I have an interest in making RF filters with low distortion (for
example, as preselection filters for receivers). I've found that very
often I need to use air-core coils to get the low distortion I want,
but it would be convenient if I could use powdered iron or ferrite
core toroids to get higher Q at low frequencies without having to make
the coils physically big. I've always assumed that powdered iron
would give me the best results, but someone suggested I try low-
permeability ferrite. I just did some experiments with that. I wound
three toroids, all 0.37 inch OD cores; one is on mix 2 powdered iron,
and the other two are on type 68 and type 67 ferrite. I put each
coil in a simple 1MHz high pass filter: two 2.2nF caps in series,
with the coil connected between them to ground. I tested for harmonic
distortion with a 1MHz input, and for the 2.42MHz intermod product
with 0.62MHz and 1.52Mhz inputs.

With about +16dBm coming out of the test filter at 1MHz, I measured
second harmonic at about -95dBc for the T37-2, -90dBc for the FT37-67
and -57 for the FT37-68; the third was at -74dBc, -61dBc and -51dBc
for the three coils in the same order.

The 2.42MHz intermod, with each tone at +18.5dBm out of the combiner,
measured -71dBc, -58dBc and -38dBc for the coils, again in the same
order.

My conclusion from this is that at least these two ferrites aren't
going to help me in my search for a better low-distortion core
material, but at the same time, it was interesting to note that the
higher permeability -67 performed better by far than the lower
permeability -68. It gives me some hope that if the ferrite materials
guys went looking for a formulation that would be specifically as low-
distortion as they could get, they might find something pretty good.

Before someone asks, no, I didn't run a T37-6 core coil along with
these. I expect it would be a little better than the T37-2. And I've
measured a few coils wound on (unknown formulation) powdered iron rods
that show significantly lower distortion than those on toroid cores;
but of course those have a large "air gap" and relatively high
external field.

Cheers,
Tom

Good info Tom, I'd like you to try -61 material, but I don't have anything
but T50 in that ferrite.

W4ZCB

I have an interest in making RF filters with low distortion (for
example, as preselection filters for receivers). I've found that very
often I need to use air-core coils to get the low distortion I want,
but it would be convenient if I could use powdered iron or ferrite
core toroids to get higher Q at low frequencies without having to make
the coils physically big.

K7ITM wrote:

snip
I wound three toroids, all 0.37 inch OD cores; one is on mix 2 powdered iron,
and the other two are on type 68 and type 67 ferrite.

snip

How do you determine the formulation of the cores that you use (I am
presuming they are taken from the junk box...)?

Michael

On Apr 29, 6:17 am, msg wrote:
K7ITM wrote:

snip

I wound three toroids, all 0.37 inch OD cores; one is on mix 2 powdered iron,
and the other two are on type 68 and type 67 ferrite.

snip

How do you determine the formulation of the cores that you use (I am
presuming they are taken from the junk box...)?

Michael

Well, I trust places like Amidon and Bytemark to supply what they
say... The ferrite toroids were delivered last week from Bytemark.
The powdered iron come mostly from old stock we have upstairs, some
from ones I've ordered in the past from Amidon. (I have a different
batch of cores in front of me right now that I'll try running the same
test on. These cover a wider range of formulations, but all in (F)T50
size -- 'cept I couldn't find a T50-6, so I may look for a T37-6 or
just run a T68-6A I do have.)

But you bring up a good topic. You can measure the dimensions of a
core and figure its relative permeability from the inductance of a
coil wound on it, and make a guess about the formulation from that,
but short of doing more complicated tests, I don't know that you can
really determine the formulation. Powdered iron toroids generally
come with a color code paint job, but the ferrite toroids I've always
gotten have nothing on them to tell you what they are. I've found
that an "extra fine" tipped paint pen is a good way to put a marking
on them. You could color-code them, but I just write what they are on
them, such as "FT50-67". That's not so important if I keep them
nicely sorted, but if I've wound a coil on one and then want to re-use
it, it's a big help.

Cheers,
Tom

On Apr 28, 5:25 pm, "Harold E. Johnson" wrote:
Good info Tom, I'd like you to try -61 material, but I don't have anything
but T50 in that ferrite.

Hi Harold,

Well, I'm curious about -61 too. Turns out I have some FT50-61's (or
so claimed by Baggy Bob...), and I have FT50-67, FT50-68, FT50-43,
T50-1, T50-2, T50-3 and T50-7 to try for comparison. Unfortunately,
we're out of T50-6, but maybe I can arrange a test on a different -6
against the same size -2, at least. Not sure if I'll get to the coil-
winding and testing today, though.

Cheers,
Tom

Hi Harold,

Well, I'm curious about -61 too. Turns out I have some FT50-61's (or
so claimed by Baggy Bob...), and I have FT50-67, FT50-68, FT50-43,
T50-1, T50-2, T50-3 and T50-7 to try for comparison. Unfortunately,
we're out of T50-6, but maybe I can arrange a test on a different -6
against the same size -2, at least. Not sure if I'll get to the coil-
winding and testing today, though.

Cheers,
Tom

Well, if you're going to put it off, I can send you a few T50-6's. Probably
some -10 material too. BUT! I want your results when you get them! If you
have them in a filter, would need to know the operating Q as well, but if
just the toroid itself, wonderful to have.

W4ZCB

On Apr 29, 8:58 am, "Harold E. Johnson" wrote:
Hi Harold,

Well, I'm curious about -61 too. Turns out I have some FT50-61's (or
so claimed by Baggy Bob...), and I have FT50-67, FT50-68, FT50-43,
T50-1, T50-2, T50-3 and T50-7 to try for comparison. Unfortunately,
we're out of T50-6, but maybe I can arrange a test on a different -6
against the same size -2, at least. Not sure if I'll get to the coil-
winding and testing today, though.

Cheers,
Tom

Well, if you're going to put it off, I can send you a few T50-6's. Probably
some -10 material too. BUT! I want your results when you get them! If you
have them in a filter, would need to know the operating Q as well, but if
just the toroid itself, wonderful to have.

W4ZCB

Sure, I'll test 'em. I'll go ahead and test what I have, and add the
-6 in when I get them. I have the test setup documented well enough
that it will be easy to repeat the same tests later. Of course, the
test is always with some particular test setup. Even if I just
connected the toroid shunt across a 50 ohm line with source at one end
and analyzer at the other, it's still a particular environment, and
the results will be different if, for example, I add in my 1MHz
bandpass filter at the generator output (to kill generator harmonics)
and add my 1.5MHz HPF at the analyzer input (to kill the 1MHz so the
analyzer can be set to a lower range without contributing significant
distortion of its own). Each of those filters changes the source or
load impedance. I suppose I could have made diplexing type filters
that maintain a constant impedance vs frequency, but these guys are
big enough as it is, using high Q air core coils.

Anyway, the particular environment I chose for the coil test is in a
simple high-pass filter, nominally third order 0.1dB ripple Chebychev
with a 1MHz -3dB point: 2.2nF series, 5uH shunt (coil being tested),
2.2nF series. The distortion products I'm reporting are all 2MHz and
above, where the output capacitor passes things with very little
attenuation, and the excitation voltage across the inductor for a 1MHz
input is about 1.7dB higher than the 50-ohm source indicated output.
I suppose I could make the test more interesting by exciting a high
circulating current in the inductor, but it seems like if I excite
them in the test circuit to where distortion is clearly measurable,
the relative results be a good indication of relative performance in
other circuits.

Cheers,
Tom

Did you look at the B-H curves for the material you're using? Any magnetic
material will be non-linear to some degree. Normally the lower the peak flux
density you operate the material at the more linear the B-H curve. Generally
ferrite will have lower losses compared to powered iron. However powered
iron has more tolerance to small DC bias in the core without saturating the
material. Which brings up another point you should not operate the core with
a DC current flowing through the winding if you can avoid it. The best you
can do is find the combination that provides the inductance you need while
operating the core material at the lowest flux level you can get away with
using which tends to suggest using a high permeability material using the
least number of turns to get the needed inductance. The flux density is
proportional to the number of turns on the core and the current in the wire
(ampere-turns which is the "B" field on the B-H curves). As you noticed the
higher permeability material seems to work better, that 's what I would have
expected. That's my 2 cents worth of advice.

--
Regards,
Leland C. Scott
KC8LDO

"K7ITM" wrote in message
...
I have an interest in making RF filters with low distortion (for
example, as preselection filters for receivers). I've found that very
often I need to use air-core coils to get the low distortion I want,
but it would be convenient if I could use powdered iron or ferrite
core toroids to get higher Q at low frequencies without having to make
the coils physically big. I've always assumed that powdered iron
would give me the best results, but someone suggested I try low-
permeability ferrite. I just did some experiments with that. I wound
three toroids, all 0.37 inch OD cores; one is on mix 2 powdered iron,
and the other two are on type 68 and type 67 ferrite. I put each
coil in a simple 1MHz high pass filter: two 2.2nF caps in series,
with the coil connected between them to ground. I tested for harmonic
distortion with a 1MHz input, and for the 2.42MHz intermod product
with 0.62MHz and 1.52Mhz inputs.

With about +16dBm coming out of the test filter at 1MHz, I measured
second harmonic at about -95dBc for the T37-2, -90dBc for the FT37-67
and -57 for the FT37-68; the third was at -74dBc, -61dBc and -51dBc
for the three coils in the same order.

The 2.42MHz intermod, with each tone at +18.5dBm out of the combiner,
measured -71dBc, -58dBc and -38dBc for the coils, again in the same
order.

My conclusion from this is that at least these two ferrites aren't
going to help me in my search for a better low-distortion core
material, but at the same time, it was interesting to note that the
higher permeability -67 performed better by far than the lower
permeability -68. It gives me some hope that if the ferrite materials
guys went looking for a formulation that would be specifically as low-
distortion as they could get, they might find something pretty good.

Before someone asks, no, I didn't run a T37-6 core coil along with
these. I expect it would be a little better than the T37-2. And I've
measured a few coils wound on (unknown formulation) powdered iron rods
that show significantly lower distortion than those on toroid cores;
but of course those have a large "air gap" and relatively high
external field.

Cheers,
Tom

On May 3, 9:17 pm, "Leland C. Scott" wrote:
Did you look at the B-H curves for the material you're using? Any magnetic
material will be non-linear to some degree. Normally the lower the peak flux
density you operate the material at the more linear the B-H curve. Generally
ferrite will have lower losses compared to powered iron. However powered
iron has more tolerance to small DC bias in the core without saturating the
material. Which brings up another point you should not operate the core with
a DC current flowing through the winding if you can avoid it. The best you
can do is find the combination that provides the inductance you need while
operating the core material at the lowest flux level you can get away with
using which tends to suggest using a high permeability material using the
least number of turns to get the needed inductance. The flux density is
proportional to the number of turns on the core and the current in the wire
(ampere-turns which is the "B" field on the B-H curves). As you noticed the
higher permeability material seems to work better, that 's what I would have
expected. That's my 2 cents worth of advice.

--

I had no particular interest in looking at the B-H curves, since it's
distortion in filters I'm interested in, and that's easy for me to
test directly.

With respect to flux density: there seems to be a bit of a disconnect
between what I know and what you've written. From Faraday's law of
magnetic induction, I expect that for the same inductance with fewer
turns, the flux density must be higher: if the inductance is the
same, the voltage must be the same for a given current at the test
frequency. The same voltage with fewer turns around the same cross-
section core implies higher flux density. No?

So far, my tests suggest that you can't draw any conclusions one way
or the other from only the relative permeability of the core
material. Ideally, you should do tests in the actual application. My
testing is slightly derailed (I hope temporarily) because the cores I
got that were supposed to be FT50-67 are clearly something different,
almost certainly FT50-61. I am testing all the other low-to-medium
permeability (F)T-50 cores I have available at the moment.

Cheers,
Tom

"K7ITM" wrote in message
...

I had no particular interest in looking at the B-H curves, since it's
distortion in filters I'm interested in, and that's easy for me to
test directly.

Looking at the B-H curves will save you the trouble of testing core material
that is obviously non-linear. If the B-H curve is perfectly linear there
would be no distortion in the signals. Some material from just a quick look
at the curves would eliminate them from consideration without the time
wasted testing them.

http://www.ferroxcube.com/appl/info/3B46.pdf


With respect to flux density: there seems to be a bit of a disconnect
between what I know and what you've written. From Faraday's law of
magnetic induction, I expect that for the same inductance with fewer
turns, the flux density must be higher: if the inductance is the
same, the voltage must be the same for a given current at the test
frequency. The same voltage with fewer turns around the same cross-
section core implies higher flux density. No?

With all things being equal that's true, however not all core material
reacts the same. As you can see from the link above things are a bit more
complex than they at first appear. Somebody could do a PhD thesis on the
subject of designing low distortion magnetic components.

http://info.ee.surrey.ac.uk/Workshop/advice/coils/mu/


So far, my tests suggest that you can't draw any conclusions one way
or the other from only the relative permeability of the core
material.

I think it's worth another look myself.


--
Regards,
Leland C. Scott
KC8LDO