Parallel inductors

[analityk]

Hi,
i want use inductor/choke for buck converter but with current value about 10-15A and inductancy about 10-15uH.
One choke should be big, expensive, hard to wiring on pcb and hard to hand soldering (due to huge copper mass).
So i find 4 smaller chokes about 47uH but for current 4A-5A (and 4 are cheaper than 1 big). Summary it should represent ~12uH and 16A. But i have never seen something like this in any circuit. Why?
Thanks for any help.

[Benta]

10...15 uH, 10...15 A ain't that big. Where have you been looking?

[analityk]

10uH is absolute minimum, so 10uH in paper usually falls off because with constant current flow decrease inductance.
I usually looking for tme.eu (in Poland is most known distributor), mouser, digikey but only for reference because taxes. If i can't buy something in mouser or digi i assume it is nonexistent.
47uH 4.2A: https://www.tme.eu/pl/details/de1209-47/dlawiki-smd-mocy/ferrocore/  ~0.32 euro and it gives 12uH @ 16A under 1 euro.
Why i have to not parallel it?

[mariush]

You say 10-15A yet you link to a 4.2A inductor...

Lots of higher current inductors that are easy to solder on Digikey, starting from around 1.3$ : https://www.digikey.com/short/nf77n884

[analityk]

I've wrote up that 10uH is to low value due to constant current flow through choke decrease inductance. So you should check this value under 10A constant current, I bet it will be half.
Yes, I show you 4.2A 47uH and connect it by four parallel, and this is clue of this thread. Give me reason to not doing it, not parts that is useless.

[mag_therm]

Parallel inductors are not common, I can only recall one time I saw it in day job, connected in Z to help sharing.

However it should generally work Ok.

Parallel transformers are common.

I am going to try parallel inductors in hobby
 20 A power supply because I have lots of small surplus E cores and magnet wire.

I suggest you go ahead with a trial prototype.

[analityk]

Prototype is currently under development, this is time for asking questions like this.
Another important factor is how many parts (of some sort) is on stock in distributors. It is a lot of gold parts, only you have no idea when you can buy it. Chocke typed by me is popular. Bigest part are not common and can level up to unobtainum very quick.

[eugene]

The first problem that I can think of is not sharing current equally. If one of the four happens to have fewer turns than the other then it will have lower inductance and lower DRC. That one will get more than 25% of the current.

The other potential problem that I can think of is interaction between them. Some cores can couple together when close together which can make the inductance go up or down depending on how they are arranged. The type in the data sheet has a fair amount of leakage of magnetic field, but they should not cancel each other. Give them space, and as always, make sure there is no copper on any PCB layer under them. If there are ground and power planes remove the copper in that area.

Personally, I would do it ...

[MrAl]

Hello there,

i had considered using multiple inductors in parallel for a buck circuit too.  The reason was that the larger chokes were more expensive, and for this project it was a one-off so i could get smaller chokes discounted a lot more than the bigger ones, so it made sense to use two or more.

Of course with 4 inductors in parallel the inductance drops by 1/4 of any one inductor which i am sure you know, and the total current goes up by 4.  If the total current has to be 16 amps and you use 4 amp inductors that could work if the saturation current is much higher.  For example if the saturation current is 8 amps i would say it would work unless you have unusually high current transients.  They should all be the same model too of course from the same manufacturer.

As far as connecting them together, use thicker traces than usual and make sure each lead can be soldered effectively to help ensure proper current sharing.

[analityk]

I am thinking about saturation state and chocke tolerance and it give me bad feelings. There is need to group/sort inductors by initial inductance. As long as load can handle excessive current (or small overvoltage by short time) there is overheating part by internal chocke resistance.
If only one chocke go to saturate rest of them turn instatnly like open circuit and all current should flow by this one. 15A x 15A x 46mohm gives 10W. Way to much.

[tszaboo]

If you want smaller inductors and capacitors, the proper way to do it is to use multiple phases. That way the inductors are switched 180 degrees or so (depending on the number of phases) apart, reducing the ripple current, the amplitude of the noise and so on. It costs more silicon, that's the drawback. This would be the proper way of implementing it, otherwise you can run into the many issues listed above, and maybe more.

[jonpaul]

Need the frequency, voltage, duty cycle min max to specify a,buck inductor.

We have done designs output at 50..100A on a single ferrite core gapped inductor.



An SMPS usually needs a magnetics custom designed and wound, but some stock inductors exist.

Unsure for,Poland or EU, but in USA try OEM mfg, Coilcraft, Pulse Engineering, etc.


Just the ramblings of an old retired EE

Tak!

Jon

[Siwastaja]

15uH at Isat=15A should be available, look at Wurth Elektronik offerings for example.

Failing that, do not parallel inductors, instead do multi-phase. Inductors and capacitors are the most expensive part of the design anyway, and multiphase helps with capacitors. MOSFETs and controllers are usually low enough in cost so that a 2-phase converter (with the two inductors in separate phases) is not much more expensive than 1-phase converter with the same two inductors in parallel or series. You might be able to save the added cost in capacitor savings. In any case, the result will be better with less ripple current and ripple voltage. Especially on the input when we are talking about buck.

If you must, in general, series connection works better than parallel, then current sharing won't be a problem. (Equivalent to how capacitors work better in parallel; voltage sharing is a problem in series.) So if you must, use two 8.2uH inductors in series. Being able to use choose 8.2uH instead of 15uH means you have bigger Isat available with the given size.

[mag_therm]

Uneven saturation is a problem I hadn't considered.
One option could be to design each inductor to not saturate at a transient full current limit.
 However that would probably need custom build and would increase their  core size although not conductor size.
I will do some qucs and bench test as I was also thinking to parallel inductors here to keep the conductor size down.
The converter in progress here is essentially two phase as it is push pull with 4 half wave rectifiers. A single L_out needs 200 uH 20 A 45 kHz pwm.