysight sponsoring many electronics chanbels but it would be nice if you could make a video how it looks inside one of their mesuring devices (maybe you could get a broken one or something from keysight from their rma or something like that )
50 cents can be a lot on a large scale production. Aluminum caps are easily replaceble and do less damage the board than the ceramic ones when they fail.
You have quite an assortment of nice tools, that and the knowledge to use them properly makes all the difference
i liked how you took the images of the components from the circuit and put them in a sketch
Thanks :-)
I love that hot plate for component removal at the end.
I have some experience in real market production and can take an educated guess to why such things occur:
Case A - The Manufacturer is not aware of the issue due to one of the following:
1 - prototype worked well but somewhere in production someone deviated from the design due to the lack of some component
2 - a partner provided a batch of bad components
3 - lazy (minimum testing) or non-existent QC
4 - following (copying) a design blindly not knowing how slight changes might affect the result
5 - Damaged components due to bad quality / bad storage
Case B - The Manufacturer is aware of the issue:
1 - Greed
2 - discovering the issue in late stages of production (happens a lot) so they lower the price to not throw away the stock and to avoid fixing the issue which is sometimes very hard to near impossible / costly
3 - they just don't care
4 - indented for uses with high tolerance to error
This was a very interesting and also useful video!
I found especially interesting the bandwidth limit on the oscilloscope. Can you make a video to explain why it's limited during official measurements, the implications about having strong peaks beyond that bandwidth limit and how to absorb them if so desired (I know this part: ferrites rings/beads around the wire).
Thanks!
At high frequencies such as in these switching power supplies, standard electrolytic capacitors have too high of an “equivalent series resistance” (ESR) and should not be used in this type of application; that ESR leads to excessive heating. Low ESR electrolytic capacitors are designed for these applications.
The small ceramic capacitor heating up is a bit surprising since they are intended for use with high frequencies; perhaps that was a cheaply made part, and that is why it showed a higher than expected temperature.
Or the ceramic one had to do all the work and was more or less "overloaded"
Oftentimes, these boards are based around one key component, i.e. the regulator IC. If you were to reverse engineer the board and compare the resulting schematics to the IC's reference design, you'll find probably little to no differences. And very often, the designers doesn't really pay much attention to the textual parts of the datasheet that explains the values and different usage cases, or just use one calculation for all their product runs (i.e. the same capacitors for fixed 3.3V, 5V, 12V and variable variants of the board). Also, the reason they won't add it is cost and availability. They already most likely have tens of thousands of cheap small mlcc caps around, adding one to the output won't change the BOM cost by much, but also doesn't have to affect the output in the desirable way, or even hurt it.
Thank you for showing us this
Thank you Scott I have that exact same boost converter now I have to just add that capacitor in parallel to my output terminal. Love your videos 😍😍
In my opinion a part of noise came from the switching elettronic load... Try to add a capacitor on the input of load.
I bought the UPS circuit in 5V and 12V versions, but I've received 9V instead of 12V, so from this video I've easily recognised the resistors to swap for the voltage divider, but buying the exact components is not feasible due to the shipping costs, so I went to the harvest route. The problem is that my multimeter is not very good at measuring resistance, so I harvested some similar resistor from another circuit that had similar resistance without really knowing the exact number. Well the good part is that I managed to obtain 7V, 10V, 13,5V and 20V and left it at 13,5 that is the most similar (also should allow more power and staying close to 12V during high loads). It will be useful to know the exact resistance, maybe you could pick it up again and make a couple of final upgrades: a potentiometer to regulate the voltage and an interruptor to disconnect the batteries when not needed. Maybe design a cool box 🙃. Having a clean output is not always critical, especially if I know such regulation is done at the receiving end.
Just received mine, a 5V and a 12V. Will be using 1 for a solar project. Just be careful with the voltages! Capacitors seem rated at 16V max and I would stay at least 10 to 20% below that. Checked my 12V version and that smd resistor R7 has a marking 9092 which is 90.9 KOhm or 90.900 Ohm. Also the board has been changed as his has R7 and R7* . The R7* has turned into R12.
I would always opt for a clean output. Spikes can produce noise on radios if you use longer wires and in some cases cause strange effects. I always also place a 100nF cap on the input and output.
Actually ended up using one of the smaller ones for an amp, and had added a larger capacitor to the output to support the bass. good to know it's probably helping with noise as well.
I actually got the UPS from the last finding hidden gems video and it works really well
Glad to hear it :-)
@@greatscottlab I did too. Can the same value caps be used when the output is 12V?
I also want to know. I bought 2 of each, 5V, 9V and 12V, for alarm/cameras and just in case. One seller has it for 4$ each and there was also 1$ from 5$ coupon for playing (and losing) in Go go match game, but this promo ended. So they were really cheap.
@@oyvindrepvik I think so
A lot of people don't understand ceramic capacitor dielectrics. Y5V ceramic capacitors are cheap but can lose 90%+ of their capacitance at elevated temperatures and also generally have poor ESR/ESL. X5R and X7R etc are much better but cost more. I never bother with Y5V and stick with X5R/X7R/X8R/C0G depending on the application.
Standard aluminium electros usually also are only effective up to maybe 200kHz. Above that you really need to use polymer and/or tantalum caps or paralleled ceramics.
Replacing the output capacitor with a low ESR one reduces noise, but also the power supply ripple. Often that ripple is needed for stability of the voltage regulator. I would always recommend to stick with replacements to similar components and not deviate too much from the original implementation, unless one can measure the stability afterwards. Bad stability is often much worse than noise.
BTW, good on ya, for showing the error in probing technique! It is *such* a common issue that it's kind of like "is the power plugged in" question for oscilloscope AEs !
Great walk-through of the process as always
Thanks for sharing your experience and solutions with all of us 👍 😀
Well... I've got some of those mini boost converters and i already knew that something wasn't right. Only a small SMD capacitor on the output. On my projects when i used those boost converter modules, i always added some extra capacitors on the output (Both electrolythic and MLCC).
DC-DC converters are INCREDIBLY useful in basically everything.
I wanted to ask if you can make a video where you analize a 'special kind' of DC-DC converter, the one used in ALL car audio amplifiers: A single rail to double supply rail DC-DC converter. I could not find any appropriate schematics anywhere. This DC-DC converter is really useful but i just cant find a way to make a proper DIY version. A lot of people is probably searching a way to make this DIY... I think that a video about this topic will explode on views! (And help a lot of people too).
A DIY version is really versatile, because you can have full control over both the output voltage and the power that the circuit can handle. It would also be a nice learning experience.
With this circuit, a battery can run ANY audio amplifier!!
Respect for owning up to your mistake and educating others! Thank you!
I ordered the parts from Mouser, so now the two UPS devices going to get the upgrade. I even have 5 of those boost converters to that I will do the same with.
Thanks for this great fix video. Great stuff as always. Now I just need to find a cheap-ass oscilloscope, so I can test other electronics I have lying around.
Great Scott makes a mistake: false noise signals
I make a mistake: *MAGIC SMOKE ESCAPES*
😂
Yeah I've been trying to measure a gas sensor, and I've found that every USB power supply I plug my NodeMCU board into produces a different result on the gas sensor. An Apple iPad charger makes the analog measurement noisy. A Samsung charger makes the measurement get cut in half. It really makes a huge difference in my circuit.
Great video! It's amazing to me how often a single capacitor can fix big issues in electronics 😁
Two 100K ohm resistors, a 2.5 V reference diode, and an OpAmp and you can also get a very stable 5V with current up to the opamp’s output current rating. Also, thank you for admitting a mistake, many wouldn’t be willing to say that, very cool!
Great video again! 👍
When dealing with digital circuits, I rarely find myself having to deal with noise problems. Noise only becomes a factor when I'm working with analogue circuits. I often use a linear voltage regulator in combination with a low pass filter(LC or RC with low-value resistor) and of course decent decoupling for every chip. And it rarely stops there, one must consider whether circuit isolation is necessary to prevent current loops, as well as determine if guard traces should be implemented(for example an opamp circuit with high impedance load), etc.
I bought some of the small MT3608 boost converters
and i also measured some pretty bad noise at higher output currents. Adding a 1,5uF ceramic capacitor and the problem is fixed. The ceramic caps I used literally cost 6 cents per piece
Technically, they didn't need to raise the price by 50 cents, the other 2 capacitors won't be needed anymore therefore the cost would cover these new 2.
The reality is that this board was designed by someone that either didn't take noise into consideration by accident or didn't care about it at all.
@GreatScott! In your video you say you used a 2.2 uF capacitor but in the description you linked a 22 uF capacitor.
Which one is correct?
I say 22uf because his mlcc's are really big in physical size.
0:01 VTEC just kicked in, yo!
100% about cost. Sure the part may only be 3 cents for them at bulk rates, but they already have the tooling setup for these PCBs and the components on them, and they're making truck loads of them per day, those 3 pennies per unit add up.
This man is getting so powerful he will eventually turn a nail clipper into a gun
What?
ayo Heisenberg send a pack of blue stuff
🤣🤣🤣🤣 true.
First hackSmith, then the boys, now great Scott I see you literally everywhere 😂
Aren't nail clippers already declared as weapons of cruelty?
There is a reason why I always use a high quality 100nF capacitors over Voltage rails and most of the times in parallel over bigger caps. You can clearly see a lot of garbage, aka noise, when not doing so. For reducing RF that these switched things can cause also use ferrit to stop interference. Some of these Chinese pcb's are really well designed and then you order another one and measurements are different. Bad capacitors, change in pcb layout to make it cheaper. A bit of a lottery.
That last pcb, released some smoke on those. They will go into self destruct like you said, but I still like them a lot.
'If your PCB design is garbage' - had me in stitches, hope the sellers of that circuit see this video and take note!
I was confused when you mentioned the very high pp noise value in the prev video cz I have this power supply and it works according to specs and the manufacturer is very keen on improving their products.
I also assumed noise pickup. Assumed adding a very small resistive load would fix it.
Thanks for revisiting the test!
There are plenty of off-the-shelf USB Battery Banks that can replicate the 5Vdc "UPS" function. They'll output 5Vdc even when charging, or not charging. You don't need to go out of your way to buy one; you probably already have several.
They might just do that now. We are constantly reinventing the wheel and working towards making something as efficient as possible. Thanks for the videos you put out man it has significantly added to my progression of circuit design.
You can probably make a fair bit of money by designing neat little circuits like these that actually work right, and sell them on your store. I'd happily pay reasonable markup for something I know was designed properly, characterized, and tested.
I was struggling so hard to find the right keywords to google this exact problem… THANK YOU❤
You have my respect that you publicly admit a mistake👍 true engener
I always try not to use the GND spring, it is a bit of a pain. But yeah it makes a big difference for noise and "high" frequency signals
I used to do component level troubleshooting for a manufacturer of cable tv monitoring equipment. Here is the answer to the question of why don't they use better components, at least for them. Their engineering team used 1% tolerance parts when designing and prototyping. They would hand off the list of components needed never specifying what tolerance (if any) of parts should be used.
Naturally, procurement bought the cheapest stuff they could find. I practically begged management to let me talk to the engineering team. They finally brought one guy down. I sat him down, showed him the log I had been keeping, and the exact parts that needed tighter tolerances. In this case it turned out to be the inductors in a frequency generation circuit.
The board pass/fail rate jumped to over 70% (it was in the upper 50's). My big reward for figuring that out was I got to be the end of the line guy. If I couldn't fix it, it went into the trash. I didn't last long there
Very clearly presented information, even I have understood this with my limited knowledge. Love electronic and what can be done with it. Thanks so much.
My guess for why a manufacturer wouldn't include the fixing product is just the cost of the better capacitors most likely being higher than the ones they used, and most of their customers probably don't check the outputs noise, but honestly I couldn't tell you if that is truly the reason.
06:22 Additional to further reduce output noise, add capacitor parallel to R1. Start with say 0.1uF. This will inject output noise in feedback path & can help reduce output noise further.
that makes for a PID controller, you are injecting a part of the noise in the PWM regulation which could create instability... If the dynamic regulation is OK, you should not touch it but improve output filtering IMHO
@@adrienletellier6132 Yes that is why capacitor value should be specific based on output noise.
bois, i m new here with this type of technology and soldering etc, i'll buy soldering iron from aliexpress costs 75 SARS, but im not sure if its working perfectly for me but i'l buy it and testing and Learn how to solder like a pro!, thank you great scott for a knowloge of things like this and ofc it will take me a long run to a level pros
Thank you for this video, especially the info on how to check for noise on the DC supply using an oscilloscope.
They don't do it because they don't need to -- 99%+ people will not be analyzing the signal, and a sloppy 5V is fine enough for the things people buy these for; either the device isn't sensitive, or has its own built-in regulation.
Nonetheless, good info. Next time I need to use a cheapo boost/buck converter, I'll check the signal and maybe throw a post-regulator on there.
8:14 i love how you put the images of components next to the schematic components -- for those of us trying to take self learning electronics seriously, it massively helps our brains put to mind the components as if they were legos (but the various values of each 'lego' is like the lego color). Next after looking at this for a while as a reference during test building on our own, the schematic components begin to fit like a catalog in our brains before we even realize it. I hope i not sounding too nuts but this really works for my brain very exceleratredly, ty for this idea and i hope you and others do that much more often -- even if just once before you do the red pen work on the circuit would be a big help ^_^ p.s. i have loved your work for last many years, ty for keep doing your channel ! 😀
Awesome! Ordered the power supply. Thank you so much for testing 👌
You're welcome ;-)
That's what they say, if your circuit misbehaves, put capacitors
Haha true
I'd guess that they don't include those capacitors because they're aiming for minimal viable product, minimum cost, and maximum profit. If they add $1 of parts and need to bump the price $1, a competitor might not do that and sell more because it's cheaper... Especially since these probably sell like hotcakes and are acceptable for most things people use them for like lights or things with their own regulation/filtering
Mostly anything where you don't want noise (not audio/frequently changing output) you usually just have to add a capacitor to level the voltage
thanks for all elctronic helping video and contents,good luck
So you achieved that by replacing 220uf electrolytic with 22uf multilayer ceramic SMD cap(huge one). But ceramic capacitors are capable of extreme current spikes, much larger then electrolytic ones, though since it is used in place of 220uf electrolytic total capacitance load may not be as bad
Tantalum / MLCC Caps seems to solve most noise problem which is why a lot of audiophiles just spam them a lot around their board like Onkyo or Denon when they designed their board.
This is not a complaint about this fantastic video on what can be done to achieve better results from inexpensive switching power supplies, just a reminder that buying what appears to be inexpensive component parts can be very expensive when you come to pay for them to be shipped to you!
50c for a component is a great fix, but let us remember that no component supplier is going to ship you a 50c component for free, most suppliers have a minimum order price + shipping, which can make small projects VERY expensive to build.
This is even more frustrating here in the UK where buying let us say GBP 10.00 in components can result in you paying GBP 10.00 in shipping, so when you want a GBP 0.50 regulator you have to either buy 100 off of them or pay the minimum order charge plus their shipping.
Sorry for the whine, but a lot of TH-cam videos on building small electronic items forget these hidden costs.
Thanks Jeremy❤🎉❤
Need a video on how to design a LPF for a switching converter.
Well done. Fix the problem rather than ridiculing it and recommending something that costs 5 times the price... or worse not recommending an alternative at all!
Hi Scott, I was wondering if you could do a video to show how to connect a VFD with a single phase capacitor start motor. Thank you
Congratulations from Belgium and thank you for sharing.
Thank you very much for the video. How could you make the UPS not charge the battery at such a high voltage? For example, charge it only at 4.1v to increase battery life. Thank you so much
Nice, I have a project where I need this ups, thanks a lot, I wasn't looking for this at this exact moment but it just what I needed
Having been exposed in the power supply side of things for so long, the main failure had always been due to some cheap capacitor. And this applies to even big brands.
It's very simple, cost! Just because it costs 50c more on a component level, it will raise the final sale price by $5 or more.
Also, don't forget products are designed to fail after warranty!
Thanks, I use those and don't have the knowledge or time to dive into this like you do. That was extremely helpful, since the fix doesn't require more than basic skills :)
Was just getting ready to send some Gerber files to the PCB place, now I have to make a few capacitor changes first. AAAARGH!...more work but I'm thankful for the knowledge!
I get to Deutschland again I owe you a bier!...Prost my friend!
Ah… the good old “didn’t check the SRF of the caps” problem
you are the smartest man that have EVER EXISTED!!!!!!!
Oja, mit dem MT3608-Board (das letzte vorgestellte) habe ich auch so meine Erfahrungen gesammelt. Mein Use-Case: Reste eines 24V-LED-Bandes an einer etwas größeren 12V-LiFePO4-Batterie betreiben. Im Test an schwächeren Batterien und mit Netzteil kein Problem, am großen Akku ist mir eins nach dem anderen abgeraucht. Nach etwas Messerei mit dem Oszilloskop: Der Akku kann aus Sicht des kleinen Schaltkreises quasi unendlich viel Strom liefern, so dass der Ausgangskondensator beim Einschalten einen starken Impuls durch die Induktivität erzeugt hat. Dabei kam es, nachdem der Kondensator voll war und der Stromfluss abgebrochen ist, zu einer Spannungsspitze von über 40V. Eventuell hat da der DC-Bias der Billigstkondensatoren auch eine Rolle gespielt. Hübsch zu viel für den bis zu 30V spezifizierten Chip. Lösung: ein klassischer Elko parallel, außerdem den Regler am Ende des nicht ganz so dicken Anschlusskabels platziert statt am Anfang. Jetzt funktionierts...
paused near the start when you showed the noise in the circuits, I assume an LDO or just some resistor cap additions would smooth it out.
I see a lot of what I call "hopeful" or "expectant" engineering especially in the cheap but plentiful switching power supplies online. I expect noise value to be absurd, and often wind up modifying supplies to remove noise. Generally, I try to design my stuff as though the cleanest DC power it will see is half-wave rectified AC. I'm an electrician and hobbyist. Cool discussion of the incredibly low cost of doing it right, especially the first time.
Have you tried asking the Chinese seller why they don't use the other capacitor? In my experience the Chinese have quite good customer support. The answer may be difficult to understand due to machine translation, but they usually do answer. Your request may even prompt the Chinese to change their PCB to use the other capacitor in the future.
I would imagine the capacitor on the MT3608 boost converter is not included as it would reduce efficiency slightly and maybe in most cases isn't functionally needed? I could be wrong.
alternative way to reduce noise are to include ac filter in the output