LC power filter (Pot Core) inductors for PCB mounting

[mike buba]

Hi All,

I need advice on the power filter inductors. I am planning to build a three-phase PCB inverter board for 5 kW. At 400 V output, that is approx. 7.2 A rms output line current. The switching frequency is 30 kHz and the LC power filter LC values are calculated as 3 mH and 6 uF.

In the past, I built a 5 kHz inverter, so the L was much higher (8 mH), and I ordered it as a stand-alone pot core type from HKR where I filled their form with data: switching frequency, dv/dt, current, etc.

Now, since I increased the switching frequency, apart from a lower value, a lower number of turns will also be needed, so I was thinking of putting the LC power filter inductors on the PCB board, since capactors are already there. Something similar like this: (a), (b), (c), (d), or (e)
Are there any inductor manufacturers that do Pot Core inductors for PCB mounting, where I would send waveforms and other data and they would design affordable bespoke inductors?

From the off-the-shelf product, I was able to find only this for 3 mH and 7 Arms.

Thanks, MB

[moffy]

Are there any inductor manufacturers that do Pot Core inductors for PCB mounting, where I would send waveforms and other data and they would design affordable bespoke inductors?

Thanks, MB

I would guess that a Toroidal Core would be preferable to a Pot Core for thermal reasons, the windings are not enclosed by the core material. The main reason for a Pot Core is if you want a shielded inductor plus the examples you provided all use a Toroidal Core.

[mike buba]

I read Pot Core inductors have lower EMI emissions..
Regardless, any companies that do bespoke Toroidal Core inductors?

[jbb]

I did some design work on a similarly sized inverter some years ago. With a nice high switching frequency like this, the 50 Hz AC current is much larger than the 30 kHz ripple current. This would mean you would be limited by Bmax, and need either a lot of ferrite or something with higher Bmax.  We used nanocrystalline (aka amorphous) iron UU cores because nano crystalline saturates at above 1500 mT versus most ferrites at 200 - 400 mT

RM cores may be of interest. They’re available in PCB mount and can pack quite densely. I don’t know about who might manufacture them for you.

If EMC is a concern, you might wish to come down a little from 30 kHz; some of the conducted emissions standards have a sharp drop in allowed emissions at 150 kHz, which is the 5th harmonic. Dropping to 29 kHz would place the 5th harmonic at 145 kHz which may help.

If you’re wanting to push down the size and weight, I can happily recommend 3-level topologies such as the T Type. They can (approximately) halve the required inductor size versus a basic 2-level inverter and also reduce switching losses (assuming they use the same switching frequency…).

https://www.rohm.com/topology/dcac_inv_npc_t_800v

[jonpaul]

Bonjour M. Buba:  Bravo for the good question.

Pot cores are last choice:  cost, fragility, and power dissipation: ,   potcore  surrounds the copper wdg, ferrite is aceramic with poor heat cond.
Pot core bobbins have small and delicate pins.

1/ We use toroids of ferrite or powdered iron, 
Most power toroids are self lead and glued to the PCB. See typical SMPS/UPS/PFC.

2/ Off the shelf power inductors  readily avail :

CoilCraft,
https://www.coilcraft.com/

Pulse Electronics  and many others.
https://www.pulseelectronics.com/
 
Micrometals for, calculators and cats for powdered iron toroids and other shapes.
https://micrometals.com/

3/ EMI and compliance are achieved by good design, shielding, grounding anf illters (CM and DM) rather than by core choice.



have an absolutely fantastic day!

Jon

(EE, power, HV  and magnetics since 1970s)

[mtwieg]

I read Pot Core inductors have lower EMI emissions..
Regardless, any companies that do bespoke Toroidal Core inductors?

Pot cores and toroids are both renowned for reducing external fields, but I've never seen a direct comparison between the two. Could be that pot cores are somewhat better at reducing external flux due to the core covering the windings. But I also would bet that this makes power handling of pot cores much worse than toroids: that core covering everything makes dissipating heat far more difficult for pot cores, which is probably why toroids dominate power entry filters.