20-25% is actually what Fair-rite gives as a tolerance for their cores (at
least for the EMI suppression family) - Chip beads are 25%
So if you've measured 20% variation, that's probably within spec.
I note also that their properties change with temperature. 43 mix is 1.5%/deg
C for permeability - the graph shows mu= 600 at 0C and mu = 900 at 25C.
Impedance shows similar variations with temperature (partly due to the physical
size of the core changing, but more for the change in ferrite properties - and
the change is different for different frequencies).
31 material has impedance that goes up as temperature rises from 25C at 10 MHz,
and down at 100 MHz.
In the typical application, one designs for "more than enough" impedance, so if
it's 20% low, you're still ok.
Transformer core materials tend to be a bit better, both tolerance and tempco
wise. The %/C for some materials is 0.1%, versus >1% for the EMI suppression
mixes.
for instance, 79 Material mu is pretty flat vs temp from 75 to 150 C, and given
it's use in power converters, they probably run fairly hot.
The folks designing power converters at work pay a lot of attention to this
kind of thing. And I found out more when looking at using ferrites to suppress
RF currents on a cable laying on the lunar surface, which can have huge
temperature swings between day and night.
On Mon, 3 Nov 2025 04:52:34 -0800, "jim.thom jim.thom@telus.net"
<jim.thom@telus.net> wrote:
The issue is those Fair rite cores.....
which are now made in China . The tolerance is all over the map... like
+/- 20%. . Good luck trying to find 2 that are the same.
And hats off to everybody who is trying to enhance the state of the art.
Jim VE7RF
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|