Yesterday, I put the VNA away and invoked the signal generator (HP 8648C)
and an o'scope (Agilent DS06104A). I specifically looked at the amplitude
balance and phase relationship on the DM side of the chokes referenced the
CM side from 1.8 MHz through 30 MHz. Most did OK with the amplitude
balance. Most were good to excellent with phase skew and a few were not so
great. I'm updating my table to reflect those observations.
Addressing only the 2-stack 400-31 CMC: Even though the CM impedance was
lower than the others across from 1.8 through 30 MHz, it proved to be a
stellar performer in both amplitude balance and phase skew at the DM side
of the CMC. The 22 turns on that CMC are spaced about the width of one
bifilar winding of two parallel AWG #12 enamelled copper wires so the
windings are not close spaced. Only on 14.2 MHz did I detect a very slight
phase skew on the DM side. It was by far the best of all the CMCs I've
constructed and measured in that respect. In my application, except for
7.15 MHz, it should prove more than adequate wrt CM impedance. W/C on 7.00
MHz in the shack, the open wire feeders measure 1160 - j 1110. The rest of
the bands all come in less than 250 ¡À j 300 in the shack. Most are in the
range of ¡Ü 150 ¡À j 175. Measurements of the impedance presented in the
shack by the antenna/feedline were made with a small VNA with nothing else
connected to it other than the VNA itself -not even my hands.
Still interesting is the 2-stack 240-31 CMC bifilar wound with #12
enamelled copper wire. I managed (crammed) 18 bifilar turns on that
stack. The windings are certainly close with really no spacing between
windings on the inside diameter of the core going against 'common
knowledge' that they should be spaced for lower distributed C of the
assembly. Believe me, it was a challenge to get 18 close-spaced bifilar
windings on that core stack!! That CMC is the absolute best for DM
impedance and amplitude balance and lack of phase skew from 1.8 through 30
MHz.
As a final CMC, yesterday I also wound a 2-stack 240-31 with 12 spaced
bifilar turns of #12 solid enamelled wire. In comparing it to the CMC
wound in similar manner with stranded #14 wire, it resulted in about the
same CM impedance but far less phase skew than the stranded wire
'equivalent. It also faired excellent with the amplitude balance and lack
of phase skew on the DM side.
Well, enough babbling on my part. All this will be reflected in my final
table which I have yet to update.
Again, I have no idea, after winding all these CMCs, how K9YC managed so
many bifilar turns on the 240 cores without overlapping as he presents in
his "Choke Cookbook". Maybe he didn't wind bifilar chokes but, rather,
coax cable on the cores which does not constitute a true CMC???? There is
no way I could cram any more than 18 bifilar turns on the 240-31
stack!!!!! And the CM input of that choke is right next to the DM "output"
side (these are all bilateral devices).
Dave - W?LEV
PS: Winding the 240-31 cores with solid copper wire is a major challenge
and if you attempt it, you WILL work up a minor sweat.
On Mon, Jan 25, 2021 at 4:02 PM Manfred Mornhinweg <manfred@...>
wrote:
Dave,
Yesterday I got quite a surprise in winding the 4" 31 material cores. I
stacked the cores so the external dimensions of the core cross section
are
0.5 x 2.0" (1.27 x 5.08 cm). I wound 22 bifilar turns,
Too many!
no crossovers or
twists, on the stacked cores using AWG #12 enamelled solid copper wire.
I
did my best to keep the conductors of the transmission line on the core
in
parallel fashion and close to each other (the pair). When measured, the
CM
impedance was actually less than that of the many cores I've wound in
similar manner of 31 material and the 2.4" OD.
Because you get too much stray capacitance. Probably that choke with 22
turns on a huge toroid stack has so much capacitance that the capacitance
dominates over the entire spectrum, or most of it.
The reason why I think that such a large core could be good is because it
allows you to increase the space between turns. Don't fill it with as many
turns as you can! Instead optimize the number of turns to get the best
possible compromise between capacitance and inductance+resistance, for the
bands of interest.
In addition, the DM losses
were greater.
You have a long parallel line in close contact with lossy ferrite. Some of
the field between the two wires is in that ferrite. Maybe you could reduce
this loss by wrapping some inert material on the core, before winding the
wire. Some sort of thick adhesive tape. Maybe applying the tape just along
the edges, so that most of the wire ends up in the air,, somewhat separated
from the ferrite.
So,.....I spent the next couple of hours with my supply of
nylon cable ties tightening and closing the gaps between the two
conductors
of the transmission line on the core. Once I had a good tight pair of
conductors around the core, I remeasured. The CM impedance didn't change
much, but the DM mode losses more than halved.
That seems to confirm that part of the losses are due to the ferrite
capturing a significant part of the magnetic field between the two wires.
Closing the space between them reduces the amount of that field in the
ferrite.
Also consider just the length of that line, and the fact that unless its
impedance matches the load, there will be standing waves on it, leading to
parts of it having higher loss than others. When you compare chokes with
longer or shorter wires wound on them, it's quite possible that the choke
with the shorter wire doesn't have a high-loss point somewhere along the
wires, but the one with the longer winding has.
The windings (the conductor
pairs) are spaced generally 2 to 3 conductor pairs apart.
That's not bad at all.
My question to myself: Is it possible to put too many turns on the
cores?
Definitely yes. There is always an optimum, which depends on the band of
interest, or range of bands. On a large core like that, made from high
permeability material, you could easily have so many turns that its peak
performance happens below the lowest ham band!
Another bit of a surprise: I built a small piece of test equipment to
measure the current in a relative manner going out the balanced line
transmission line to my set of wires (450-foot long doublet). Those CMCs
which measured a relatively large CM impedance (6 to 15 k) did not show
good balance between the DM feed to the balanced line.
Check how much of the CM impedance is resistive, and how much reactive. I
suspect that most might be reactive, more specifically capacitive. That can
resonate with anet inductive common mode impedance of the feedline+antenna,
and result in poorer performance than expected.
Even more
interesting, the measured CM impedance did not correlate to well with the
measured balance between the DM pairs.
That's the typical symptom of resonances.
Be aware that such resonances can happen both with the antenna system, and
the measuring equipment. Proper calibration should take it out of the
latter, but I'm running myself into situations where it doesn't, at least
with the NanoVNA, when measuring through a coax cable of any significant
length. After all the NanoVNA is natively a 50? instrument, and when we
measure CMCs on it having impedances of several k?, it's working with tiny
signals that are prone to large errors. With your big VNA this should be
better, but not perfect.
When I built the little piece of
equipment, I checked its ability to measure good balance by loading it
with
a non-inductive 50-ohm resistor. It showed good balance between both
sides
of the DM load. So,......the home brew piece of equipment is measuring
properly.
I don't know... There are so many possible pitfalls. Keep in mind that a
stray capacitance of just 5pF is a reactance of around 1k? at 30MHz. With a
CMC having 10k?, a stray capacitance of 5pF in your measurement setup will
totally mess up measurements even at 3MHz!
RF work is easy between about 30 and 200?. Below that range stray
inductance is nasty, and above that range stray capacitance is equally
nasty. Your CMC is very high up in the domain of mischievous stray
capacitance!
Try shooting for lower inductance/resistance, and much lower stray
capacitance, by winding fewer turns. With two FT-400-31, I would try 10
turns. The pair close-spaced, the turns spread around two thirds of the
core, the input well separated from the output. And sticking some separator
(cardboard with double-sided tape, for example) over the edges of the core
before winding the wire, to keep the wire a little separated from the
ferrite.
10 turns won't give you fabulous impedance on the lowest bands, but might
be a better compromise over the entire HF range. Fewer turns will probably
improve performance on the highest bands.
If I got this right, one FT-400 has the same cross sectional area as two
FT-240, at nearly twice the path length. So, a single FT-400 wound with the
same number of turns as two FT-240, of the same material, should give you
only 55% of the inductance, at very low frequencies. Up to what frequency
this ratio is maintained, is something I don't dare to predict...
So, with two FT-400 you should get slightly more inductance at VLF than
with the same number of turns on two FT-240. Again, I don't dare to predict
what happens at higher frequencies. The turns are longer, so more stray
capacitance. But they are much more spaced, so much less stray
capacitance... And while Amidon and Fair-Rite say that material 31 is free
from dimensional resonances, I don't know if this holds true even with such
large cores.
So, it's really something to experiment with!
Manfred
--
*Dave - W?LEV*
*Just Let Darwin Work*
