Measuring saturation of a toroid core

[Recyclojunk64]

I'm trying to find a suitable toroid core to wind an inductor for a buck converter I'm working on. I've chosen to use an amorphous core as it requires less windings and thus will be lighter (this is a weight-sensitive application). The core I'm thinking of using is shown below, it's dimentions are labled as 40-25-15 (for the actual core not including the plastic case).



The numbers A-8067-030  10 21670 REV3  34/10 are printed on the fibreglass spacer underneath it, though I couldn't find a datasheet of course. It had on it nine 14AWG wires with two turns each, and there were some 40A relays next to it...

What I'd like to find out is the maximum amp*turns (correct terminology?) it can tolerate before saturating. So to test it I wound 100 turns on it (990mH) and connected a shunt resistor (0.01ohms) and looked at the current when a capacitor (15000uF) charged to about 20V was discharged through it. As it's about one Henry of inductance, I would expect that the current would rise by about 1A every 0.05 seconds until it suddenly spikes up as it saturates (assuming the capacitor doesn't drain first).


But as can be seen, it instead rises by about 1A every 4us! So either my $20 LCR meter is out by a few orders of magnitude, or this core can't take more than a few amp*turns. Neither sounds very plausible, so I guess I must have overlooked something? Or made a terribly large mistake somewhere.



Here's a more macro look at that curve. Note that I forgot to correct the voltage offset on the scope. (the capacitor is still connected to the battery hence it levels off afterwards)




PS. Sorry for the photographed "screenshots" I still have to install a floppy drive on my scope

[mij59]

Looks like an iron core, what's the switching frequency of the buck converter, at what output current ?
Inductance 9.9 mH ?

The properties of the magnetic material ought to be measured with an AC signal.

[johansen]

As it is now, that core is useless for use as buck converter.
if you want to experiment with it, you might try and unwind it, and roll it back up with a strip of paper in between each layer.
I can't tell how thick that metal is, but you might need to use something as thin as a VHS tape to space the layers out.
also, i don't know how to estimate what the permeability will be if you do that.
as it is now, the permeability of that core is probably more than 5000, and it is a 1 H inductor, but only up to like, maybe 10 milliamps?
50 microjoules of energy storage i'd guess.

try and find the saturating flux density, using an ac signal with a capacitor in series so as to avoid any dc bias.

i suggest you use my choke calculator spreadsheet here http://johansense.com/bulk/spreadsheets/
use the open office version.
calculate the energy needed to be stored in the buck inductor from sheet 4, as well as the peak current.

then use the other tabs to find an appropriate size of inductor.

remember these key principals:
energy stored is proportional to flux density squared
energy wasted in copper losses is proportional to flux density squared.
energy wasted in the core is generally proportional to both frequency and flux density squared.
energy stored in the core is almost always zero, it is stored in the air gap.
copper losses are proprtional to air gap volume squared.
be reasonable with the copper fill factor. 30% is about all you can get for a machine wound toroid. 60% might be reasonable for hand winding.


if weight is real sensitive you might try aluminum wire.

[Recyclojunk64]

Looks like an iron core, what's the switching frequency of the buck converter, at what output current ?

Haven't quite decided on the switching frequency yet, but thinking somewhere between 30 to 100kHz. Output current will vary, but I'm pretty sure I won't want more than 20 to 30A out of it. Perhaps 10A normally.

As it is now, that core is useless for use as buck converter.

You're absolutely right. In fact it's only useful for one thing; a paperweight. It seems that it saturates almost immediately with even just a singe amp flowing through one turn. I gusess all the wires going through it originally must have canceled each other out somehow? I did some tests today with some other cores, and all the powdered-iron and ferrite ones seemed to work fine, but the ones that had a very large inductance for their number of turns exhibited the same ridiculously low saturation point (of which most were amorphous cores mostly from LC filters).

if you want to experiment with it, you might try and unwind it, and roll it back up with a strip of paper in between each layer.

I measured about 7 ohms between the outer and centre layer, is it likely that it already has a thin insulative layer on it? Anyway I don't think I'll be using it in anything AC so it probably isn't worth rewinding it and getting metallic glass shards all over the floor.

i suggest you use my choke calculator spreadsheet here

Thanks, I'll use that.

if weight is real sensitive you might try aluminum wire.

It's going to be mounted on my mountain-bike so weight is really just a tradeoff, I'd like to keep it lightish but in reality I really won't notice an extra hundred grams.


Anyways, I've found a core that I think is suitable. I believe it's a powdered iron core judging by the soft saturation. It measures 230uH and has about 54 turns. I think I'll stick with that number of turns, but I'm not sure if it will be worth rewinding it with thicker wire for less resistance loss. I guess I might as well I have the time.


I tested it with 40V this time and a larger capacitor bank

[johansen]

i'm guess you want a kilowatt of power?

i think you should be able to do that with a single ferrite core from a 500 watt or bigger computer power supply.
you will need to stuff the core absolutely full of copper. don't use a bobbin at all, and keep the copper smaller than 22 awg.
100 khz is a good starting point. looking at a physical air gap of about 1mm.

but for the sake of input current ripple you should use more than one phase. there are a number of off the shelf dual phase synchronous buck converter chips available. too bad most are surface mount and you'll need a professional board made for them though.

[T3sl4co1l]

There are two kinds of amorphous/glass/nanocrystalline cores: low and high permeability.  I'm not sure how the low-mu kinds are made, if they're just powdered iron (but... powder-ier...), but the most common kind are stripwound (like your example) and extremely high permeability, with 10k mu_r being on the low end.  As such, the energy storage is utterly miniscule, so they are completely useless as an inductor.

The primary applications are for common mode chokes (where the high permeability offers a higher attenuation factor than anything else, and the lossiness helps to suppress spurious signals), or transformers with very high power density (the frequency range is comparable to high-mu ferrites, but Bmax is 2-5 times higher, allowing for extremely compact transformers -- if you can keep them cool, that is!).

So, if you want to use that core, use it for a transformer in a forward converter topology, in the 20-100kHz range would be best.

Tim