Maker Pro
Maker Pro

ferrites

R

R.Lewis

Jan 1, 1970
0
I need some small(ish) inductors at around 100mH (more precise value to be
determined) for some filters for use at a number of frequencies between 200
and 700Hz
It is critical that the inductance stays constant over several years - the
current will be in the order of 10mA with a minimal dc
component.
If I use a ferrite will I encounter any problems at these low frequencies
(book data is always in the kHz and MHz range) and will a ferrite remain
stable over, say, 5 years?
Temperature variation of the ferrite is not a problem (at this time).


Thanks in anticipation.
 
A

Adrian Jansen

Jan 1, 1970
0
Ferrite has a property called disaccomodation, where the permeability drops
roughly logarithmically with time since the last major magnetic upheaval (
eg core saturation ). Once you get beyond a few days, the effect is usually
negligible compared to the temperature coefficient. A data sheet for the
ferrite in question will show the disaccomodation value, and how to estimate
the long-term effect.

--
Regards,

Adrian Jansen
J & K MicroSystems
Microcomputer solutions for industrial control
 
Q

qrk

Jan 1, 1970
0
I need some small(ish) inductors at around 100mH (more precise value to be
determined) for some filters for use at a number of frequencies between 200
and 700Hz
It is critical that the inductance stays constant over several years - the
current will be in the order of 10mA with a minimal dc
component.
If I use a ferrite will I encounter any problems at these low frequencies
(book data is always in the kHz and MHz range) and will a ferrite remain
stable over, say, 5 years?
Temperature variation of the ferrite is not a problem (at this time).

You can use ferrites at low frequencies. Just beware of flux density
problems at lower freqs. For filters, you want to keep your flux
density very low.

You don't say what kind of stability you want. So, blanket answer is
maybe. You can improve the long term stability of your inductor if you
temperature cycle the inductor. I forgot what the temp cycling
parameters are - something like 5 or 10 cycles where you ramp up from
0 to 80 deg C in one hour, hold at 80 for 30 minutes, ramp down to 0
in one hour, hold at 0 for 30 minutes, .... This settles the inductor
mechanically (wire, ferrite interfaces, spring interfaces). If your
inductor is subjected to vibration, you will have problems with the
inductance changing. If you lack a programmable temperature chamber,
place the inductors on the dashboard of your car and park in the sun
for a couple weeks.

Be sure to use a gapped core.

Have you considered an active filter?

Mark
 
R

R.Lewis

Jan 1, 1970
0
qrk said:
You can use ferrites at low frequencies. Just beware of flux density
problems at lower freqs. For filters, you want to keep your flux
density very low.

You don't say what kind of stability you want. So, blanket answer is
maybe. You can improve the long term stability of your inductor if you
temperature cycle the inductor. I forgot what the temp cycling
parameters are - something like 5 or 10 cycles where you ramp up from
0 to 80 deg C in one hour, hold at 80 for 30 minutes, ramp down to 0
in one hour, hold at 0 for 30 minutes, .... This settles the inductor
mechanically (wire, ferrite interfaces, spring interfaces). If your
inductor is subjected to vibration, you will have problems with the
inductance changing. If you lack a programmable temperature chamber,
place the inductors on the dashboard of your car and park in the sun
for a couple weeks.

Be sure to use a gapped core.

Have you considered an active filter?

Unfortunately an active filter is not possible (just believe me).
I am looking for a max (absolute max) 10% drift in inductance over about 5
years but data on disaccommodation seems to be rarer than rocking horse
s**t.
All the manufacturers have reference to it in the preambles of their data
books but any numbers then seem remarkably absent.
I was rather hoping to use something like an E20 size wise (maybe even an
EF25).

Any idea as to where to look?
 
T

Tony Williams

Jan 1, 1970
0
Unfortunately an active filter is not possible (just believe me).
I am looking for a max (absolute max) 10% drift in inductance
over about 5 years but data on disaccommodation seems to be rarer
than rocking horse s**t. All the manufacturers have reference to
it in the preambles of their data books but any numbers then seem
remarkably absent. I was rather hoping to use something like an
E20 size wise (maybe even an EF25).
Any idea as to where to look?

Siemens (aka Siemens-Matsushita, aka Epcos) publish DF
data for most of their materials as standard, and give
a formula for using it.

(L1-L2)/L1 = DF * ue * log(t1/t2).

In their worked example (for material K1):
Air-gapped pot core, so has an effective ue = 15.9.
DF for material K1 is listed as 35exp-6.
t1/t2 = 500weeks/5weeks.

They calc'd (LI-L2)/L1 as 0.0011, or 0.11%.

Epcos data for various (prefered) materials is below.

Material Initial permeability DF (* exp-6) Rt/c
~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~ ~~~~
K1 80 <35 4
M33 600 <12 1.6
N58 1200 <6 0.9
N48 2000 <4 0.7
N41 3000 <6 4

N48 material is one of those listed as suitable for high-Q
inductors (with an air gap normally), and for operation
down to 1000Hz.

That cryptic Rt/c on the right is the average value of the
relative temperature coefficient of permeability for each
material, over the range 20C to 55C. N58 and N48 have the
lowest t/c.
 
R

R.Lewis

Jan 1, 1970
0
Adrian Jansen said:
Mostly numbers for disaccomodation are published for core materials designed
for filters. Check around the Philips site - if you can find the right
area. Else I might be able to dredge up some very old data, to give you a
start. 10% over 5 years sounds quite ok, but it depends heavily on the
material and permeability ( core gap ) you have chosen. Post some numbers.

--
Regards,

Adrian Jansen
J & K MicroSystems
Microcomputer solutions for industrial control
Thanks for the response.
The (sort of thing) I am looking at is an EE20 using Philips 3C90 or,
preferably, AVX F1 material gapped to give an Al of 100.
I need 64mH (with an initial tolerance of +/-12% max) and a coil resistance
no greater than 16 ohms. Coil current, at this time, would be 10mA or less.
I can then stand an inductance drift of 10% (up or down) due to temp and
disaccommodation during service.

Philips do not seem to quote any disaccommodation figures in the data book I
have but do conveniently quote the Al figures for the various standard gap
sizes for many of their cores except the EE20!!
The AVX databook mentions the disaccommodation phenomenon but quote no
figures (that I can find) and rather helpfully intimates that the EE20 F1
cores have an Al of 63 to 250 (or something similar) without informing of
anything else.

I need a service life of 5 years (or longer).

Any help at this stage is gratefully received.
 
T

Tony Williams

Jan 1, 1970
0
R.Lewis said:
Philips do not seem to quote any disaccommodation figures in the
data book I have but do conveniently quote the Al figures for the
various standard gap sizes for many of their cores except the
EE20!! The AVX databook mentions the disaccommodation phenomenon
but quote no figures (that I can find).....
[snip]

These Exx cores seem to be power cores, for which any
Disaccommodation is secondary, and their materials tend
to have no spec for DF. The traditional inductor core
shapes are pot cores and RM cores, and most materials
used in these shapes do have a DF quoted.

[required spec shifted down]
I need 64mH (with an initial tolerance of +/-12% max) and a coil
resistance no greater than 16 ohms. Coil current, at this time,
would be 10mA or less. I can then stand an inductance drift of
10% (up or down) due to temp and disaccommodation during service.

That spec can be achieved with an RM7 core with Al= 400
with no problem whatsoever and is just about possible
on an RM7 with Al= 250.

Note that RS 228-236 is an RM7 core with Al= 250, and the
material is grade A13/Q3/N28 (depending on manufacturer).

An RM7 bobbin will take just about 506 turns of 0.16mm wire
(0.177mm od including enamel). That would be assuming a
space factor of about 60%, which is reasonable for a single
winding without insulation layers. 506T should produce an
inductance of 64mH (+/- 3% unadjusted), with an initial
resistance of about 15.5 ohms.

Assuming a typical max DF of 6exp-6 for such materials, and
ue of 150, gives an effective disaccommodation, (L1-L2)/L2,
of about 0.25% over a t1/t2 ratio of 500/1 weeks.
 
B

Boris Mohar

Jan 1, 1970
0
I need some small(ish) inductors at around 100mH (more precise value to be
determined) for some filters for use at a number of frequencies between 200
and 700Hz
It is critical that the inductance stays constant over several years - the
current will be in the order of 10mA with a minimal dc
component.
If I use a ferrite will I encounter any problems at these low frequencies
(book data is always in the kHz and MHz range) and will a ferrite remain
stable over, say, 5 years?
Temperature variation of the ferrite is not a problem (at this time).


Thanks in anticipation.

Have you considered metglass or amorphous metal core? Just a thought.

--

Regards,

Boris Mohar

Got Knock? - see:
Viatrack Printed Circuit Designs http://www3.sympatico.ca/borism/
Aurora, Ontario
 
R

R.Lewis

Jan 1, 1970
0
Boris Mohar said:
Have you considered metglass or amorphous metal core? Just a thought.

--


No, I have not considered them because I know nothing about them.
Where would I best start to find out?

Thanks
 
R

R.Lewis

Jan 1, 1970
0
R.Legg said:
"


The philips/ferroxcube databook quotes disaccomodation only for those
materials designed for use in LF inductors. These will generally start
with the designator '3B..". This is quoted in the material properties
section.

http://www.ferroxcube.com/appl/info/HB2002.pdf

For this application, they also expect a sensible self-shielding part
shape to be used; like a gapped pot core, with or without a tuning
stub.

If you are designing for an application, follow the mfrs
recommendations, or be prepared to use your own time and resources
finding out why these recommendations are innitially made.

Applications requiring precision inductance values must be built
around this requirement. It is generally avoided, wherever possible,
in current design practice, for many good reasons, some of which you
may be experiencing or anticipating.

The last thing I want to (ever) do is to invent a new wheel when it is not
necessary and I heartily concur with the sentiment of your post.
However I am having difficulty in finding the data for a core to suit my
task.
As you state an RM or Pot core would be the obvious first choice except
Philips do not do either in a 3B(n) material.
The materials that are available (in these shapes) do not seem to be
specified below 10kHz and I am looking at 200 to 700Hz.
Initial accuracy is not as important as long term stability.

I am current treading water by wading through a megatude (?) of alternative
manufacturers data in search of the holy grail. Any help you can offer would
be appreciated.
 
A

Adrian Jansen

Jan 1, 1970
0
Philips 3B5 and 3B7 / 3H1 were both specified at 4 KHz for permeability and
losses. Mainly because they were designed for telephone circuits in the
voice band - 400 Hz to 4 kHz. I doubt things have changed that much, except
the telephony. Any material spec at a higher frequency will ( in general )
hold at a lower frequency.

--
Regards,

Adrian Jansen
J & K MicroSystems
Microcomputer solutions for industrial control
 
R

R.Legg

Jan 1, 1970
0
R.Lewis said:
The last thing I want to (ever) do is to invent a new wheel when it is not
necessary and I heartily concur with the sentiment of your post.
However I am having difficulty in finding the data for a core to suit my
task.
As you state an RM or Pot core would be the obvious first choice except
Philips do not do either in a 3B(n) material.
The materials that are available (in these shapes) do not seem to be
specified below 10kHz and I am looking at 200 to 700Hz.
Initial accuracy is not as important as long term stability.

I guess lack of 3B(x) parts is due to the historical age of this
material type. Current Ferroxcube recommended materials for LF
filtering (page 52 of their product selection guide) are 3D(x) and
3H(x) materials. Philips finally divested itself of it's ferrite
mfring arm (the self-same 'ferroxcube' facility) about 5 years ago.
I am current treading water by wading through a megatude (?) of alternative
manufacturers data in search of the holy grail. Any help you can offer would
be appreciated.

We still have no real idea of your application.

Temperature compensation of ferrite parts has traditionally been
achieved using polystyrene capacitive components in resonant or filter
applications. The two tempco's reacted to give an S-curve over a
fairly wide environmental range. This is 50 year old technology, using
70 year old materials, but it should still work if the part sizes are
practical and available, in your case.

Stability over time basically boils down to the careful construction
and packaging of the pot core. All those spacers, clamps and vibration
absorbing layers described in their selection guide, are intended to
make this a practical proposition. Doesn't look very simple, does it?

The polystyrene capacitors themselves are large, extremely expensive
in any size over 10nF and have an upper temperature limit (<<85degC)
and physical delicacy (they benefit from 'aging' and moisture control)
that restrict their use in many applications. Larger C values are also
typically offered only in low voltage (<63V) varieties.

They experienced a period in the 70's when many exotic and ungainly
sizes were offered in the surplus market, as the analog application
technologies were phased out. They are now quite rare, though some
fairly large parts (22nF) were still mfred by Philips at last count.
The polystyrene dielectric itself is avoided in other applications
where low-loss characteristics are still required, by substitution
with polypropylene or polysulphones, which have a different tempcos.

RL
 
Top