a larger green toroid core from the same vendor:

5X(Product Name : Ferrite Core;Material : Metal
Color : Green

 

Saturation current was 3A and still 3A.
Series resistance was 40mOhm and became 0.64Ohm.
Exactly as TS want.
What is wrong?

I was not questioning whether it was possible to achieve the specs with inductors in series. My point is that it is unlikely to be more compact than a single core with the same specs. The volume and hence saturation goes up with the cube of the linear dimensions and the inductance goes up with the square of the number of turns. So, take one of your cores, and to get the required inductance you would not need to double the size or resistance to double the saturation current and the ESR, which you do by using two inductors.

Bob

 

Here are some numbers for some Micrometals cores and windings:

T184-52 250 turns for 10 mH zero bias, down to 5.5 mH at 2 A DC bias
T184-26 250 turns for 10 mH zero bias, down to 4.6 mH at 2 A DC bias
dimensions of core 4.7 od x 2.4 id x 1.8 h, cm

2 stacked T225-52 233 turns for 10 mH zero bias, down to 6.6 mH at 2 A DC bias
dimensions of core stack 5.7 x 3.5 x 2.8

same T225s, 500 t for 46 mH, down to 15 mH with 2 A DC bias

You can do better with MPP cores - higher initial permeability and less swing, but a T184 size core in small quantity is going to be well over $10 unless things have changed since I last bought any (probably increased - MPP is mostly nickel). There are some high-flux powdered iron formulations that will swing less - more expensive than Micrometals' 26 and 52 materials but still much less than MPP.

 

Again, ferrite core toroids are not suitable for the application. A common mode choke is intended to be operated with essentially zero net magnetizing force at low frequency.

There are really only two options for toroids that can handle DC bias and exhibit tolerable loss at high frequency.

Tape-wound cores consist of a very thin high-permeability metal that is wound to make a toroid shape. Each layer is insulated from the next by oxide or some other method to prevent eddy currents from flowing layer to layer. Core like this are not at all common and the alloys used are expensive.

By far the most popular core material for this sort of application is "powder", as I mentioned previously. Powder cores have vastly lower permeability than the ferrites used for common mode inductors. The ferrites will typically be between 5000 and 10000 perm. for CM chokes. There are very few choices in powder cores with permeability of more than about 150 - and that gets you into molybdenum permalloy powder that is very expensive. Inexpensive powdered iron tops out a permeability of about 100. In practical toroid geometries that are reasonably easy to wind, the ratio of cross section to magnetic path length does not vary by a lot. This means that regardless of the size of the core, there isn't much variation in inductance index, perhaps two or three to one, from one to another in a particular formulation. For example a Micrometals T106-26 (suitable material but not size for task at hand) is just over an inch OD and has an AL of 93 nanohenries per turn squared. A T400-46B which is 4" in OD has an AL of 205 nH/t^2. Micrometals type 26 material has an initial permeability of 75, is inexpensive and popular (and counterfeited!).

Taking the T106-26 for 10 mH at zero bias:

inductance is 10 x 10^6 nanohenries
divided by AL is 107527
the square root of which is 328
You would need 328 turns for a 10 mH inductor - that would be considered a "full" winding with about 25 AWG.
A 40 mH inductor would require twice a many turns.​

As soon as you apply any DC bias, the inductance will be reduced (my computer with my spreadsheet for doing this calc isn't running at the moment; I think Micrometals has on-line and down-loadable free tools for doing all the necessaries)

You can stack two toroid cores for twice the inductance index. Theoretically you could stack more, but practically the winding become a big problem.

Magnetics Inc. is a nice place to go core shopping for some of the more exotic core materials. Take lots of money with you.

Maybe getting one like this could offset that effect caused by DC.

 

@cmartinez:
How about using digital (I2C) controlled PWM chip with frequency 200...400kHz.
Then filters will be very tiny and MC will be free of continuous PWM pulses forming.
And programing will much easy?

 

@cmartinez:
How about using digital (I2C) controlled PWM chip with frequency 200...400kHz.
Then filters will be very tiny and MC will be free of continuous PWM pulses forming.
And programing will much easy?

It could be .... I'm going to check if the optos I'm using are capable of such high frequencies.

Did you have a particular pwm chip in mind?

 

It could be .... I'm going to check if the optos I'm using are capable of such high frequencies.

Did you have a particular pwm chip in mind?

May be optos for I2C only? Not high frequency.
I need some time for chip, may be with driver inside, searching.

 

See this one.
I will add to this post more.
https://www.idt.com/document/dst/zspm2000-datasheet
Very good controller IXDP610 but with parallel interface:
http://ixapps.ixys.com/Datasheet/91600.pdf
Seems controller IXDP610 (link above) is not so bad,
if you will use it with PCAL6408APW expander:
https://www.nxp.com/docs/en/data-sheet/PCAL6408A.pdf
Danko.

 

Higher frequencies are better to a certain extent in SMPSs. Inductive reactance goes up, and capacitors filter more easily. Those components do more. This means you can use smaller value components, saving space and minimizing the unwanted inductances/capacities. So you may only need a fraction of that inductance, giving you far more options. This also saves cost. However, there is a reason above a few MHz is so uncommon in SMPSs. There are also many unwanted inductances/capacities that have more of an effect. You get unwanted resonance, skin effect losses, and large current through unwanted capacities.

Additionally, this brings into play the switch itself. The gate/base of your transistor is a picofarad capacitor. This has almost no effect at lower frequencies. However, when you get into a few hundred kHz, it starts to need more current to charge and discharge it. This issue increases with frequency. A good driver can help you do this quicker, but it still must consume more current. And if you do not provide sufficient current, it will completely fail. It may also operate in the linear/ohmic region, dissipating large amounts of power, causing more issues.

When using a microcontroller, higher frequency means less resolution. Let's say you have a PIC or something similar with a 100MHz clock. This is certainly fast, but may still lead to issues. If you switch at 1MHz, you only get 0-100 for your PWM. Let's say it translates directly into a voltage (I know this is an oversimplification). Let's say you have an output range of 0-30V. that means 30/100 or steps of 300mV. That is not good at all! If you chose 100kHz, you get 0-1000, or about 30mV resolution. That is much better. You can get up to maybe one or two GHz, but it becomes impractical and expensive to operate a microcontroller at such high frequencies.

One last note, operating at audible frequencies will produce annoying sounds, so try to avoid that. Your inductor (or other component) may turn into a speaker, especially if vibrations are easily allowed. When balancing all these things, people generally go with 100kHz to 3MHz for switching. More details about your specific application would certainly help.

 

I connected two 100 µH in series like this one and, as expected, their total inductance added up perfectly to 200 µH when I measured it with my meter, no matter which phase was connected to which.

But something funny happened when I drew them close together. If both of their phases are connected forward-biased (that is, the non-dot phase of the first inductor connected to the dotted phase of the second one) their total inductance would come down to 165 µH, but if I connected them back-to-back (with both dotted phases connected together) then their total inductance would increase to 365 µH.

Question, can I take advantage of that phenomena to build a, say, single 44 mH inductor out of a two-line 12 mH choke by connecting its two windings back-to-back?

If the original 100uH inductors have a saturation current of 2.1A, then when you orient them to get 365uH total inductance, the saturation current will be much less than 2.1A, more like 1.5A. This is because the saturation current is when the magnetizing field from the winding saturates the magnetic flux in the core, and when you orient them to increase the inductance you are orienting them in a way that the magnetizing field from each increases the magnetic flux in the other, causing it to saturate at a lower current. I got the 1.5A estimate from assuming that saturation occurs at approximately the same stored magnetic energy (1/2 LI^2).

There's no such thing as a free lunch.

 

That's been the whole point of this thread ... the inductance values needed for what I want are not in stock at the major online suppliers ...

Mouser has updated its line of products concerning chokes:

https://www.mouser.com/Passive-Comp...1ya6qnbSGT&Keyword=choke&Ns=Pricing|0&FS=True

But now I'm intrigued as to @Danko 's suggestion of using a higher PWM frequency. With the AT89LP4052 running at 22.1184 MHz, a PWM frequency of up to 86.4 KHz can easily be generated. It's a little short of the 100KHz that was suggested, but it should be a more workable solution. I'm now going to start playing with the sim, and then jump into the real world and see how it all works out. Stay tuned.

 

But now I'm intrigued as to @Danko 's suggestion of using a higher PWM frequency.

@cmartinez, basic problem, we can not solve, is filtering PWM pulses in power line.
But it is possible using SPWM method for AC motor control.
Check this link
EG8010 - Single Phase Sinusoid Inverter ASIC:

 

@cmartinez, basic problem, we can not solve, is filtering PWM pulses in power line.
But it is possible using SPWM method for AC motor control.
Check this link
EG8010 - Single Phase Sinusoid Inverter ASIC:

Thank you so much, Danko. Right now I'm continuing this subject on this other thread. This because I discovered that filtering improved things significantly.

 

I've been trying to decide which inductor would be best for testing purposes in one of my PCBs, and I have a few questions:

• When it comes to filtering, what's the deal with the inductance value? Is it the more the merrier, or would the intended filtering frequency be affected? I mean, I'm sure there are applications that require a minimum of X mH in them (and of course, their current capability before saturation too). But what would happen if i.e., a 100 mH inductor were to be used instead of an (allegedly recommended) 20 mH one?

• The way I understand it, for filtering purposes it is better to use a toroidal inductor because the magnetic field generated by the transients being filtered remains contained. Whereas a cylindrical inductor generates an "open" axial magnetic field that could cause EMI. Is this correct?

• Would using a cylindrical inductor instead of a toroidal one for filtering purposes represent a significant difference in a, say, 12VDC digital circuit consuming less than 2 amps total? I don't know if the following info is relevant, but the particular circuit in question has a few MCU's running at 14.7456 MHz, and this MCUs in turn pwm a few nFet drivers at 10 KHz.

 

Hi cmartinez,
Maybe these articles (links below) will useful for you:

https://www.schaffner.com.tw/filead...ation_note/Schaffner_AN_RB_common_chockes.pdf
https://www.coilws.com/index.php?main_page=page&id=128