Wow. What a demonstration! This sets a whole new standard for this kind of videos. You seem to show every functional detail of the circuit. I especially liked the slow mo footage of the sliding transient voltages. It goes a little over my head what I'm doing as a electronics enthousiast, but it's nice to know there is a full explanation that's easy to follow.
shouldn't there be another primary coil with a diode to reset the charge of the main primary. In this case I see only a diode for the reset. Somehow they avoided adding an extra winding.
Such a great video series! I would love it if one day you could focus on the various ways these SMPS units do feedback, showing the signals and waveforms for that. (Going back to the magnetic coupled feedback on early SMPSs)
LOL, lately I have been consuming a lot of Adrian’s content and a lot of Diode as well with some of Noel’s retro lab and Marco Reps. They are definitely my favorite RU-vidrs by far! Keep up the great content!
This really should be a million-subscriber channel. That was some of the best circuit analysis and demonstration I've ever seen, and it didn't take $20k worth of test equipment to make it happen. Pure gold!
I totally agree with you. Brings to mind a wise, old saying that goes: "It is not the tools that make the mechanic, rather it is the mechanic that makes the tools". I firmly believe that this saying is true. Fred
I always thought you had a old chunky oscilloscope, I now realize you have a classic irreplaceable tool that you operate with amazing precision and great ability to demonstrate it..Thanks
It is an awesome idea to measure the transformers with simple half a turn or a full turn of a piece of cable. I am always scared to measure with my oscilloscope directly fearing I may oversee some isolation and blow it up, using those wires it is almost impossible to blow it up. Thank you for another great video Diode!
could you plz show us how to remove or decrease 25mhz overshot spikes ? Actually in most PC power supply is 2 switch forward and 80% of them have this spikes which is not good for mainboard healthy. REGARDS
That fake slowmo footage was really good, could you have achieved a similar (maybe worse) effect, but easier by setting the trigger holdoff to something a little over 10ms, so you sample different parts of the 100hz variation in each successive capture?
a very fine example why analog still matters, i was a long time a fan of CRTs, when lcds were used all around, don't care of the space, weight and power, as long it showed me a bit more ground truth
Whenever you get your осциллоскоп out... I get sidetracked from what you're talking about because I'm trying to read the Cyrillic. I have to keep rewinding! ;)
I love the little ‘Sdyelano v SSSR’ badge! I taught myself to read the Cyrillic script (which I absolutely love) when I was about 12 years old, and that’s 42 years ago, so I hope that’s right, lol! I’m a bit rusty now. I don’t understand Russian, though. I’m guessing it means ‘Made in USSR’?
@@edgeeffect - I understand almost nothing except hello, thank you, etc. lol! I really wanted to learn as a kid, but there wasn't enough interest at my school. I would probably have struggled anyway, I couldn't even get to grips with German!
Try to measure output voltage overshots by connecting tip and ground of your scop together and put it on negative of output voltage, you can see the spikes on any place of negative with this method!!!
The "horrible 3V ripple" is almost certainly the ground lead pickup on your scope. Try a low inductance "spring-clip" ground lead and it will go away. The loop area of a regular 10:1 probe ground lead makes it not useful for any PSU ripple measurements. Also don't forget you need to have a 20MHz BW limit on scope to measure ripple in a standard way.
Regarding the high frequency ringing noise on the output signal, could that be ground bounce? What happens when you connect the probe tip to the same ground used to make the measurements shown? I bet you still have the ringing. Remove the long ground clip lead and probe clip, wrap a piece of wire around the probe tip ground to form a short ground lead. More than likely the amplitude will be much less with same low frequency ripple amplitude.
Generally the ground loops introduce noise, but in this case it seems to me that it is due to the filter coil being saturated, the diameter of the wire seems too small for the current involved, and also as regards the poor quality electrolytic capacitors to remove those high frequency rings and improve the ESR, it is better to put ceramic capacitors of a few hundred pf at the ends of each capacitor. It would be better to redesign the printout with a double-sided design, precisely to obtain a clean and low-noise output voltage. But all these design subtleties, for such a low-priced power supply, I don't think the manufacturer has ever thought about it. However my most sincere congratulations to Diode, exhaustive and detailed exposition as always!
Damn, that fake-slow-mo was awesome. That combination definitely beats a digital scope on its own for stuff like this! It shows how the duty cycle slowly increases as the capacitors discharge and then quickly goes back down at the peak of the AC input. How else can you even show the varying speed of the duty cycle changes in an intuitive way? Do any digital scopes come with a slow-mo replay mode?
My rtb2004 is not particularly fast (maybe in a keysight scope you can see it clearly), but in the history mode you can replay your captures at different speeds, so I'm pretty sure you can do something similar. If you don't have that option I guess you can set the trigger holdoff to a little more than the 100hz of the variation to get a slowly moving waveform, you essentially sample a different part of the 100hz cycle in every trigger. Another alternative is to trigger on different levels of the input capacitor voltage, by moving the trigger level you can see the changes in the waveform.
@@Gengh13 Mx Hantek DSO4072c has so little sample memory (only 40K) that there is no chance of recording anywhere near enough points to do any kind of slow-mo. Yeah maybe you could do it with tons of memory. Trigger holdoff might be an option, but you're still limited to the refresh rate of your scope and it's gonna be just as fiddly to set up as on the analog scope. Just triggering on the filter capacitors doesnt give you the time aspect of it though. You could look at all the waveforms in the cycle, but wont get an idea of how fast they change into each other. Of course you cant do any of this with just an analog scope either. You need a half-decent camera for it as well.
Hi there, can you please make a video on how to bypass safeties and protections in ATX for another use besides computers as they can be a very good power supplies to charge batteries, make 12v fridges work as well as car air compressors. But SUPERVISORS with OVP, OCP turn SMPS off and stay useless... THANKS !!!
Is this way for calculating transformer ratio will be correct in case of flyback? In ex. this orico charger from previous video : ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-jdISnlydUMk.html 320 * ( 10/138) = ~23,1V, even including voltage drop from diode it seems high and that charger doesn't have an output inductor. What I'm missing?
fantastic overview of operation , you have taken some time to get these "cro" shots ect , alot would not realize the time to do so , thanx so much for these vids ect
You measure with only "half a turn". Is that even possible? It seems to me you measure with one turn but you only turn it around half of the cores cross section area which means only half of the flux. Is that correct?
It is possible to run your big oscilloscope without the ground. Just build a ground defeating extension... It's the only case where you can use it of course, otherwise it is very dangerous
@@cekpi7 great idea. I was going to say that the power supply is 600W so the transformer would be crazy big, heavy and mostly expensive, but using it with the scope is brillant. Still you have to disconnect the ground from it but since it is powered via an isolation transformer it is not needed anymore.
@@cekpi7 There would be no gain by running the scope on an isolation transformer because its internal supply will be a regular isolated step-down transformer. Either way if you defeat its ground and hook it to the DUT's high voltage supply it will raise the oscilloscope's chassis to dangerous voltages relative to earth ground. It's one thing to do that with a small plastic-cased battery scope but I would definitely not want to do that with a large metal-cased oscilloscope because the shock risk is greater.
Hi Danik, we've seen that you've got less material to show, so I second the opinion that some theory, like snubber networks, would be great. You have a good command of electronics, so I trust you'd make an excellent example with actual parts. Good luck with your channel!
I just disconnected the earth wire on my scope #yolo Of course for measuring two transistors at the same time I still need a differential probe, but I actually use a small transformer.
10:09 That short huge overshoots are not real. I am pretty sure about that, because of I know where they come from. Look at the huge area loop created by the ground wire feeded from the middle of the probe to the GND. This and the tip part of shielding of the oscilloscope probe makes a huge one turn inductor, terminated by the input capacitance of the oscilloscope probe, the output capacitance of the DUT, etc., which in fact creates an serial resonating circuit, which produces some huge ringing somewhere in the range between higher MHz to lower hunderts of MHz. On top of that all, the area of the GND wire loop makes inductive coupling to the core of the power transformer, the second stage inductor and who knows what else, which leads to induction into it. Simply short the tip of the probe with that GND wire and put it into same location with DUT running without touching it to the DUT. You will observe a large induction, because this probe configuration behaves like a kind of H-probe. And this is in fact what happened. That great overshoot is not the real output of the DUT, this is caused by improper probing. Scope probes are sometimes a little bit tricky.
A half turn of a winding does not exist, somehow the circuit is always closed. The winding is only affectet by half of the magntic flux because of the core design. Only the inner core has the full flux, the two outer arms split the flux between them. However very informative!
Of course, such a half turn only works when not loaded much. If it was loaded more, it would force majority of the magnetic flux into the other half of the core. But you can create half turns by connecting turns on both sides in parallel.
You're The Best SMPS teacher I've ever had. You make me look on this topic from other perspective. Thank You! Have You seen prices of server power supplies? They can be bought from 100Kć and they seem to be not dodgy at all! And they also seem quite complicated. Maybe some video in future? Now Im modding one made by dell to use it to power hydraulic car lift :)
Why dont you float your 2 channel scope? Basically i have an extension cord and i removed the ground contacts on a few sockets and i plug my scope in those when measuring mains stuff. On my scope its very unsafe because the scope body is mostly metal, but with your scope its pretty safe to do this (mostly plastic).
