All times UT unless otherwise noted
WWVB disciplined clock
by John Wright
( "Attachments" are at the bottom of the page )
I designed a 1 Hz crystal controlled clock (attachment 1), and its
output pulses drive a gate follower/emitter follower pair and matching
transformer that puts seconds ticks on the DVD recorder's audio
channel. The audio going to the DVD recorder now has a versatile mixing
network so that I can have any
combination of WWV or CHU shortwave or the microphone input for marking
times by voice input or the seconds ticks from the 1 Hz clock. It works
The 1 Hz clock pulses also drive a cascaded CD4529 pulse generator circuit (video marker) that creates a distinct broad stripe/narrow stripe pattern on the TV monitor (see attachment 2). The spacing of those two stripes straddles the equalizing pulse region (vertical sync) so that you will at least have one or the other stripe in a given interlace field. That way you know exactly where the even second occurred. Even seconds are at the top edge of the broad stripe. I had to use gate and emitter follower circuits to get the micropower CMOS pulses adapted to the real world of 75 Ohm video. There is a 20 kOhm attenuator for the video marking pattern output (what goes to a video summing network), and it is set close to 300 Ohms for what you see in attachment 2. As you can see, it doesn't screw up the video even at saturation level. By increasing the attenuation the pattern can be set for much less contrast -- translucent -- so that underlying image features are not lost. I liked that. The clock interface that extracts even seconds strobes from WWVB's modulation (attachment 3) has not change much from what I sent you earlier. I added a new strobe generator and buffer that sends a reset strobe (via coax) to the 1 Hz clock. The strobe is applied to the clock's CD4059 chip (master reset), and that locks the clock's second pulses to WWVB.
The system is simple to use. You press and hold the "WAVE" button on the back of the Westclox until the yellow PWN LED indicates low, the press the white button switch on the clock interface circuit board until a seconds strobe goes by (a green LED flashes seconds), then release the switch (the switch is spring loaded). Thus, as far as the 1 Hz clock is concerned, the only purpose of the Westclox is its WWVB receiver output. But if shortwave WWV or CHU are completely missing (and it certainly happens when you really need good time signals), the Westclox provides an accurate display after a "WAVE" reset. It can thus be used for adding time marks to the DVD's audio channel by voice.
The fourth attachment shows that large loop that I described earlier, which now has windings and capacitance to resonate at 60 kHz. Going to more gain will not necessarily improve the WWVB signal, especially in summer weather with T-storms around (it amplifies the lightning impulses, too!). But high gain is preferred because you can always add attenuation and explore the full range of antenna gain from a boost of 0 db to maximum. I suspect that the loop will perform impressively in winter conditions. The naked Westclox works well enough even now.
I may send an account of this project to the ALPO journal, and that would be the end of much further development. It's something to use. I'm a scientist, not an entrepreneur. I have no interest in trying to develop and market a new gadget for amateur astronomy!
Details of the WWVB-based time stamping device.
I'm attaching the circuit diagrams in the order of information flow from the clock to the marking circuit. The Clock's manual follows the diagrams.
With everything connected and powered up, the yellow PWN lamp will be lit. When the WAVE button on the back of the clock is pushed and held for a few seconds, the PWN lamp will turn off and the red DATA LED will flash the WWVB timecode. When this is happening the clock is updating itself with reference to WWVB, and it will stay in that state for at least a full minute. The interface will detect the exact seconds, and the green LED will strobe the seconds. The PWN lamp absolutely has to off and the Red DATA LED flashing before the circuit will lock the 1 Hz clock to WWVB. Pay attention to the signal strength display on the Westclox's display (marked #9 in the manual's figure). If it isn't showing all four of those lines that step outward, the signal isn't fully robust. Also the red LED shouldn't show any spurious chatter, nor should the green LED show multiple seconds strobes.
So if the signal indicates strong, push the momentary switch until a strobe goes by, then release the switch. Also, record the time when you did that. If the marking pattern flashing on the video display was out of step before you pushed the switch, it will now coincide with the ticks from WWV's shortwave signal. In fact, for the fun of it you should turn the 1 Hz clock on and off several times, until the mark and the WWV tick are actually syncopated (i.e., by trial and error), then you'll note the distinct lock-in when the switch is pushed.
My 1 Hz clock runs slower than the NIST standard, but the drift is systematic, because the clock is crystal controlled. I simply make a correction in the fasion of the ship's navigators of old. The time is noted when the local clock is synchronized with WWVB. The time of an event is also recorded, and delta-t is calculated (by subtraction) and multiplied by the drift constant. The calculated drift error is then added to the local clock's apparent time.
The audio going into my recordings now includes seconds clicks from the 1 Hz local clock, and if the shortwave signals (WWV or CHU) are poor, all is not lost. Actually, I can mix or match all the signals, however desired. I punch record on the audio tape recorder and manually mark the click of an even minute (and add any other necessary comments, etc.). This can be repeated as often as I wish. So I count seconds from the noted point in time to the second that immediately precedes an event (actually, I use a stopwatch). That gives me time down to a second. The fraction of a second is worked out by single framing through the video and counting frames from the immediately preceding marked frame.
The device is not inherently expensive, but it doesn't give you a digital time stamp on every interlace field. It simply marks every 60th field. The even second is at the top of the broad stripe.
|PARTS - Listing and Suppliers|
Costs of components for the WWVB based video marking device:
If you build these circuits on prototyping boards, those will be the most expensive components. A 6 inch board can be obtained for under $20.
Westclox 70026 "atomic clock" (proxy atomic clock) - $10.88 (Wal-Mart)
Anyone who is serious about building the device should go to Wal-Mart and buy one right now! Most of the time (here in North Central Arkansas) the WWVB signal is robust. I may add an external antenna in the future, and it will be inductively coupled.
The electronic devices are actually cheap.
MPF102-ND - $0.27
2N2222 - $1.09
CMOS integrated circuits:
CD4059AE (programmable divide by N device) - $3.57
CD4049UBE (inverting buffer) - $0.53
CD4040BE (divide by 2^N) - $0.48
CD4528BCN (dual monostable multivibrator) - about $2.00 (Digikey doesn't have it
in their current catalog, but they will usually have it or an equivalent device.
ask for an equivalent IC with the same pin-outs.
Crystal crafted to a specific frequency. $?
I haven't bought a crystal cut for a specific frequency in a long time, but that would be the best approach to obtain a 1.312 MHz frequency standard. I suggest looking around supply houses such as Tanner Electronics. If a 1.312 MHz crystal can't be found, consider one near that frequency and try re-programming the jam inputs of the CD4059 IC.
|PDF Data Sheet Downloads|
|PROGRAMMING THE CD4059|
1 Hz Synthesizer - programming the
A 1.312000 MHz crystal was found in my box of spare RF crystals. Plan: Use a CD4049 for the oscillator and buffer and follow that with a CD4040 binary divider to obtain an intermediate frequency of 5125 Hz:
2^8 = 256
1,312,000 Hz/256 = 5125 Hz
Then use a CD4059 IC to divide the intermediate frequency by 5125, to get 1.000000 Hz; this is long division, not binary division:
There is a solution for division mode 4, which sets a divide-by-four pre-scaler; this divides the intermediate frequency by exactly 5125: 5125/4 = 1281.25, which is 1281 with a carry of 1
Set Ka Kb Kc = 0 1 1
Jam Inputs 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Status 1 0 1 0 1 0 0 0 0 0 0 1 0 1 0 0*
___ ___ ________ _________ _________
Meaning 1* + (1000 + 1 + 80 + 200) x 4 = 5125 = N
*zero means connect the jam input to Vss (ground) and one means connect to Vdd (+12 V)
* Jams 1 and 2 set the division's carry (maximum of 3), which is 1 in this case. RCA's tabulations of example calculations shows the order of jam inputs with the least significant bits on the left, which is an odd notation. I used it to find the IC's programming inputs. The K inputs can be cycled through 0 0 0 (master mode) to synchronize the divider to an even second from WWVB.
The circuit will go into a thermostat enclosure, so that -- like a ship's chronometer of old -- a divergence of rate (+ or -) will be constant from the point in time when the above free-running clock is set (locked) to a standard (in this case, the WWVB atomic standard), to any later point in time, and a correction factor may be applied to measured times.
This site was last updated 07/10/17
© Copyright, Derek C Breit. All rights reserved.