For a given inductance, is it desireable to use large toroid cores with just a few turns of wire, or a smaller core with more turns? I understand the I^2 losses from the wire, but not sure about the losses in the ferrite core. For a given power level, how would you size the ferrite core?
--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---
As big as you can accommodate - both space-wise and cost-wise.
A bigger core with fewer turns has smaller wire losses, and probably smaller losses in the ferrite, since the flux density is going to be lower (same volts.seconds spread over a larger cross-section).
With fewer turns you've got less winding capacitance. If you can get away with a single-layer winding this is less of a worry, since layer-to-layer capacitance ends up as a lot higher than turn-to-turn capacitance after you've lumped it.
Inductor and transformer design is a total can of worms - there's a lot going on, and not a lot of good guidance around.
The Siemens - now EPCOS - soft ferrite application notes were a whole lot better than their Mullard - now Ferroxcube - equivalents in the late 1970's.
E.C. Snelling must be the worst technical author ever. The Siemens technical notes did include the transformer equation, which was what got me comfortable about transformer design. People who teach the stuff don't seem to like it as much.
-- Bill Sloman, Sydney
You don't want to get it into saturation. For those converters wot do store energy in the core. nonisolated step up,down,buck converter stuff. Also isolated flyback... For those that work more like transformers, it's less of a problem... push pull isolated and similar... I think.
For the `cyclic power loss'(in the core) there's applications around... (I forget what they were called). It depends on the material... (Larger Hysteresis = Bad) The usual spice inductor models won't do that..
LT Spice will let you model an inductor - not a transformer - core with hysteresis, using the John Chan model.
You do have to dig a bunch of parameters out of the core data sheet, and the saturation data out of the materials data sheet, but it's pretty straightforward, and seemed to work the one time I used it.
-- Bill Sloman, Sydney
Engineering being an optimization of all things, cost is certainly part of it!
Ambiguous proportions. Same turns for larger A_e gives proportionally lower Bmax and therefore lower p_c. v_e is larger, so P_c isn't necessarily lower, but p_c(Bmax) is usually about square law, and v_e is
3/2 with respect to A_e, so it should be P_c(A_e) ~= A_e^(-1/2), which is a net win, but not proportionally so.Which means you can reduce turns by A_e^(-1/4), which is a rather small advantage really. Or it's linear and I'm confusing myself, which is possible. Run the numbers, in any case.
An ideal eddy current material has p_c ~= Bmax^2, but real materials are a little bit above or below that, with some ferrites having an exponent up to 3, which means big savings for especially low Bmax.
You also incur higher copper losses, due to proximity effect. Especially for the first layer(s), and especially especially near the gap (if the center leg is gapped, which is typical). This can make Litz look especially good.
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com
Thanks, that's what I suspected. I see there are some toroid cores designed for filters so they have heavy losses to absorb transient voltages, and others designed for low loss transformers.. Are there more than 2 cores designed for low loss or heavy loss? I have a bunch of unmarked torroid cores and just wondering if they are classified low loss or high loss or something in between. Some of them I know are for filters since they don't work very well as a transformer or high Q inductor.
--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---
An excellent reference on inductor and transformer design is Col.William Lyman's lasssic book available at :
There's no easy way of finding out that I know of. There are quite a few different ferrites used to make ferrite cores.
The gross distinction is between manganese-zinc ferrites - higher permeability and lower resistance, typically used below 1MHz - and nickel-zinc ferrites with lower permeability and higher resistance, typically used above 1MHz.
They aren't expensive. If it's been worth the effort of posting a query, here it's worth buying something identifiable from a distributor.
-- Bill Sloman, Sydney
Bookmarked for future reference. Thanks.
Sylvia.
Not really. Having been through the design process many times, I can safely state: inductors are never* chosen for damping, only for cheapness (once everything else is met, that is).
The yellow-white and green-blue (#26 and #52 respectively) cores are the cheapest of all, and just so happen to be fairly lossy (typical Q around
10 or 15, at power switching frequencies; dropping to ~1 in the MHz).Let me guess, I just pegged over half your pile of cores, eh?
Micrometals has a list of power and RF types, their colors and standard sizes. Most manufacturers follow these conventions. I don't know who all actually make the things; I've got to believe Micrometals buys from a verified Chinese vendor. And likewise, if you go shopping on Alibaba, you should find a half dozen others, plus a hundred resellers, of varying reputation.
Solid colors are usually something else. White, gray, green, blue, black and others. Arnold used to be pastel (Robins-egg) blue, but they're part of Magnetics Inc. now. I've most often seen white and blue on coated ferrites, and others on powder, but you really need to measure them to be sure. Black is usually Kool-Mu, a reasonably low loss powder material.
*I have seen the rare use of powder cores in EMI filters. Just a few turns. Clearly for tweaking the performance at high frequencies, probably to modify or dampen a resonance due to winding capacitance.Note the specific difference between ferrite and powder cores: ferrite toroids are ungapped, and therefore store extremely little energy. They are only suitable for transformers (whose purpose is to have a high reactance relative to the circuit impedance). Powder cores effectively contain distributed air gaps, so have lower apparent permeability (almost always 75 or less, with MPP going up to 250 as the exception), and are only suitable as inductors. Whether those inductors are coupled (such as the transformer for a flyback supply), doesn't matter -- I personally define an energy-storing transformer as a coupled inductor, just to keep them clear.
Note that common mode chokes count precisely as transformers: their purpose is to have a high common mode impedance.
Ferrites come in toroids as well as cut shapes; the shapes can be gapped (by grinding or shimming) to cover either purpose.
Tim
-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com
I downloaded all those chapters last summer and made them into a single "book" PDF....
...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
I'm not sure how Snelling got into this conversation, but I'm appalled at your statement. I've collected most of the books written by various authors (and companies) on the subject, and his detailed book, Soft ferrites: properties and applications, has been so useful I can hardly overstate its value to my work. Highly useful discussions and graphs on the proximity effect, winding fill-factors, winding capacitance, and other exotic topics have proved invaluable. My copy is carefully annotated. I've scanned some annotated chapters so I can refer to them wherever I'm working.
It's just a shame his book has been out of print and various attempts to resurrect it have failed. There aren't many copies out there.
--
Thanks,
- Win
In Slowman's opinion the only intelligent being on the planet is Slowman >:-} ...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
Only $594...
...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
A follow-up text...
...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
I just checked my supply of cores and found lots of pretty cores with different colors. Some are blue with a yellow stripe on top and others are yellow with a white stripe. And others are solid blue, green, black, white and one red one with a narrow gap.I wasn't aware the little solid red core with the gap would store more energy. I understand the gap would prevent the core from saturating, but the gap would also reduce the inductance. So if the energy is 1/2*LI^2 and you increase I while reducing L, how do you get more energy storage? Or is it some square law thing?
--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---
I'm not being critical about the content - which I've used off an on over years, but about the organisation and presentation, which is unspeakably horrible.
As a technical expert on ferrites he's fine. As an author he's a catastrophe.
It's not an easy book to read, and you won't read it unless you know what you want to find, and have some idea where to find it in the morass.
It's not something that you'd give to a junior engineer to read - you'd have to point them at a section, and explain the kind of information they were to look for.
-- Bill Sloman, Sydney
Jim-out-of-touch-with-reality-Thompson confuses subject-organising skills with general intelligence. I've never tried to write a text-book, or even a narrow technical monograph, but I've read enough of them to know that E.C.Snelling is uniquely bad at it.
Soft ferrites are a very specialised area, so his text is useful, despite it's horrible presentation and confusing organisation, because it's close to unique, but getting information out of it is like having teeth pulled.
-- Bill Sloman, Sydney
Saturation is exclusively about flux.
Saturation is specified in terms of Bsat == V.s / m^2 (or uVs/mm^2). And per turn, if you want to keep count of that too.
When you multiply this by the core area A_e, you get a flux (V.s == Wb), which is the area under the curve, the time integral of voltage. If you have square pulses (a handy thing about SMPSs...), this is simply width by height. (Note that, if it's a single ended, return-to-zero circuit, like a DCM flyback, then this is the pulse width and height, period. For a full wave, above-and-below-zero circuit, take half, because half the flux charges from -pk to zero, and the other half from 0 to +pk. For an unbiased sine wave, take 1/(sqrt(2)*pi) of the RMS voltage instead.)
The flux is precisely the EMF applied to the winding, but real windings have resistance and leakage, so mind that the terminal voltage won't be exactly this, but if you're not going out of your way to make a shitty winding, it'll be pretty close.
Current is all about magnetization, the H field, and gapping. Think of it as the current drawn /as a result of/ an applied flux.
The conversion between B and H is permeability: B = mu*H. mu_0 = 1.257 uH/m (or nH/mm if you prefer).
Oh, and henry == V.s / A.
The length unit in H is the core length, l_e.
This becomes very apparent if you set up a test, where you're pulsing voltage into an inductor and measuring the current. You see current transition from smooth ramp to pointy spike as it saturates. The slope of the ramp varies with gap length, /but the time saturation occurs at does not/. Saturation flux is the invariant (well, nearly).
Clearly, a high-mu core has very little magnetization (usually a few A/m), so you store very little energy in a hi-mu core. Indeed, the Maxwell Stress is: e = B^2 / (2*mu) (substitute mu == mu_r * mu_0 of course) e has units of energy density (J / m^3) or pressure (== N / m^2 == Pa), which is, in itself, a useful unit equivalence to keep in mind (atmospheric pressure is also energy density, and the reason should be obvious after a little thought! :) ).
If you add gap to a ferrite core, you get a region of size l_g * A_e (actually a little more than both due to fringing, but we can ignore that for l_g
gauss, oersteds, copyright 2004?? wtf...
joe
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.