Hi Ali,

if this converter is built for the lab (no industrial series production) I would think twice about a liquid coolant because on the one hand it could cause a lot of trouble (e.g. leakage and its consequences for the other components), and on the other hand, in order to have a low thermal resistance you need to separate the windings by canals for the liquid, and the question arises: If you would fill the canals with copper (i.e. use a wire with a larger cross-section, without canals) would not the generated thermal power decrease as far as to render the liquid unnecessary?

Instead, I would do a rough calculation or a simulation first, on the topic if - with a well suited core and wire as thick as possible - filling the inductor with a thermally (but not electrically, of course) conducting epoxy would do the trick (e.g. 3M TC-2810 or Electrolube ER2220). The epoxy could be shaped to have a flat surface on one side, proving low thermal resistance to the cold plate. One prerequisite is that you have access to a vacuum chamber in order to avoid air bubbles in the epoxy; if not, you may look for a company providing potting as a service.

Hope this helps.

Hi Ali 10 kw is low power in this game , air cooling should be fine up to max conductor temps of 140 Deg. If you do wish to LQ, you can use a tube for your windings and run distilled water through all your components with insulating tube inter connects. I have used this arrangement for 2.5 MW rectifiers and induction loops with a cooling tower. Using a sensor to measure the electrical leakage is a good idea as well as flow, level and turbidity sensors. Plan dual circuits if you need to maintain on the runs and allow redundancy should the circulating pump fail.

Thank you so much, Joerg and Kris! Very nice feedback.

What material do you have in inductor? What wire do you use? At what frequency do you operate? Do you have a gap? If so - how big?

Mikhail,

We are using powder toroid cores with no gaps (MPP 26 from Magnetics Inc.). Wire gauge is 12 AWG and the inductor works at 200 KHz.

You use 55735 MPP core.

At this frequency you have a very thick wire! You need to use a litz wire! I would suggest 1020strands*0.072mm . Or if you don't have it, then use AWG 18 x 100 strands. In this configuration you would have about 20w core loss and 0.5w copper loss.

And from my experience, use ferrite cores if you can (N95 material or alike), At 200kHz - MagInc - not that good

Hi Mikhail,

Thanks a lot for the comments!

We are actually using 55256A2 MPP (which is a little smaller than the one you cited.)

You suggested using a AWG 18 wire. I'm a little confused about the size; is it not too small for this current level (i.e. 30 A average current)? is there any reason (e.g skin effect, proximity effect etc.) that using this wire gauge makes sense at this frequency?

Also: One nice feature that powder cores have is that, unlike ferittes, they show soft saturation (i.e. less risk of sudden saturation and destruction of the system). But in terms of core losses, in general, do ferrite cores show less core losses or powder cores?

Thanks!

Dear Ali,

You can use water cooled indcutors. Such indcutors are available in the market.

Please for more detailed information please refer to the link: https://www.weldotherm.com/products/water-cooled-coil-inductor/

Best wishes

At 200kHz skin effect 0.05mm. Therefore 18AWG is maximum effective diameter. 100 strands of 18AWG - is OK for 30A current amplitude!

The feature you have described - is not that nice. If you use it for boost, buck or filtering , then it might be OK. In resonant SMPS, only ferrite with a gap. A gap adds great parameter stability, and also if you have a gap - it is more stable for saturation. Usually you make additional 20% current saturation, and inductor will work fine under all conditions.

Thanks Mikhail!

If I understand correctly, you are suggesting to use 100 strands of 18AWG wires. But is this not going to be a too thick wire to wind such a core with?! (or are you suggesting to use a 18AWG wire with 100 strands?)

What is your desired Inductance? For what application do you use this inductor? 30A is it RMS or peak to peak?

The desired inductance is 90 uH. And 30 A is the average current (it's an output filter inductor for a DC/DC converter and has about 4 A peak-to-peak current ripple riding on the 30A average current.)

I thought you needed 100uH :-), was pretty close. Well! The core is too small! You have 2 options.

1) stack up 2 cores of yours and use 10 strands of 18 AWG wire. It will take some skill to wind 40 (yes 40!) turns with 88% fill factor. Reduced inductance at full load - not to worry, probably you have a buck where it is not that important.

2) If you are not that skilled at winding, take the core I have suggested, use 20 strands of 18 AWG, do 34 turns (fill factor about 40%). That will yield you the best possible efficiency, and precise inductance at full load.

AND - no need to water cool your inductor!

Remember, well calculated inductor MUST have about 60C-80C working temperature! BUT - this it true for the ferrite!

N95 PQ50/50 - with the same inductance will be 4 times more efficient :-)

Thanks a lot Mikhail! Very useful feedback!

Dear Ali Shahabi,

Preferably, the inductor comprises an inductor core, such as an annularly shaped core, having a plurality of sub-millimeter sized coated particles having alternating magnetic and non-magnetic layers. This equipment comprises a liquid cooled inductor system including at least an inductor and an immersion fluid directly contacting at least a portion of the inductor, such as an inductor core and/or an inductor winding. The immersion fluid is a substantially non-electrically conducting fluid. A liquid cooling system, said cooling system configured for use with an immersion coolant directly contacting a portion of said inductor; a cooling line, said cooling line configured to pass through the immersion coolant, said cooling line configured to physically separate a circulation coolant within said cooling line from the immersion coolant outside of said cooling line.

Ashish