Oh goddammit...

[c4757p]

Well, since I am pretty much a complete magnetics novice, it was to me.

[c4757p]

Well, I think I know why some worked and some didn't. I took one that wasn't working properly (not this one - it was 1mH), unwound it, cracked it into four pieces, glued them back together and rewound it. The results: 1mH, saturated at 150mA (which makes sense considering that the 200uH one saturated around 800mA). After modification, the inductance went down to 32uH (yep, a whole 31-fold loss), and the saturation current went up to 5.8A (39-fold increase).

I bet the "good" ones were actually slightly damaged in shipping!

I'll post pictures of the tester and scope captures over at the other thread in a few minutes.

[Gall]

I had to build a 100W 3kV flyback at one time. This was quite a challenge, I learned very much about these beasts. In a flyback the inductor has to store all the energy of one working cycle. Ferromagnetic materials like ferrite act like a conductor for magnetic field, they cannot store much. Most of the energy is stored in air gaps of the core. That's like an electric wire that just carries electric current without storing energy, while a piece of insulator across the wire (a capacitor) does. Unlike a wire, ferrite core still may hold a small amount of energy but usually not enough for high loads. That's why it's easy to saturate it. A small air gap is usually not needed in buck converters, while most boosts and flybacks cannot go without it. Most ferrite cores made of two parts have a "natural" gap that is enough for most purposes. High-power ferrites are made with an artifical gap. One can add a gap just by placing some thin paper between the parts of the core.

Using a big ferrite core, it is possible to light (and even overload) a 240V 40W light bulb from just three turns of wire. The primary winding was pumped from 12V with IRF840. I did that in one of my experiments. I used an improper diode for secondary rectifier so it got really hot. (I decided to save time and sacrifice an "incorrect" diode instead of spending time finding the "correct" one).

[Gall]

Well, I think I know why some worked and some didn't. I took one that wasn't working properly (not this one - it was 1mH), unwound it, cracked it into four pieces, glued them back together and rewound it. The results: 1mH, saturated at 150mA (which makes sense considering that the 200uH one saturated around 800mA). After modification, the inductance went down to 32uH (yep, a whole 31-fold loss), and the saturation current went up to 5.8A (39-fold increase).

I bet the "good" ones were actually slightly damaged in shipping!

I'll post pictures of the tester and scope captures over at the other thread in a few minutes.

Yes, this IS the air gap!

[BravoV]

Glad to hear it works as expected. 

Whatever it is, I'm sick of Mr. 34063

So no more Mr. 34063 hatred ? 
C'mon c4757p, you are the authority for this MC-34063 chip here in this forum, please don't hate & abandon it. 

[c4757p]

MC34063 can screw itself and die in a fire.

OK, OK, it does have one advantage: it's just about the cheapest damn DC-DC regulator you can find. It's not particularly good, but it is cheap. I will definitely continue to use it, but nowhere near a kit.

(Yeah, it was my fault it blew up, but that was established from the beginning! LM2575 does not become a DC-to-projectile converter when the inductor saturates.)

[wraper]

Saying that MC34063 is shit is at least unfair. You blow MC34063 because left it without current protection by shorting the current sense resistor. LM2575 have internal current limiting so of course you couldn't blow it because it was impossible to disable current limiting. Therefore this is not telling anything about reliability.

[c4757p]

LM2575 have internal current limiting so of course you couldn't blow it because it was impossible to disable current limiting.

That's my point. I'm kind of kidding about it being shit - as BravoV remembers, I was quite a fan of it. I don't like the idea of using it in a kit though - all it takes is a simple short to make it put your eyes out. The LM2575 has inherent protection against many things that is impossible to disable. I tried pretty hard to blow one up - the only way I finally managed to wreck it was to reverse polarize VCC, and it still didn't do anything too dramatic, even with the PSU current limit at 5A. (MC34063's tantrum happened so fast it could have done it on the energy in the decoupling cap.) Does that make it superior? Not necessarily. But it does put it higher on my list of chips to use with beginners.

[wraper]

Did you try the same stress test on MC34063 without omitting current sense resistor to be fair? As the main difference is that in LM2575 this resistor is inside the IC. Well you could include two extra MC34063 to the kit and it would still be cheaper than LM2575 

[c4757p]

It would take it just fine without shorting the sense resistor, but that's the whole point! Properly implemented, it'll stand up to just about anything, but properly implemented, it wouldn't have to! For use in the kit I want something that can be completely screwed up and still reasonably safe.

I would just include two, but the problem is how violently it dies. That generally seems to be a problem with power devices in DIP format, they become guns when abused. The cost savings aren't really worth it for this specific application. It takes a lot of power to make pieces come flying off a TO-220 or DPAK.

[c4757p]

OK, I just had to take a screenshot. F*king dyslexia...

[wraper]

If LM2575 output transistor gets short I'm pretty sure it would fire the same bullet or at least emit magic smoke from the hole. That's just how power devices die with unlimited input current.

[c4757p]

The current isn't unlimited, though. The 34063 blew up almost instantaneously, even with a relatively weak power supply it could have done it just by eating the contents of all the power supply caps. I did short the LM2575 output transistor directly to ground for quite some time, with the power supply set to a limit of 5A. It didn't blow up, it overheated slowly enough for the overtemp cutoff to kick on. This thing is likely going to be powered by something much weaker than 5A.

[wraper]

Not exactly LM2575 but i have replaced LM2577 which was dead the same way as this MC34063. With big hole in the middle. Psu was sure <3A on 25v rail where it sat.

[c4757p]

Oh, I'm sure it's possible to catastrophically kill them. It's just more difficult/unlikely. There are a lot of ways to make electronics dangerous, I'm not going for "completely covered in bubblewrap". Just a bit safer for really not much more cost.

[wraper]

Regarding to beginners, reverse polarized electrolytics will make more sound effects and fly further  . One friend of mine, being beginner managed to catch blown cap into his ear.

[Bored@Work]

all it takes is a simple short to make it put your eyes out.

Do not run with sharp objects.

[madires]

Regarding to beginners, reverse polarized electrolytics will make more sound effects and fly further  . One friend of mine, being beginner managed to catch blown cap into his ear.

Always wear ear protection at the shooting range :-) I've managed to make a LM2576 "pop" by reversed voltage some while ago.

[Flávio V]

I destroyed a LM2576 by shorting the input capacitors(2 3300uF 35V) multiple times, i still can't figure out why it happened.

[c4757p]

The input capacitors? Because you reverse polarized the output transistor - the output capacitors were charged up, and the input was now at zero. This is the same reason why you're supposed to put protection diodes around, for instance, the LM317. You really ought to put a protection diode across any step-down regulator, as well as diodes antiparallel to all the rails in any system with positive and negative supplies. This prevents polarity inversion - if two are shorted together, one regulator will "win" and the negative rail could become positive or the positive rail negative.

Reverse polarity is the best way to kill even the most robust chips.

[dr_p]

I destroyed a LM2576 by shorting the input capacitors(2 3300uF 35V) multiple times, i still can't figure out why it happened.

but... why??