WWVB Loopstick Antenna & C-Max Receiver
© Brooke Clarke, 2003 -
2010
 7.5" rod
|
 2.5" rod
|
Background
Bank Winding
Litz Wire
Measured Data
C-Max
Coil Design
CME8000-Bus-LP-01
CMMR-6P-60
K5JHF Clock Kit for the CMMR-6P-60
Thermal Time Constant
Links
Background
I wanted to make a loop antenna for
receiving WWVB based on
the Amidon
R33-050-0750
Ferrite rod. This uses the type 33
material and is 1/2" diameter by 7.5" long. After trying a number
of
different windings I think the current version may be optimum for WWVB
at 60 kHz. It's bank wound using 2 layers of 5x36x44 Litz wire.
Bank Winding
The following explanation of "Bank Winding" is for a 4 layer coil.
"Bank Wound" means that you start the winding at the left end, then
make 4 turns then wind backwards for 3 turns then forward for 2 turns
and then backward for one turn. Now there are 4 layers of wire
all at the left end. This process is repeated until the length
of the rod is covered. The important fact about this type of
winding is that the electrical distance from any one turn to it's
immediate neighboring turn is is just a few turns. If the coil
was wound a layer at a time, then the electrical distance between turns
may be hundreds of turns. The self capacity of the completed coil
depends on both the capacity between adjacent turns and also the
voltage between the turns.
If
you look at the equivalent circuit for a single layer coil where the
turns are spaced about one wire diameter it can easily be seen the the
total coil voltage is divided so that 1/N of the voltage is across each
turn-to-turn capacitance. That's to say the capactance is very
closd to the capactance of the first to the last turn. But if you
added a second layer then
the
two coil terminals are at the same end and the turn-to-turn capacitance
for the turns adjacent to the two terminals has 100% of the coil
voltage
across it. The next turns away from the terminals has (N-1)/N of
the full coil voltage, etc. This results in a huge self
capacity.
Litz Wire
There are a number of effects that cause the A.C. resistance of a wire
to be higher than the D.C. resistance.
Skin Effect
In a straight wire, with no other wires near by, as the
frequency increases the electrons carrying the current move away from
each other and so end up more and more near the surface of the wire and
less and less along the centerline.
Proximity Effect
Is similar to Skin Effect, but is caused by the
interaction between wires that are near to each other.
Litzenhardt wire is made up of a number of independent conductors that
are insulated from each other. There are a large number ways that
the strands can be arranged.
The correct term may be
Litzendraht, meaning Litzen (braided) Draht (wire)
Thanks to John Portune W6NBC |
The wire I used is made from AWG number 44 wires that are in bundles of
36 wires and there are 5 of these bundles in the final wire. If
you needed a cable with 180 seperate wires you could use this Litz wire
that way. Since a single
# 44 wire has a diameter of 0.00198" the total area of the 180
wires is equivalent to a solid wire of diameter 0.02656", or between
AWG 21 and 22.
The problem is the insulation takes up space that could be used for
copper. The result is that Litz wire typically only works well
below a few MHz. It's interesting that at mains power frequencies
(20, 50 or 60 Hz) the largest size single strand wire you can buy is
determined by skin effect. If a load requires more square mills
of copper area then you need to use multiple insulated conductors.
MWS - Litz tool - input
the AWG size to get the cross sectional area and it tells you the
available strand count and wire sizes.
Measured Data
The following plots were made using the HP (Agilent) 4395A
in the Network Analyzer mode and using the Z Transform technique where
the unknown device is placed in series with the center conductor.
The problem with this winding is that it's very difficult to bring it
to reasonance because of the difficulty in finding fixed caps of the
exact value you need.
A much better approach is to make the coil on a sliding former so
tuning can be done simply by sliding the former. That's the way
the commercial WWVB antennas are made.
|
Imaginary Part of Impedance Plot
|
The marker shows 1.004 k Ohms at 32.960891 kHz. The network analyzer
has already separated the real and imaginary parts so this is pure
inductance.
L = X / (2 * PI * F) = 4.83 mH |
Real Part of Impedance Plot
|
You can see that at 100 Hz the real part of the impedance is about 1
Ohm. i.e. the DC copper resistance. The marker shows the
resistance has climbed to 29.325 Ohms
at 60 kHz.
Without making a plot of "Q" vs. frequency and analyzing the slope it's
hard to say what's causing the resistance. It may be skin effect,
proximity effect, or dielectric losses or some combination of these. |
Resonated with Mica caps
|
Marker shows 94.308 dB at 62.485345 kHz.
Ohms = 10 ^ (dB/20) = 51, 927 Ohms.
The shunt capacitance is given by:
C = 1 / [ L * (2 * PI * F)^2] = 1.34 nF
The HP 4332 LCR meter measured the caps at
1.35 nF.
The loading capacitance value was chosen so that when the coil, caps
and a small variable cap and the input capacitance of an amplifier are
connected the resonance point can be brought right on 60.0 kHz.
The Q = XL / R = 51,927 / 32.96 = 1,575
A
similar plot of just the coil shows that it's self resonance
frequency is very close to 500 kHz. That's good in that you don't
want to run a high Q coil near it's self reasonant frequency and here
we are well away from it.
|
Coil Design
This appears to be an improved version
of the Temic LF time receiver chips. They specify the resistance
at reasonance [40k to 100k for the CME8000 = (1/Q) * SQRT( L/C) ] of
the loop for best signal to noise. They have an
antenna design -
tool
to help. The Q should not be between 40 and 150 for the CME8000
if higher temperature effects may tune the cirucit out of
reasonance. Note that they have a single IC that can be used with
a single loop antenna and by switching caps will tune to LF time
stations at 40, 60 or 77.5 kHz and knows how do decode them.
The above loop with the caps that are installed has a
resistance
at
reasonance (60 Hz) of about 52 k Ohms.
The reasonating cap should not be an X7R type since they change
capactance up to 10% during soldering.
The desired loop would be 1.5 mH for WWVB and the cap would be 4.7 nF
with a Q of 100.
To get the above coil down to 1.5 mH would require removing a lot of
turns which would reduce it's series resistance increasing the Q,
probably to way over 100. Then either a fixed resistor can be
added to control the Q or the reasonating capactance needs to be
designed to track with temperature.
Reasonating Caps
C-MAX has an app note on the choice of resonating caps. In order
to get surface mount caps they have gone with multilayer ceramic
cape. But that makes for a big problem. If the wrong type
is used there will be large permanent changes in the capacity because
of the soldering operation (maybe measure cap first then use silver
epoxy for attach?). For the loop above I used a combination of
mica caps and a variable cap to tweak the total.
Q
When the loaded Q is high there can be a problem with temperature
causing the center frequency to shift to the point the received signal
drops. For most commercial applications the Q is limited to 120.
There is an upper limit on Q imposed by the signal bandwidth.
L.F. time code stations use a modulation that changes once per second
so a bandwidth of say 10 Hz would be plenty. This implies that at
a Q of 6,000 you will start reducing the modulation.
CME8000-Bus-LP-01
In the photos above remember that my
loopstick is 7.5" long and the CME module is only 2.5" long. It's
interesting that both of the ICs on this module were made by
Atmel. The 3 station LF time
receiver is the Atmel CME8000 and the micro controller is the
AT89LS52
(8051 based) driven by a 11.0592 MHz crystal.
They make a number of different versions of the LF time signal board
and all those that have an on board RS-232 chip, like the MAX232, can
NOT be used in real time because the switching noise of the RS-232 chip
for generating the negative voltage is many 10s of dB higher than the
received LF time signal. The model shown here was chosen because
it has only TTL/CMOS outputs, i.e. does not have the RS-232 chip on the
board.
CMMR-6P-60
The
CMMR-6P-60 is the 60 kHz version (they also offer 40 kHz and 77.5 kHz
versions) for WWVB. It's based on the C-MAX CME6005 which is the
analog front end without a CPU.
Digi-Key p/n: 561-1014-ND.
The C-Max data sheet has a table calling this the CME6005 w/o CPU and
that's the way Digi-Key has listed it. But it's really an
evaluation kit for the CME6005 and comes with an attached (and tuned)
60mm x 10mm loop stick antenna. The C-max bag has a sticker
"Tested Sample". This is important because of the difficulty in
tuning the antenna to reasonance using surface mount caps.
Notice that the wire between the antenna and PCB is twisted. That
is required to minimize interaction between the magnetic field caused
by any digital circuitry and the very low level input signal.
Note there is a through hole type cap on the loop stick and in addition
there are pads on the board for C4, C3 & C7 to reasonate the
antenna. Since those three have no cap installed it's clear that
the thru hole cap attached to the antenna is reasonating it at 60
kHz. A method to do that is described in the CME6005 data sheet.
The cylindrical package is a 60 kHz series reasonant crystal acting as
a filter. It's possible to connect two crystals in parrallel (and
two of the three frequencies and this board would then output the time
code for whichever station was within reception range. But C1 and
C2 may need to be different values for different stations.
This circuit operates on 1.2 to 5.5 volts aned draws about 100
uA. That means all inputs and outputs are very high impedance.
13 Jan 2009 - have a receiver module powred at 3 V and a DMM watching
the output (1 second per bit should be visable) but so far the output
is between 2.9 and 3 V. Maybe later tonight it will start
working. Looking with a scope shows pulses about 4 us wide at
random times. The outputis mostly at +3 V and the pulses go to 0
V. Using a GPS receiver to supply a trigger pulse once per second
confirms that the narrow pulses from the CMMR-6P are not correlated
with the time.
4:30 pm it's starting to work. See the scope photo below:
The gaps in the bottom
part of the trace correspond to the time code. But they are not
solid.
by 7:23 about the same, dots, not solid at the bottom.
This is connected to TCO and not working very well. Maybe there
was a wiring probllem?
Changed from powr supply to battery operation and added LED and
resistor to monitor.
First connected LED to the TCO output, but that means the LED is
normally on all the time draining the batteries and it didn't seem to
work very well.
Connecting the LED to the logic inverted output TCON works better and
saves the batteries.
This is a very handy gadget in that you can take it to different places
to see how well it works.
Put in the bedroom at night you can see that the data is more solid,
but far from noise free.
Next to put a scope on TCON.
A couple of wires were soldered to the board so that other antennas can
be connected.
This is a kit to make a clock based on the CMMR-6P-60. Shown below as
received.
 |
|
Kit Assembled Running Without
WWVB Receiver
|
The clock comes up in
hh-mm-ss mode and with
CW ID active. The LED
flashes along with the CW.
|
|
|
This kit makes use of some surface mount parts but is straight forward
to hand assemble with a soldering iron. But there's one
gotcha. The2x5 header for the LCD is located inside the 2x8 DIP
socker for the micorcotroller so you have to install the 2x5 first then
the 2x8. Also be sure that pin 1 on the LCD matches pin 1 on the
kit PCB. The piezo has seperate "Tic" and "Toc" sounds that are
cute.
Now, where did I put the CMMR-6P-60?
Got another CMMR-6P-60 from the maker of this clock kit.
When first powered on around 3 pm local time the clock did not sync,
but did sync around midnight, but ever since it has not displayed any
starting letter when I've checked.
I'll be checking closer to noon to see how it's doing.
When looked
at
|
yy mm dd hh mm ss
|
19 Feb 2010 about 11:51 (Day of
Year = 50) UTC = 19:50
|
10 02 19 05 01 59
|
| 19 Feb 2010 about 12:01 |
10 02 19 05 01 59 |
This
means that the clock lost lock at 5:02 am this morning, but no first
letter has appeared since so am confused how they work.
|
20 Feb 10 am
|
10 02 20 01 01 59
|
23 Feb 9:32 am
|
10 02 20 01 01 59
|
This
menas the clock has NOT synced for a few days. The time show is
12 seconds slow.
|
This photo of the clock
running a little before noon shows no prefix letter. Note I have
assembled the kit so that the setting buttons are facing the same
direction as the LCD face to make it easy to operate the buttons.
Top button is "Up/Increase/On"
Center button is "Down/Decrease/Off"
Bottom button is Mode,
First Letter of Display
L
|
Loss
of
detectable
signal |
U
|
detected a signal but has
not detected a 200ms sync (Un
Locked)
|
E
|
received noise and is
starting the sync detect again |
S
|
detected a 200ms Sync and needs a second sync before
it sets the clock |
<blank>
|
two consecutive syncs detected
and the clock is set/synced |
Better Antenna for the CMMR-6P-60
I have some Russian Ferrite rods on the way. The eBay ad says:
"Large Balun Ferrite Rods 10x200mm"
They are very similar to the 7.5" rod above, but this time I'll make
provision to tweak the reasonant frequency in a simple manner.
Patents
Temic (now C-Max)
10/999,339 Radio-controlled clock and method for acquiring time
information from a (C-Max), application - ignores non data bits
US 2005/0122952 A1Radio-controlled clock and method for automatically
receiving and evaluating ...(C-Max), application - multi frequency
receiver
Horst Haefner et al
7369628 Method for gaining time information and receiver for
implementing the Method, (C-Max), May 6, 2008, 375/324 -
discriminates based on width of pulses
Calls: same as 7317905 below except for the last one
7317905 Radio-controlled clock and method for gaining time information
(C-Max), Jan 8, 2008, 455/265 ; 368/47; 375/354 -
AGC control in the front end rather
than by the uC
Calls:
| 4440501 |
Method of
automatic adjustment of self-contained radio-clock by means of time mark |
na
|
Apr 3, 1984 |
| 4768178 |
High
precision radio signal controlled continuously updated digital clock |
PST
|
Aug 30, 1988 |
| 4823328 |
Radio
signal controlled digital clock |
PST empl
|
Apr 18, 1989 |
| 5105396 |
Autonomous
radio timepiece |
na
|
Apr 14, 1992 |
| 5349570 |
Method
for operation of a radio-controlled clock and radio-controlled clock
for use in an environment subject to interference fields |
Temic |
Sep 20, 1994 |
| 5528560 |
Timepiece
receptive of a broadcast time-signal for correcting a time error |
Seiko
|
Jun 18, 1996 |
| 5727022 |
Method
for improving the signal-to-noise ratio in a transmission system by the
formation of area equivalents |
Temic |
Mar 10, 1998 |
| 5805647 |
Method for
detecting the beginning of time messages |
Temic |
Sep 8, 1998 |
| 5818851 |
Method
for detecting the time messages in the faulty signal of a time-signal
transmitter |
Temic |
Oct 6, 1998 |
| 6456831 |
Amplitude
change time activated phase locked controller in a selective call
receiver |
NEC
|
Sep 24, 2002 |
Calls:
4768178 High precision radio signal controlled continuously updated
digital clock, (Precision Standard Time, Inc.) Aug 30, 1988 - see
PST1020 WWV receiver
4823328 Radio signal controlled digital clock, (not assigned, same
names as on 4768178), Apr 18, 1989
5528560 Timepiece receptive of a broadcast time-signal for correcting a
time error, (Seiko) Jun 18, 1996 - time receition only when watch
error needs correcting based on past history to save battery
Calls:
| 3281806 |
Pulse
Width Modulation Representation of Paired Binary Digits, Honneywell
|
Oct 1966 |
| 3806656 |
Decommutation
Device
In
Use
In
Particular
in
a
Transmission
Link
with
a
Missle
|
Apr 1974 |
| 3939304 |
Decommutator
for
extracting
zero
and
one
bits
from
a
coded
message
of
duration-modulated
pulses |
Thermal Time Constant
If a free running 32 kHz oscillator was
enclosed in a combination of thermal mass and insulation where the time
constant was much longer than a day then the daily temperature
variations would not be such an influence. To get a feel for how
big a mass and what insulation that represents some experiments follow.
1#
A brass rod 1/2" dia by 3 3/4" long
weighing 102 grams is heated in the oven, removed and lightly wrapped
in a towel. The time and temperature are recorded. The time
for the temp to drop by 1/e is 6 min 35 seconds. So to get a
multi-day time constant would take much more mass and much better
insulation.
Links
DCF77
vs. GPS - an I&Q type LF time receiver - the builder, Kasper,
suggests that a direct to DSP approach would be better, see:
LORAN-C Receiver by
Poul-Henning Kamp - an improvement would be to band limit the loop
antenna, but not high Q since sferics will cause ringing.
MSF Radio
Receiver - the MSF signal suffers the same problems as the WWVB
signal.
Back to Brooke's Products
for Sale, Military
Information,
Electronics,
Home
page
13598 hits since May 07 2007 page created 22 Sep.
2003.
7.5" rod
2.5" rod