Author Topic: Machined ferrite cores  (Read 1537 times)

0 Members and 1 Guest are viewing this topic.

Offline nimishTopic starter

  • Regular Contributor
  • *
  • Posts: 144
  • Country: us
Machined ferrite cores
« on: February 08, 2023, 09:52:28 pm »
Has anyone ever gotten a custom machined ferrite core for a prototype? How much did it run/ who did you go with?

 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3319
  • Country: fr
Re: Machined ferrite cores
« Reply #1 on: February 09, 2023, 02:23:29 am »
diamond lapp
ELNA Ferrite Sales NY
TDK, Siemens Philips ELCO  make some gapped cores

We just use a DIY gap across the face

Jon
« Last Edit: February 09, 2023, 03:05:07 am by jonpaul »
Jean-Paul  the Internet Dinosaur
 
The following users thanked this post: Someone, pardo-bsso

Offline Someone

  • Super Contributor
  • ***
  • Posts: 4510
  • Country: au
    • send complaints here
Re: Machined ferrite cores
« Reply #2 on: February 09, 2023, 02:35:02 am »
Asked TDK for pricing and lead time on production qty.... never heard back.
 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3319
  • Country: fr
Re: Machined ferrite cores
« Reply #3 on: February 09, 2023, 03:06:24 am »
contact in USA is TDK importer MHW international Mahwah NJ
Jean-Paul  the Internet Dinosaur
 

Offline TimNJ

  • Super Contributor
  • ***
  • Posts: 1649
  • Country: us
Re: Machined ferrite cores
« Reply #4 on: February 09, 2023, 02:04:52 pm »
How big is the core and how many do you need to make?

I have ground many dozen ferrite by hand. If the gap is large, I start with a CBN/diamond drum on a Dremel for bulk material removal. Then diamond hand file to finish it off. Usually 30 minutes or less to get it done, for a core of roughly RM10 size.

Here is some previous discussion: https://www.eevblog.com/forum/projects/tips-for-gapping-ferrite-cores-on-a-surface-grinder/
 

Offline nimishTopic starter

  • Regular Contributor
  • *
  • Posts: 144
  • Country: us
Re: Machined ferrite cores
« Reply #5 on: February 09, 2023, 03:07:47 pm »
This is more than just gapping or lapping, I want to machine core shapes like this odd one: where I'd mill it from a block if ferrite weren't nasty hard ceramic that shatters if you look at it funny

« Last Edit: February 09, 2023, 04:06:18 pm by nimish »
 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3319
  • Country: fr
Re: Machined ferrite cores
« Reply #6 on: February 09, 2023, 03:58:08 pm »
One off/ prototype / experimental

NC mill, diamond tools. Fixtures will be critical as the thin x section shape will be very prone to cracking. Very poor design

Production... send RFQ to Fair Rite if you can pay for setup and tooling.
Practical for 10k/100k EAU

Jon
Jean-Paul  the Internet Dinosaur
 

Offline nimishTopic starter

  • Regular Contributor
  • *
  • Posts: 144
  • Country: us
Re: Machined ferrite cores
« Reply #7 on: February 09, 2023, 04:08:10 pm »
One off/ prototype / experimental

NC mill, diamond tools. Fixtures will be critical as the thin x section shape will be very prone to cracking. Very poor design

Production... send RFQ to Fair Rite if you can pay for setup and tooling.
Practical for 10k/100k EAU

Jon

Oh yeah that's just an example, but it's indicative of the kinds of geometries I want to explore, obviously designed to be manufacturable. Still in the prototyping phase so not interested in setting up large run tooling/NRE -- I'd really like a "please make this STEP" type arrangement

I've been using COTS cores for a while even with client projects but there's a rather massive improvement with shifting to custom integrated magnetics
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
Re: Machined ferrite cores
« Reply #8 on: February 09, 2023, 04:13:40 pm »
The shape does not look easy to machine and there can be areas that see a low field and thus keep remanent fields.

I would not count on the flux to share equally between the 4 outer connections: the µr is not very homogenous and already moderate stress can have quite some effect on the µr.  The transformer with fractional turns may work as a relatively sensitive (though not accurate) fore sense. 
 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3319
  • Country: fr
Re: Machined ferrite cores
« Reply #9 on: February 09, 2023, 04:48:01 pm »
What is a "COTS" Core ?

What is your motivation to DIY a new core shape?

We found the existing core shapes are fine, and bobbins TH, SMD are generally available

If you DIY a custom core, what about the bobbins?

Have used RM, EER, ER, PQ, UI, UU, in designs SMPS, EBU, HV, pulse.

What's the intended market or,application ?

Power thru put ? frequency min/max ?

Jon
« Last Edit: February 09, 2023, 06:09:05 pm by jonpaul »
Jean-Paul  the Internet Dinosaur
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 7934
  • Country: us
  • Retired, now restoring antique test equipment
Re: Machined ferrite cores
« Reply #10 on: February 09, 2023, 06:45:49 pm »
"COTS" is the normal abbreviation for "Commercial Off-the-shelf", as opposed to custom or customized.
 

Offline nimishTopic starter

  • Regular Contributor
  • *
  • Posts: 144
  • Country: us
Re: Machined ferrite cores
« Reply #11 on: February 09, 2023, 06:46:19 pm »
What is a "COTS" Core ?

What is your motivation to DIY a new core shape?

We found the existing core shapes are fine, and bobbins TH, SMD are generally available

If you DIY a custom core, what about the bobbins?

Have used RM, EER, ER, PQ, UI, UU, in designs SMPS, EBU, HV, pulse.

What's the intended market or,application ?

Power thru put ? frequency min/max ?

Jon

COTS = commercial off the shelf i.e. no customization standard part

The application is <1kW HF converters with planar and/or integrated magnetics. Material would be TDK PC200 or Fair-Rite 67, 80 etc.

You cannot easily find stuff like fractional turn cores, of which that's simply an example of a geometry. The actual one would be FEM optimized to a much better shape for the application and for manufacturability. This is for prototyping to see if it's even worth it, and to explore how we can increase density and efficiency.

Standard cores are ok but have limited shapes to really push the needle. You're working around the geometry not with it.
 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3319
  • Country: fr
Re: Machined ferrite cores
« Reply #12 on: February 09, 2023, 07:02:27 pm »
if using soft switching, resonant mode, or other modern topologies with SiC, GaN  at high frequency, then I am at a loss why a new shape is required.

As for fractional turns, every circuit is by definition closed.

The current loop eventually results in a whole number of turns.

Jon
Jean-Paul  the Internet Dinosaur
 

Offline Hawaka

  • Regular Contributor
  • *
  • Posts: 93
  • Country: ch
Re: Machined ferrite cores
« Reply #13 on: February 09, 2023, 08:58:12 pm »
That is quite a hard shape to machine. It would most certainly need a custom mold, even for a ferrite core manufacturer that has the capability of machining cores. Kolektor seems to have something very similar in their catalogue pictures, but it is not listed in it (https://www.kolektor.com/uploads/pdf/kolektor_soft_ferrites.pdf).

If it's just for prototype I would try to buy small standard pieces and then glue them together. If needed slicing or grinding, but not in an interrupted way. See attachment for an idea.

In any case the hardest part will be to find the right pieces in the right material.

 
The following users thanked this post: Someone, nimish

Offline nimishTopic starter

  • Regular Contributor
  • *
  • Posts: 144
  • Country: us
Re: Machined ferrite cores
« Reply #14 on: February 09, 2023, 09:05:56 pm »
if using soft switching, resonant mode, or other modern topologies with SiC, GaN  at high frequency, then I am at a loss why a new shape is required.

As for fractional turns, every circuit is by definition closed.

The current loop eventually results in a whole number of turns.

Jon

It's never *needed* of course, the goal is R&D :)

If you can customize magnetic cores, then you can do stuff like https://cpes.vt.edu/library/viewnugget/927 and simultaneously lower losses, size and occasionally large scale cost. For fractional turn, really fractional flux, https://ieeexplore.ieee.org/document/9214982 makes the point that it greatly reduces secondary side losses. You can get neat effects like flux cancellation that can have rather impressive effects on core loss, size and weight.

Absolutely not at all necessary in 99% of cases but that's the point here. Don't over index on that one pic, it's just an example of the kind of unavailable geometry that would need machining.
« Last Edit: February 09, 2023, 09:09:36 pm by nimish »
 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3319
  • Country: fr
Re: Machined ferrite cores
« Reply #15 on: February 09, 2023, 09:23:22 pm »
Actually the industry has achieved excellent efficiency, low EMI and fine control using modern switches, topologies and conventional core shapes.

I feel the R&D in cor shapes is a fine work but will not be implemented as the economics to tool, proto, qualify and manufacture a new shape will be practical only in 1M and over quantities.

 ferrite factories are closing and very overbooked already, what mfg will fab th production if you succeed? Who will design and fab the bobbins?

You may be able to assemble the proto  by modifying an existing shape to your new design.

Bon Chance ]


Jon
Jean-Paul  the Internet Dinosaur
 

Offline nimishTopic starter

  • Regular Contributor
  • *
  • Posts: 144
  • Country: us
Re: Machined ferrite cores
« Reply #16 on: February 09, 2023, 10:02:55 pm »
Actually the industry has achieved excellent efficiency, low EMI and fine control using modern switches, topologies and conventional core shapes.

I feel the R&D in cor shapes is a fine work but will not be implemented as the economics to tool, proto, qualify and manufacture a new shape will be practical only in 1M and over quantities.

 ferrite factories are closing and very overbooked already, what mfg will fab th production if you succeed? Who will design and fab the bobbins?

You may be able to assemble the proto  by modifying an existing shape to your new design.

Bon Chance ]


Jon

Sure, the industry has basically optimized the last penny out of cost. But one client had a situation that was not ferrite cost limited because the savings from even 0.2% gain in peak efficiency would pay for a million ferrite cores. Obviously I have no desire to use a non-standard core but it's very much an issue that I can't even find a good R&D process to see what could be done!

 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3319
  • Country: fr
Re: Machined ferrite cores
« Reply #17 on: February 09, 2023, 10:30:59 pm »
Rebonjour Cher monsieur

Fine notes, agree with small gains,  scaling with quantity

The greatest benefits can be the Pfe core loss optimization eg matériel optimiséd for freq, or hot spot temp min losses

The core shape affects the thermal resistance Hotspot to ambient as well as cooling by radiation convection and conduction.

Pcu is reduced with minimal turns, Litz, bunched wire or foil

These other factors may be more significant for efficiency and size  than an exact core shape.

Your r&D might be interested to simulate, and model, rather than cutting ferrite.

The FEM SW is well known, here is just one

https://www.melexis.com/

Jon

Jean-Paul  the Internet Dinosaur
 
The following users thanked this post: nimish

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3319
  • Country: fr
Re: Machined ferrite cores
« Reply #18 on: February 10, 2023, 05:22:14 pm »
Rebonjour, many fine papers, seminars and text on optimization of magnetic shapes, materials and transformers.

Before you proceed I would check the simulation FEM SW sites they may have good examples and results along this same line.

Then see IEEE Power Electronics and Magnetics proceedings and seminar /conf papers in the past.

The general optimization of power magnetics was nicely covered in 1980s by ou old friend Josef HEYDRIKS of Teledyne.

Unsure if it was published but I have his monograph/manuscript somewhere.

The trend in low profile has been the use of planar cores with multilayer PCB traces as the windings, at 100kHz..25 MHz.
 
Bon chance


Jon

Jean-Paul  the Internet Dinosaur
 
The following users thanked this post: nimish


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf