[TowerTalk] Ferrites 31 vs. 77 material

David Gilbert xdavid at cis-broadband.com
Thu Oct 31 03:54:14 EDT 2019


Many (many) years ago I was a component engineer at Collins Radio with 
responsibility for passive components, including ferrite cores.  I was 
told by the core vendor that fabricating them was best likened to baking 
bread ... with all the potential variability that entailed.

73,
Dave   AB7E



On 10/30/2019 8:46 PM, Richard (Rick) Karlquist wrote:
> To expand on what Jim said:
>
> I was once trying to design a conducted power line EMI
> filter that had to have a lot of attenuation at one
> frequency around 100 kHz, and the power level was several
> kW.  I needed a rather huge expensive choke which was
> problematical.  I got the bright? idea that I could
> resonate the choke at 100 kHz with a capacitor.  I went
> into the lab and breadboarded the filter and it worked
> beautifully with a fairly small common mode choke.
>
> Fast forward:  all known vendors of chokes had extremely
> loose tolerances on inductance, like +/25% or even +/-50%.
> The vendors blamed this on the core manufacturers, and
> I confirmed that none of them made closely specified cores,
> and on top of that, the tempco was high, so I couldn't
> even hand tune them by picking the cap values.
>
> I ended up using two large expensive chokes that I could
> barely shoehorn into the allotted area.
>
> Rick N6RK
>
> On 10/30/2019 10:51 AM, Jim Brown wrote:
>> On 10/30/2019 6:43 AM, Roger Parsons via TowerTalk wrote:
>>> K9YC wrote:
>>>
>>> "Yes, but virtually all of the recommendations for #43 and #52 
>>> material are WRONG!"
>>>
>>> Steve is sadly no longer with us and unable to defend himself. 
>>> However, he had previously rigorously defended his methodology. He 
>>> was very technically able in both theoretical and practical matters, 
>>> and I personally trust his results.
>>
>> There is nothing wrong with his methodology nor his measurements. The 
>> thing he missed, and that I also missed until recently, is the effect 
>> of component tolerances. About a year ago, I bought 40 of the #52 
>> Fair-Rite cores, split into four orders over about a month from four 
>> Fair-Rite distributors measured the characteristics of each, and put 
>> the data in a spreadsheet. I then selected cores at the limits of Rs 
>> and Xs and wound chokes following Steve's guidelines. None of those 
>> chokes were even close to his published data.
>>
>> My methodology is discussed in detail in
>> http://k9yc.com/2018Cookbook.pdf
>>
>> 73, Jim K9YC



More information about the TowerTalk mailing list