Oh goddammit...

[ElectroIrradiator]

If the toroids are from a cheap inexpensive source from the far east, then the seller may have mixed (less expensive) cores with lower permeability into the bunch, after having given the odd ones a new coat of paint.

Try measuring the inductance of a few of the bad ones, and see if they all measure the same for identical number of turns. If they do, then they are fakes/repainted. If they are all over the place, then they are just lousy junk.

[codeboy2k]

These are almost certainly all ferrite. I am not sure what can go wrong in the production of a ferrite core to reduce its saturation flux to about one tenth - I know just over "nothing" about magnetics - but they all appear to be the same type. Some work, some don't. I may break open one of the bad ones to see if there's something obviously amiss.

Maybe if you cut it open you'll find it's a cheap chinese made toroid with a plastic core, and just a few mm thick layer of ferrite on the outside 

Wouldn't that be a shocker 

edit: ElectroIrradiator beat me to the same thought by a second lol

[dr.diesel]

Maybe if you cut it open you'll find it's a cheap chinese made toroid with a plastic core, and just a few mm thick layer of ferrite on the outside 

 

Seriously, has anyone found plastic core knock-offs floating around?  I sure hope not, but wouldn't surprise me   

[ElectroIrradiator]

To the implied question: Yes, it is possible to have two cores of similar dimensions having identical permeability yet different saturation flux, if that is what you are measuring. Same conclusion: Fakes/production rejects/cheap materials.

[madires]

Ferrite cores allow high inductances with a few turns of wire but saturate quite fast. Ironpowder cores are the reverse and the built in distributed air gap helps to store more energy. So ironpowder toroids are the save bet.

[westfw]

I'm no inductor expert, but the coil in your initial picture has very few turns of wire compared to any 34063 circuit I've ever seen (or compared to your working version, for example.)

How is saturation for cores specified, anyway?  It's a magnetic flux characteristic, so surely "6A" would be a value for the finished inductor, rather than the core itself.  You have "about 7" turns, and saturated at current about 7x lower than you expected; are you sure you don't need an I*n term in the flux calculation?

[c4757p]

Yeah, "6A" was just from experience with similar inductors made on those cores. I tested them somewhere around 300uH/ten turns when I bought them. These appear to be very high permeability; those seven turns give about 200uH. I don't have a specification on the saturation (these were just sold basically as junk from some electronics shop) but it would be specified as a flux density in gauss or teslas.

[madires]

Yeah, "6A" was just from experience with similar inductors made on those cores. I tested them somewhere around 300uH/ten turns when I bought them. These appear to be very high permeability; those seven turns give about 200uH. I don't have a specification on the saturation (these were just sold basically as junk from some electronics shop) but it would be specified as a flux density in gauss or teslas.

Those are ferrite cores without an air gap, great for filters and transformers but not for storing energy.

[envisionelec]

Hmm, this regulator's not working quite right. Let's see what's wrong with it.

Oh, the current limit's kicking in really early in the cycle. Well brain, what's the stupidest idea you can come up with? (Brain: "Derrrrp.... uh... short out the sense resistor and see what happens?") OK! Here I go!

Ow! Shit!

(My best guess is that I miscalculated the inductance by an order of magnitude or so. Looked like saturation, but that core doesn't saturate south of six amps. Whatever it is, I'm sick of Mr. 34063, I'm doing something else tonight!)

Too few turns for the material type, powdered iron which has high permeability, but is too lossy at those frequencies. The numbers aren't at arm's length for me anymore (not doing SMPS as much these days), but I can tell a dud attempt when I see one.   I have a similar core here that is absolutely good for six amps and is being used in a commercial SMPS. It's the wrong type toroid in the application (powdered iron). It works but is unreliable, blowing the controller up with regularity. The better material type is 77 or F. Use a pot core for easier winding. At low current, you can maintain a reasonable expectation of flux containment.

[c4757p]

Somehow I'm not surprised I was using entirely the wrong type of core. This is what happens when a magnetics novice buys his own cores...

[envisionelec]

Somehow I'm not surprised I was using entirely the wrong type of core. This is what happens when a magnetics novice buys his own cores...

It's OK. I didn't know anything when I got started either. It's why I built a Lexan blast-containment shield when I nearly lost an eye working on a 2kW DC-DC converter. Wrong material type. Check out cwsbytemark.com - or just give them a call with your requirements. They're happy to help you succeed. I can count on one hand the number of engineers that can build a reasonably well-functioning SMPS the first time.

[rsjsouza]

c4757p, check Unitrode's magnetics design handbook, as well as their training material. I read some of it a long time ago, but it was truly enlightening...

[c4757p]

c4757p, check Unitrode's magnetics design handbook, as well as their training material. I read some of it a long time ago, but it was truly enlightening...

That looks like some really good reading. Thank you! I will definitely be spending some time with this.

In the meantime, can somebody check an inductor choice or two?
Murata 33151C. This is a higher-current version of the one I'm testing with, so I suspect it would work. (The one I'm testing with only gives 40mA margin over the maximum current) I don't see any info on the core material but it lists "switching regulators" as an application.

Bourns SRR1210-151M. Listed as for "input/output of DC/DC converters" - does that mean just filtering? It looks like the ones I've seen quite frequently as the primary inductor in DC/DC converters. Most of Bourns' SMD inductors appear to list that. The core is specified as "ferrite DR/RI" - no idea what DR/RI is and Google doesn't seem to know either. I know we've established that iron powder is superior to ferrite for switching regulator applications, but there seem to be a lot of inductors intended for DC/DC that are some sort or another of ferrite.

Application: inverting buck-boost 15-24V in, -18V out @ 300mA, 1.36A peak, 52kHz frequency.

Edit: Just double-checked my numbers and I have barely enough inductance, so I changed to 150uH inductors. That loses me a bit of max current, but still leaves at least 300mA margin (and the 300mA output current has margin as well, so probably 500mA or more above the real maximum).

[c4757p]

By the way, I am 100% a convert to the church of LM2575. Screw this MC34063 shit! I just tried to blow up a sacrificial 2575 to see how robust they are and had one hell of a time doing it. Saturating inductor, shorted output, you name it, it can take it and comes right back for more. (Not only that, but I'm not even sure these are genuine...) Does anybody know of a similar (and similarly cheap) part that works in boost converters?

[tszaboo]

So your dummy load is working after all. For a second.

[BravoV]

Yeah, "6A" was just from experience with similar inductors made on those cores. I tested them somewhere around 300uH/ten turns when I bought them. These appear to be very high permeability; those seven turns give about 200uH. I don't have a specification on the saturation (these were just sold basically as junk from some electronics shop) but it would be specified as a flux density in gauss or teslas.

Not an expert though, but I do play a lot with salvaged inductors and switching ps circuits, high permeability core ferrite is usually not suitable for switcher and will saturate at low current. Other way is to break the core into 2 or 3 pieces, and glue it together using high temp epoxy to create gaps, the core saturation will increase by multiple folds.

Suggesting you to build this, using dirt cheap jelly bean parts and very simple -> Inductor saturation tester designed by Jay_Diddy_B. 

[c4757p]

I like the TPS40210. You do need an external switch but I prefer that on anything over a couple watts anyway.

Nice. I prefer an external switch as well.

Not an expert though, but I do play a lot with salvaged inductors and switching ps circuits, high permeability core ferrite is usually not suitable for switcher and will saturate at low current. Other way is to break the core into 2 or 3 pieces, and glue it together using high temp epoxy to create gaps, the core saturation will increase by multiple folds.

That is a very good idea.

Suggesting you to build this, using dirt cheap jelly bean parts and very simple -> Inductor saturation tester designed by Jay_Diddy_B. 

Oh, I haven't seen that.  I might build that this weekend.

So your dummy load is working after all. For a second.

 

[BravoV]

Suggesting you to build this, using dirt cheap jelly bean parts and very simple -> Inductor saturation tester designed by Jay_Diddy_B. 

Oh, I haven't seen that.  I might build that this weekend.

Share it at that thread once you finished building it, please.

[Phaedrus]

Designed a 20W LED driver off a TPS40210. Never built it, but it was a very good fit for my design, aside from high-ish Vref.

[c4757p]

And the award for most gratuitous use of PCB space goes to.....

(I've got a stack of these big enough that I really can't be bothered to cut them...)

[BravoV]

I can see you're merging the power lines for the driver and for the inductor. Mine is separated, so I can hook it to "adjustable" voltage bench psu for the inductor ramp up source, its quite handy especially when you encountered a crazy huge inductor with really high saturating point (cranking up the voltage) or those very tiny one (cranking down the voltage).

[c4757p]

Good point. OK, easy enough

Edit - changed the higher current diode to MBR1535CT (15A) and added an RC snubber to keep ringing to a minimum.

[BravoV]

Cool ! 

By cranking up/down the supply voltage for the inductor, its basically saves you from adjusting the pulse width and/or pulse period, hence, less adjustments at the scope side too, especially you're using analog scope, hope you get the idea. 

Just be careful not to crank up the voltage too high beyond the mosfets or diodes specification.

[c4757p]

Just built it and it works great! Thank you for suggesting that. I'll post pictures to the other thread later.

[Gall]

Hm... A ferrite inductor without an air gap? No surprise it got saturated.