Message	Date	From	Subject
25	2006-07-16 15:49:15	Wes Hayward	Toroid Power Rating
Hi all, A question and several comments have been posted regarding the "power rating" of toroids. An abbreviated and far from complete "answer" is "Yes, we have to be concerned about power dissipation." I'll illustrate this trite comment with examples. But to say that a T50-x core is good up to 49 Watts and that we must then go to a T106 or T130-x core is not a viable answer. It all depends upon the application. It is often interesting to see what the commercial guys do. This can be difficult with the imported transceivers, but Elecraft is quite open with their designs. (Bravo to Wayne for this policy!) I took a look at the amplifier in the 100 Watt version of the K2 and noted that they build their low pass filters with T50- x cores. The low bands use -2 material while the high bands use T50-10 cores. It is easy enough to do some modeling. First, we take a core of interest and measure the Q. From this, we can model the inductor as an ideal L in series with a resistance. We then drop the model into a low pass filter circuit that we place in a SPICE simulation. A 100 Watt source consists of a 200 volt generator with a 50 Ohm source resistance. Remember that the voltages and currents in SPICE are all peak values. This generator will deliver 100 W to a 50 Ohm resistor attached as a load. I did a sample calculation. I designed a 0.1 dB ripple Chebyshev low pass filter with a 3 dB 7.5 MHz cutoff frequency with LowHi, the filter design tool that is provided with EMRFD that designs low and high pass filters. Any old program will do the job. Indeed, no program is needed for those with an urge to go back to a text. My filter was then placed between the source and load in the SPICE file described above. Results were interesting. I assumed a Q of 200 for the inductors, so the end parts had a series R of 0.34 Ohm while the center one had R of 0.42 Ohm. A sweep yielded a higher current (at 7 MHz) in the middle inductor than in the input one. The power dissipated in the input inductor was 2 Watts while that in the middle L was 3 Watts. The overall filter loss was 0.38 dB at 7 MHz. I suspect that these inductors wound on T50-x cores could run for extended periods while dissipating the calculated powers so long as they were not confined in an air tight space. Reduced duty cycles will always help. But going beyond the 100 W transmitter power level will produce greater stress. Clearly, the KW level is too much for these small cores. Conversely, we can do anything at the 10 W level and never encounter a problem. Different filters will yield new results. For example, we could easily build a single resonator LC bandpass filter at 7 MHz that had a 3 dB insertion loss. This circuit would have a high Q, perhaps 100 or more. Such a filter would be selective enough to be very useful in many applications. But this filter would dissipate 50 W if placed in the output of our 100 W transmitter. If we dropped the Q to 10 or even less, the loss becomes reasonable. Circuit details can have interesting and sometimes unsuspected consequences. For example, the low pass filters used in that 100 Watt K2 transmitter mentioned earlier are elliptic designs. These low pass filters differ from the simple ones with nothing but series inductors and shunt capacitors; capacitors are placed in parallel with the inductors. The parallel element creates a "trap" that does a lot to improve the attenuation of some low order harmonics. But the traps tend to increase the inductor current, especially as operation moves close to the filter cutoff. This will increase filter loss and dissipation in the trapped inductors. Raw power dissipation is but one criterion. Another is the allowed inductor current. Higher current within an inductor will increase the magnetic field, eventually moving toward a nonlinear relation between B and H. The nonlinearity will be small, but it is certainly present. You can detect this with a simple experiment. Build a simple LC oscillator and put it in an aluminum box with the inductor mounted on the wall of the box. You can do this without major problems with a toroid. Listen to the oscillator in a receiver as you move your hand close to the box. There should be no frequency change. Repeat the experiment with a small magnet and note that the frequency does shift a little. The stronger the magnet, the greater the shift. This will not happen if the oscillator had been built with an air core inductor. This qualitative experiment can be formalized by injecting a DC current into the inductor. RF chokes would be needed to isolate the DC source from the oscillator. The nonlinear behavior can show up in a variety of ways. The oscillator experiment mentioned is one that can cause hum in a VFO. (I got a T5 report one time, back in an era when people gave honest reports on CW. I moved the power supply and all was fine.) A more interesting situation was the observation of IMD in a 300 kHz wide LC bandpass filter operating at 10 MHz in a spectrum analyzer IF. The filter performance with a single tone was wonderful, but it was producing intermodulation distortion when two signals were present. The input intercept was around +20 dBm or so. Replacing the T37-6 toroid cores with T50-6 cores wound with the same inductance moved the third order IMD intercept up to +50 dBm, at the edge of what we could measure at the time. Bottom line: There is no hard and fast rule that tells you what core will be needed. Rather, specific measurements, perhaps coupled with calculations of one sort or another will provide the science needed to suggest answers. 73, Wes, w7zoi

Hi all,

A question and several comments have been posted regarding
the "power rating" of toroids. An abbreviated and far from
complete "answer" is "Yes, we have to be concerned about power
dissipation." I'll illustrate this trite comment with examples.
But to say that a T50-x core is good up to 49 Watts and that we must
then go to a T106 or T130-x core is not a viable answer. It all
depends upon the application.

It is often interesting to see what the commercial guys do. This
can be difficult with the imported transceivers, but Elecraft is
quite open with their designs. (Bravo to Wayne for this
policy!) I took a look at the amplifier in the 100 Watt version
of the K2 and noted that they build their low pass filters with T50-
x cores. The low bands use -2 material while the high bands use
T50-10 cores.

It is easy enough to do some modeling. First, we take a core of
interest and measure the Q. From this, we can model the inductor
as an ideal L in series with a resistance. We then drop the model
into a low pass filter circuit that we place in a SPICE
simulation. A 100 Watt source consists of a 200 volt generator
with a 50 Ohm source resistance. Remember that the voltages and
currents in SPICE are all peak values. This generator will deliver
100 W to a 50 Ohm resistor attached as a load.

I did a sample calculation. I designed a 0.1 dB ripple Chebyshev
low pass filter with a 3 dB 7.5 MHz cutoff frequency with LowHi, the
filter design tool that is provided with EMRFD that designs low and
high pass filters. Any old program will do the job. Indeed, no
program is needed for those with an urge to go back to a text. My
filter was then placed between the source and load in the SPICE file
described above. Results were interesting. I assumed a Q of 200
for the inductors, so the end parts had a series R of 0.34 Ohm while
the center one had R of 0.42 Ohm. A sweep yielded a higher current
(at 7 MHz) in the middle inductor than in the input one. The
power dissipated in the input inductor was 2 Watts while that in the
middle L was 3 Watts. The overall filter loss was 0.38 dB at 7 MHz.

I suspect that these inductors wound on T50-x cores could run for
extended periods while dissipating the calculated powers so long as
they were not confined in an air tight space. Reduced duty cycles
will always help. But going beyond the 100 W transmitter power
level will produce greater stress. Clearly, the KW level is too
much for these small cores. Conversely, we can do anything at
the 10 W level and never encounter a problem.

Different filters will yield new results. For example, we could
easily build a single resonator LC bandpass filter at 7 MHz that had
a 3 dB insertion loss. This circuit would have a high Q, perhaps
100 or more. Such a filter would be selective enough to be very
useful in many applications. But this filter would dissipate 50 W
if placed in the output of our 100 W transmitter. If we dropped
the Q to 10 or even less, the loss becomes reasonable.

Circuit details can have interesting and sometimes unsuspected
consequences. For example, the low pass filters used in that 100
Watt K2 transmitter mentioned earlier are elliptic designs. These
low pass filters differ from the simple ones with nothing but series
inductors and shunt capacitors; capacitors are placed in parallel
with the inductors. The parallel element creates a "trap" that
does a lot to improve the attenuation of some low order harmonics.
But the traps tend to increase the inductor current, especially as
operation moves close to the filter cutoff. This will increase
filter loss and dissipation in the trapped inductors.

Raw power dissipation is but one criterion. Another is the allowed
inductor current. Higher current within an inductor will increase
the magnetic field, eventually moving toward a nonlinear relation
between B and H. The nonlinearity will be small, but it is
certainly present. You can detect this with a simple
experiment. Build a simple LC oscillator and put it in an
aluminum box with the inductor mounted on the wall of the box.
You can do this without major problems with a toroid. Listen to
the oscillator in a receiver as you move your hand close to the
box. There should be no frequency change. Repeat the
experiment with a small magnet and note that the frequency does
shift a little. The stronger the magnet, the greater the
shift. This will not happen if the oscillator had been built with
an air core inductor.

This qualitative experiment can be formalized by injecting a DC
current into the inductor. RF chokes would be needed to isolate
the DC source from the oscillator.

The nonlinear behavior can show up in a variety of ways. The
oscillator experiment mentioned is one that can cause hum in a
VFO. (I got a T5 report one time, back in an era when people gave
honest reports on CW. I moved the power supply and all was
fine.) A more interesting situation was the observation of IMD in
a 300 kHz wide LC bandpass filter operating at 10 MHz in a spectrum
analyzer IF. The filter performance with a single tone was
wonderful, but it was producing intermodulation distortion when two
signals were present. The input intercept was around +20 dBm or
so. Replacing the T37-6 toroid cores with T50-6 cores wound with
the same inductance moved the third order IMD intercept up to +50
dBm, at the edge of what we could measure at the time.

Bottom line: There is no hard and fast rule that tells you what
core will be needed. Rather, specific measurements, perhaps
coupled with calculations of one sort or another will provide the
science needed to suggest answers.

73, Wes, w7zoi