EMRFD Message Archive 7014

Message Date From Subject
7014 2011-12-14 04:41:33 Ben Pharr Core Loss in High Frequency Inductors
I asked a question over on qrp-l and someone suggested I ask over
here. I'll try to keep it short.

I'm building an L-network tuner for a 1/2 wave 17m end fed antenna.
TLW suggested I needed an inductor of 4.07uH. It also said there'd be
293V RMS through the inductor at 10W, or 927V at 100W. Different
impedance values give slightly different values for inductance and
voltage, but there's not a substantial difference.

I was curious about the loss of the inductor, so I found some
equations and specifications on the Micrometals site and put them in
Excel. It showed 1.85 dB loss at 18.2MHz for a T200-6 core. I wasn't
sure my numbers were right, but I found an inductor calculator
(http://www.dl5swb.de/html/mini_ring_core_calculator.htm) that
confirmed them. That's pretty inefficient at 10W and just plain
wouldn't work at 100W.

Does this match up with anyone's experience? Do these numbers sound
right, or did I make an incorrect assumption at some point?

A T130-17 core does slightly better, with "only" 0.72dB (15%) loss,
but that's still annoyingly inefficient and won't work at 100W power
levels.

An air core coil is a possibility, but I had been wanting to keep this
tuner small, is it's mostly for backpacking. (It'd also be nice to be
able to use it for my club's Field Day effort, which would be at
100W.) Fitting an air core coil in a small enclosure (3"x2"x1") would
require putting the turns really close, which would bring into
question its voltage rating. I haven't been able to find a typical
voltage rating for magnet wire.

So, to summarize my questions, do my calculations seem reasonable? If
so, is there a way to build an efficient toroidal inductor for the
higher HF frequencies? If not, are there some good methods for
building small, high voltage air core coils? Like somehow adding
additional insulation to the wire?

Thanks,
Ben
WF5N
7016 2011-12-14 06:42:43 Bob Fish Re: Core Loss in High Frequency Inductors
Hi Ben,

I can't help you with the core loss issue, Someone else will chime in I'm sure. But I always thought the best way to match an end fed half wave antenna is with a tank circuit. Where you match the high impedance it presents by putting the tank circuit between the antenna and ground and feeding the inductor a few turns up from ground (the 50 ohm point).

Bob. K6GGO


Sent from my iPad

7017 2011-12-14 07:47:45 Ashhar Farhan Re: Core Loss in High Frequency Inductors
Actually, what will work the best is a pi. By keeping either of the
two capacitors at a minimum, you can make it act like an inverted-L
network facing either of the two directions. A few taps on the
inductor will make it far more flexible.
The old timers had got it right with the pi networks at the 807 outputs.
- farhan

On 12/14/11, Bob Fish <rwfish@comcast.net> wrote:
> Hi Ben,
>
> I can't help you with the core loss issue, Someone else will chime in I'm
> sure. But I always thought the best way to match an end fed half wave
> antenna is with a tank circuit. Where you match the high impedance it
> presents by putting the tank circuit between the antenna and ground and
> feeding the inductor a few turns up from ground (the 50 ohm point).
>
> Bob. K6GGO
>
>
> Sent from my iPad
>
>
7018 2011-12-14 07:58:21 Tim Re: Core Loss in High Frequency Inductors
My feeling (literally, with my fingers) of -6 core inductors in high circulating current matching circuits, is that you will get additional loss in the copper.

Yes I would well expect the inductor to run very warm at 100W.

If you can reduce the inductance and still get a match, that would reduce losses and circulating current/voltages.

My other recollection, that of smell, when using high circulating current matching at the 100W level: RF burns are very easy to come by. RF burns smell a lot like chicken.

I'm not an expert in end-fed half wave matching but Google turns up a couple hits where they use a toroid (-6 mix) in a tuned circuit as kind of a step-up transformer. That seems like the right idea to me, e.g. http://www.swlink.net/~w5jh/efhwa_at.htm

Tim N3QE

7020 2011-12-14 10:54:34 William Carver Re: Core Loss in High Frequency Inductors
Ben, I've been doing an antenna tuner design using the ARRL's program
"AAT.EXE" to estimate losses in a "T" tuner. An L network isn't very
flexible as a general purpose tuner, but for feeding high impedances
it's great and has the lowest loss.

I've been winding coils and measuring their Q to estimate the tuner
loss, so your question hits dead center where I am at, even though it's
a different tuner circuit.

So after some calculating I wound 17 turns of #12 over about 60% of the
periphery of a T200-6 core. It measures 4.1 uH at 1 MHz. At 18.1 MHz my
Boonton 260A Q meter gives an indicated Q of 430. Correcting for the .02
ohm internal resistance, Q=438.

I didn't correct for distributed capacitance of the coil itself.

Q measurements aren't that precise: lets be conservative and use Q=400.
That can be modeled as a series resistance of 1.16 ohms at 18.1 MHz.

Assuming your voltage number is right, for 10 watts your L network power
loss will be 0.46 watts. That's 0.2 dB. That's inconsequential given
that one S unit is 6 dB. At 100 watts the coil will get a little warm,
but just a little. It would get toasty (maybe incandescent) at 1 KW.

On some receiver filters I found that -10 (black) material produced
higher Q that -6 at this frequency. So my 20-17-15-12-10 meter filters
all use black cores. Unfortunately I don't think a T200-10 core is
available.

But Q=400 is nothing to sniff at, and the power loss won't create a
problem even up to the half KW level. That said, you might ask "what if
I use a T130-6 core? Or even smaller. Sorry, don't have any of those
sitting around.

W7AAZ
7021 2011-12-14 10:54:58 Tamás Fábián Re: Core Loss in High Frequency Inductors
I use a parallel tuned LC tank with a tapped, air core L with short and
half wave antennas (and often with antennas in between, see
http://files.qrz.com/l/ha5ftl/DSCF0166.JPG)

It is very backpackable, even in this large and fragile casing.

And end-fed antenna shows a great ohmic impedance, and this impedance can
be stepped down with the coil, using it as an autotransformer. The
capacitor is plugged in close to the ground, to form an LC tank and
therefore and easily adjustable impedance.

The variable capacitor is not designed to withstand high voltages, so care
must be taken not to connect it "too high" or else it might arc. Connecting
the capacitor too "low" one must set it near to its highest value,
resulting in large capacitance and large circulating current, adversely
affecting efficiency.

This is the type of tuner that is especially good with loads other tuner's
can't even bring close to an SWR that makes any kind of sense: short
antennas (with small radiation resistance, and large capacitive reactance),
and (n times) half wave antennas with large (radiation related) resistance,
and maybe some small reactance. It is also possible to match other
antennas, especially if those can slightly be tuned other ways (clipping or
bending for example). I recommend building it, since it is very useful, and
one can learn a great deal experimenting with it.

7022 2011-12-14 10:59:06 William Carver Re: Core Loss in High Frequency Inductors
> If you can reduce the inductance and still get a match, that would
> reduce losses and circulating current/voltages.

That is true with a "T" network. But in the limit the lowest loss T
degenerates into an L network. I believe Ben's choice of L network is
already the lowest Q/lowest loss.

> My other recollection, that of smell, when using high circulating
> current matching at the 100W level: RF burns are very easy to come by.
> RF burns smell a lot like chicken.

I agree 100%. It DOES smell. And RF burns are vertical, not just
surface, and they take a long time to heal up.

W7AAZ
7024 2011-12-14 19:15:03 kb1gmx Re: Core Loss in High Frequency Inductors
My experience..

The L network is the simplest and lowest loss. It is also the
easiest to tune.

I use L networks to match the typically high end impedance for a half wave. The usual value I use is around 3300 ohms for 40m and below
and 4200 (or 4700) ohms for the higher bands where the antenna is typically higher relative to wavelength and it's loaded impedance is generally higher.

For 50W limit I use a T80-2 for the lower bands and the t94-6 for the higher. For 10 watts and under I stack up two T50-2 or -6. Heaviest wire I've used is #20. Never had an issue with heating. The cap used must stand 2x to 4x the total voltage and my favorite cap is a length of RG316 for lower power and a length of UT141 or RG142 for higher power (higher breakdown voltage). If you have them High voltage doorknob caps are good (5KV or higher). If using a variable cap
you need wide spaced if pweor is over 20W and real wide for more than 100. For 4Uh close wound #16 or heavier is fine as the voltage gradient from turn to turn is small it's the high impedance end that may need the last few turns double spaced.

A T130 or larger is good for 100W and the unloaded Q is fairly meaningless as you want a loaded Q of under 10.

Tuning the network. Use a resistor and low power or a MFJ or other analyzer. Adjust the network for 1:1 (1.2:1 or lower) into that resistor for the middle of the band segment of interest. For 17M
I'd use 4200 ohms. Then take it outdoors and attach a wire that will be at the operating height or at least close and trim the wire for minimum SWR without touching the matching network.

Also if you backpacking I'd build a low power version good for say 25W and small and light. Unless your packing a Size 34 battery (about 53 pounds) to run 100w, low power (5-10W) is more likely.

For those that wonder at 40M using a T114-2 core A friend runs 400W
and doorknob caps for the L network. An earlier version using a
T94-2 and RG316 failed after a few weeks of use due to the RG316
(used as a cap) arcing over. The core and the enameled wire was
fine.

In the field the antenna can be used vertically, Inverted L, sloper, Inverted V or as a common flat top. Tuning of the wire will vary a small amount based
7025 2011-12-14 20:16:47 William Carver Re: Core Loss in High Frequency Inductors
On Thu, 2011-12-15 at 03:14 +0000, kb1gmx wrote:
>
> My experience..
>
> The L network is the simplest and lowest loss. It is also the
> easiest to tune.
>
> I use L networks to match the typically high end impedance for a half
> wave. The usual value I use is around 3300 ohms for 40m and below
> and 4200 (or 4700) ohms for the higher bands where the antenna is
> typically higher relative to wavelength and it's loaded impedance is
> generally higher.

To get a perfect 1:1 match both the L and C need to be adjustable. When
it's set up for that perfect match the L network has a loaded Q equal to
the square root of the antenna resistance divided by (typically) 50
ohms.

So if the antenna resistance WAS 3300 ohms, the loaded Q would be the
square root of 3300 divided by fifty, or the square root of 66, which is
8.12. As Allison says, that's much lower than the Q of the coil and
capacitor so losses are low.

If the antenna is 4700 ohms then the loaded Q is almost ten: you're
inching up closer to the unloaded coil Q and losses climb a little high.

"Low" loss is a relative thing: losses of 5% when you're QRP are also
QRP. Losses of 5% when you're running 500 watts will make a very warm
toroid, and at 1500 watts you're probably going to need a coil of copper
tubing in a big box. And a 25% increase with a loaded Q of ten gets
pretty significant at high power levels.

I read the newly announced Alpha 4040 tuner has a fan to cool the roller
coil.

I have a 375 pF 12 KV vacuum capacitor that I've used in multiple
tuners: it's a "goodie" I hang onto just in case........

Bill
7027 2011-12-15 14:37:19 kb1gmx Re: Core Loss in High Frequency Inductors
7098 2011-12-22 18:36:44 Ben Pharr Core Loss in High Frequency Inductors
My apologies for not following up on this sooner. I asked my question
on both qrp-l and the EMRFD list and got lots of helpful replies. My
apologies for not replying to each one individually.

In case you don't remember, my question was, briefly, "should losses
through a ~4uH toroidal inductor really be on the order of 1.85dB at
18MHz?" My numbers were based on equations I found on the Micrometals
site and confirmed by an inductor calculator one of you found online.
My application is an L-network tuner for an end-fed 17m 1/2 wave
antenna.

Responses were mixed. Several of you responded that those numbers
seemed reasonable. A roughly equal number of you responded with
anecdotal evidence to the contrary.

W7AAZ actually had a T200-6 core on hand, put 17 turns on it, and
measured a Q of roughly 400. Based on that, ARRL's antenna tuner
design program shows there would be 0.2dB of loss through the
inductor.

I have no reason to doubt W7AAZ's numbers, but I also have no reason
to doubt the calculations I did. So the only thing left for me to do
is build some inductors and do some measurements. I have some -6 and
-17 cores on hand, and I think I have everything I need to make some
meaningful measurements.

I'd also like to experiment with a fixed-value version of the
link-coupled approach advocated by AA5TB and others. Several people
seem to think it offers lower loss than a standard L-network.

I'll be sure to share my results when I'm finished, but it'll probably
be a couple of weeks. My tinkering time is limited these days, and
this is not the only project I've got going.

Thanks again to everyone for their help.

Thanks,
Ben
WF5N