Tests on Radio shack 100 uH coils


From: mike czuhajewski (wa8mcq@abs.net)
Date: Mon Feb 09 1998 - 02:15:20 EST

A little while back KI6DS asked me to do some tests on those 100 uH
chokes that Radio Shack sells. They are about a quarter inch in
diameter and about 1 1/3" long and have two layers of wire. This was
in conjunction with the redesign of the DeMaw Tuna Tin 2 using
modern components; the original (as well as the subsequent Sardine
Sender?) used some of these coils with turns removed and he wanted
to know how they performed. (The redesign will be using toroids
instead but he was curious.)

BTW, the forms are what is believed to be roughly equivalent to type
61 ferrite material, and one thing I found was that they are somewhat
fragile. The leads are glued into holes in the ends of the
forms, and if they get too much handling they can break out of the
form. A little bit of 5 minute epoxy will fix them up if that happens.

Here's the info that I already sent to Doug. If I made errors in some
of my assumptions or techniques, or if anyone wants to add anything
about the technical topics discussed, I'd be more than glad to hear
about it.

I've been over this a little bit and touched up some of it, and it may
still be a bit rough in places, but I want to get it on the streets
it gets lost in the shuffle on my computer, so here it is, as-is.


I bought two and checked them out on an HP 8753C network analyzer.
(That's a piece of equipment that falls into the class of Dead Serious
Test Equipment That RF Engineers Sell Their Grandmothers to Get
Access To.) One was left stock, and the other was measured stock and
then with various amounts of wire peeled off. I first stripped off the
top layer, leaving just the 27 turns (or so) of the first layer; then I
checked it again at 20 and 15 turns. The bottom line is that the stock
choke really shouldn't be used on any ham band higher than 1.8 MHz
since its self resonant frequency is about 2.6 MHz, and the 15 turn
version makes a dandy choke at least as far as 14 MHz. (I had the
analyzer set for a sweep from 0.3 MHz--lowest setting--to 50 MHz, a
good practical range to check this coil over. The instrument itself will
go as high as 6 GHz.) The use of two layers wound directly and
linearly over each other is a killer. (Remember the big RF chokes in
the old days, the three pie type, and how the wire was wound in a
criss-cross fashion? That's done to reduce distributed capacitance and
it works very well indeed, as some later tests showed, detailed
somewhere below.)

Originally I typed in this paragraph, before actually trying to do it:
"If you still want me to wind 100 turns of #28 on the form as you
requested, let me know and I'll be glad to do it. I'm presuming that
was with the intent of possibly using it as a choke at RF, and it's
pretty obvious that it would make a poor one, with a self resonant
frequeny well below any possible ham band. On the other hand, it might
be good for the broadcast band or LF, but I can pretty much guarantee
that it would be nowhere near enough inductance to be useful in any
audio filtering application; the permeability of the core is far too low
for that. If you want me to do it, let me know."

After actually trying it, the results are that it doesn't turn out that
much different from the stock coil. I wound 70 turns on the core, about
all it would take in a single layer. N2CX told me the spool of wire he
gave me was #28, but according to my dial calipers and the wire charts
in the ARRL handbook it's more like #27 :-) Unlike the stock coil, I
wound those 70 turns in a single layer. The results are detailed below,
but the bottom line is that it's pretty similar to the inductance of the
stock coil but with vastly lower distributed capacitance.

The stock, unmodified RF choke, with two layers of wire, as measured
on the Boonton 260A has an "apparent inductance" (Lapp) of 114 uH
at 790 KHz (one of the standard frequencies at which the inductance
can be read directly from a calibrated dial). The Q was 37 at that
frequency. Corrected for the distributed capacitance (Cd) which was
later measured at 43 pF for the stock coil, that is reduced to an even
100 uH. (At 790 KHz it resonated with 360 pF, but the actual
resonating capacitance was 403 pF when Cd is factored in.)

The highest frequency it could be measured at was 1.865 MHz, which
is where it resonated with the tuning cap in the 260A set to the
minimum of 30 pF. That gave an apparent inductance of 242 uH,
corrected to 99.7 uH when Cd was factored in.

I'm no expert in metrology so I won't attempt to interpret or explain
the readings I got on the network analyzer at various frequencies,
except to say that the indicated inductance changes quite a bit with
frequency. For example, at 300 KHz it shows 81 uH, 92 uH at 1 MHz,
100 uH at 1.2 MHz, 120 uH at 1.5 MHz, 216 uH at 2 MHz and 400 uH
at 2.25. The important thing is the self resonant frequency, which is
something that should not be approached when using the coil, and that
is a disappointingly low 2.5 or 2.6 MHz. Above that, the coil appears
capacitive, of varying impedance, up to about 12.2 MHz when it once
again goes over to inductive but at a vastly lower inductance (single
digit microhenries), back to capacitive between 19.2 and 43.9 MHz,
and finishing up the sweep to the top frequency of 50 MHz as

With the top layer of wire stripped off and about 27 turns remaining in
a single layer, the indicated inductance on the 8753C was around 22
uH at 3.5 MHz, rising to 25 uH at 7 MHz, and 32 uH at 10.1 MHz. As
inductors get closer to their self resonant frequency, the indicated
inductance starts ramping upward. As I said, I can't really explain or
interpret that until I talk with someone at work who uses the analyzer
a lot, to find the practical significance of it. The important thing is
that with 27 turns the self resonant frequency was up to a respectable
17 MHz.

I would suspect that what's going on is that it's seeing the entire
network, consisting of a coil AND its distributed capacitance--which is
a rather significant 43 pF, much higher than typical toroids I've
measured for HF--and since the capacitance is fairly significant the net
inductance varies a lot with frequency. On the Q meter you also get
misleading results if you don't realize what's going on with the
distributed capacitance and compensate for it. For instance, it looked
like 114 uH, higher than it was "supposed" to be, but actually turns out
to be the 100 uH it should, but with 43 pF capacitance along for the
ride. The trouble is that the 43 pF is internal to the component and
can't be disassociated from it; that's when the engineering gets to be
"fun" :-)

I pulled off more wire to get 20 turns, and the 8753C gave about 12 uH
at 3.7 MHz, 13.6 at 7 MHz, 15.6 at 10.1. The self resonant frequency
was 21.8 MHz. At 3.7 MHz it was showing a net reactance of 290
ohms, and 600 at 7 MHz, both of which easily meet the rule of thumb
that says that an RF choke should have at least 4 times the impedance
of the circuit (presuming 50 ohms).

Finally, down to 15 turns. (As I pulled off turns, I kept them close
spaced.) The self resonant frequency was a respectable 26.8 MHz. It
held at 8 or 9 uH through 7 MHz, 10 at 10.1, 11.6 at 14. That 8.5 uH
on 80 meters provided about 197 ohms, which is right about at the
limit for the rule of thumb, but on 40M it was almost 400 ohms (9.0

I measured the second stock coil and it was also self resonant at about
2.6 MHz.

At home I measured the Cd on the remaining stock coil and the one
with 15 turns, using the "approximate method" given in the Boonton
manual (which they say gives results accurate to 2 pF). I didn't bother
doing the more exacting and complicated method. As I said earlier,
the Cd of the stock coil, which had two layers wound directly over
each other, was about 43 pF. "Real" RF chokes from the old days have
criss-crossed wire to keep the Cd lower. (If I can find one somewhere
I'll measure it and see how low the Cd is.) The 15 turn coil, with a
straight, single layer winding showed 3.3 pF. On the other hand,
every single layer toroid for HF that I've ever measured has been
"down in the noise", usually getting readings of 1 or 2 pF, readings
which really can't be trusted anyhow since the accuracy of the method
is +/- 2 pF. As far as I'm concerned, for all practical purposes, "HF
toroids" can be considered to have no Cd.

On to the bare form wound with 70 turns of wire, now that all of that
distributed capacitance talk is over. It measured 5/3 pF, or 1.7 pF,
hugely better than the double layer, stock coil of approximately the
same inductance. At 790 KHz it resonated with 343 pF, or about 118
uH. Since Cd is not even 2 pF, the apparent and true inductance are
essentially the same. As a cross check of the procedures, since I did
this on a later day, I went back and measured the stock coil again, and
got 41 pF, essentially the same as the other day.

Why did the 70 turns have less Cd than the 15 turns? The wire size
was a smaller diameter, essentially resulting in smaller plates in the
"capacitor". Although there were more turns, the wire size was
smaller by a sufficient margin that the Cd was less.

I won't be able to measure the self resonant frequency of the form with
70 turns on it for a while but would expect it to be reasonably high.
Calculations using the approximate values of 100 uH true inductance
and 43 pF Cd for the stock coil comes out to a hair over 2.4 MHz, in
the ball park of the measured value on the HP8753C. Due to the
resolution of the analyzer and touchiness of the dials, the figures I
pulled from it for self resonant frequency have a bit of fudge factor
built in anyhow. For the 70 turn coil, factoring in the 1.7 pF of Cd
makes it about 117.7 uH, resonated with its own 1.7 pF for an
estimated self resonant frequency of about 11.3 MHz. Not
extraordinarily high, but hey, it's a 118 uH coil after all! And hugely
better than the 2 1/2 MHz for the stock Radio Shack coil. [Update: I
put it on the network analyzer and it is self resonant at about 8.3

All of this may well be a lot more than you wanted to know, but I
enjoyed doing it; always good to play around with the basics :-) The
bottom line as stated above remains the same, that for a good RF
choke at HF I'd recommend at the very least pulling off that top layer
of wire, and I highly recommend chopping about half of the remaining
turns from that bottom layer.

Appendix--I later scrounged up a couple of old, criss-cross wound
coils. One was a single pie RF choke on a nonmagnetic rod and the
other was wound on a cardboard coil form with a slug. The core of the
choke was 1/4", the winding was about 0.32" long, 0.420" outer
diameter and approx. 0.090" in depth, with wire size of #36 or #38.
At 646 KHz it measured 1214 uH, 8 pF of Cd, and a corrected
inductance of 1046 uH. A huge amount of wire, but very small wire
and criss-cross wound, and thus a respectably low Cd, much lower
than the Radio Shack coil. Later testing on the 8753C showed self
resonance at approximately 1.7 MHz.

The coil on the form with a slug gave an apparent L of 3564 uH at 377
KHz, corrected to 3126 uH after factoring in the Cd of 7 pF. The size
was similar; the core is about 0.28" dia, 0.274" long, about 0.487"
outer diameter and 0.110" deep. The wire on this one is also #36. Like
the other, this one had low Cd due to the cross-cross winding
technique. In both cases it's hard to tell how many layers are on them,
but they both appear to have something like 6 to 8 layers. This one self
resonates at about 2.3 MHz.


73 and Queue Our Pea de WA8MCQ wa8mcq@abs.net

Search QRP-L Archives

[ QRP-L Archive | ]
[ 1993 | 1994 | 1995 | 1996 | 1997 | 1998 | 1999 | 2000 ]


This archive was generated by hypermail 2b29 on Fri Jun 02 2000 - 11:36:37 EDT