[ntp:questions] Building/cannabalizing a WWVB radioclock...
joe at somewhere.org
Sat Aug 4 09:04:52 UTC 2007
On Fri, 03 Aug 2007 01:25:37 -0700, Pete Stephenson wrote:
> First off, I apologize for the length of this post.
> This evening, I was perusing
> http://www.buzzard.me.uk/jonathan/radioclock.html as I'm considering
> building a WWVB reference clock as GPS isn't viable here to due
> environmental conditions (buildings) and CDMA receivers being too
> I considered the cost of purchasing the various items needed on that
> list and the various electronic work needed. Me, being lazy, thought
> that there must be something cheaper, easier, and more practical.
> Thus, I present my RadioShack Radio Controlled Clock With Thermometer
> (Cat. No. 63-973). I partially disassembled it, and the interior can be
> seen in picture . The ferrite core antenna is in the lower right, and
> is connected to the small green circuit board in the top right. Four
> wires then run to the main circuit board and then connect to the
> lower left.
> Fortunately, the nice folks at the clock factory in China have labeled
> all their connection points. There are four solder spots: VCC, PON,
> WWVB, and GND.
Basically you just need to strip out the ferrite core antenna and the
small PCB in the top right. You should then be able to use this as a
direct replacement for the MSF board in my schematic.
> GND is obviously ground, and WWVB is (I presume) the modulated clock
> signal. Does anyone know what VCC or PON are? I'm not familiar with the
> terms, but I would imagine they had something to do with power (the
> whole clock runs on 3V DC provided by two AA batteries in series).
VCC is the voltage connection. Most of these boards have a fairly wide
input range. PON is a power on, basically to preserve battery power the
clock only powers the radio receiver every few hours gets an accurate time
stamp to correct itself, and then continues with the on board quartz
crystal. WWVB is the modulated clock signal and is the same as the MSF on
> My intuition tells me that the VCC wire in my clock corresponds to the
> VSS wire in the website's schematic, and that PON is also ground. It
> also tells me that my WWVB wire corresponds to the MSF wire in the
> schematic. Does this sound reasonable?
Usually you turn on the received by grounding the PON connector. When
powering the device from the serial port, we firstly want a continuous
clock signal, and secondly we are not worried about draining batteries.
Therefore in my schematic I have it permanently grounded and the receiver
> My intentions are as follows:
> - Remove the antenna, small board, and wires relating to the radio
> receiver from the clock housing.
> - Connect the wiring to a serial connector.
> - Connect the serial connector to the computer.
> - Use the software at the above site to allow the clock to communicate
> with NTP via the shared memory reference clock driver.
> As always, destruction is easier than construction. Removing the stuff
> from the clock housing is the work of a few minutes and a pair of
> scissors. The only thing I'm worried about is remembering which wire is
> which (another reason why I took the pictures).
I would remove the wires from the small PCB and attach the additional
electronics as per my schematic to the PCB.
> I have a few questions (and no doubt will have more), so please bear
> with me.
> 1. Does the radio receiver draw power from the batteries, or is there
> any current provided by the antenna? There doesn't seem to be any power
> converting apparatus on the circuit board, so if power is provided by
> the batteries, it would seem to be 3V. Is this common for WWVB radio
It does, but tiny amounts, just a few milliamps. This is low enough to
draw the power parasitically from the serial port itself.
> 2. The schematics say that it is necessary to connect the VSS wire to
> the DTR (pin 4) of the serial cable for power. What voltage does the
> port supply? Would it cause any problems to use this voltage with the
> radio circuitry?
It varies from serial port to serial port, but can be from between 5V to
15V. Typically it is around 9V, and it would be exceptional for it to be
above 12V. Most of these OEM receiver boards can operate from a wide
range of supply voltages. The EM2S receiver from HKW Electronik operates
from between 2.2-15V. Depending on how much the donor clock cost I would
just wire it up directly.
> 3. If the serial line voltage would be excessive, I could easily supply
> power from a set of batteries (probably a pair of D cells for long life)
> or a 3V AC-to-DC adaptor. If this is necessary, how would go about
> modifying the schematics for this?
Unnecessary just power it from the serial port. If you are really worried
stick a 3V zener diode in the power supply circuitry. In fact you can just
put a low current LED in there and it will work.
> 4. The website and schematics show various capacitors, resistors, and
> other such components wired in with the circuits. I presume that the
> necessary components are already wired into my circuits by RadioShack
> and that the addition of these components would not be necessary. Can
> anyone confirm (or speculate, if you have anything more than a wild-ass
> guess) if this is the case? The green box on my circuit says the
No you need to the additional bits as per the schematic on my website.
> 5. When working as a wall clock, the clock periodically syncs the time
> with the WWVB signal. I would imagine that the clock circuitry is what
> initiates the sync -- there doesn't appear to be any switching mechanism
> to turn on and off the radio receiver, so I presume the radio is
> receiving continuously. Obviously, for use as a reference clock, the
> radio would need to be receiving continuously and I presume the radio
> does run receive on a continuous basis and the clock only listens to the
> clock at the periodic times. Does this sound reasonable?
The normal thing is that it pulls the PON to ground which powers up the
receiver circuits, gets one or two time signals and then powers it off to
improve battery life.
> 6. Assuming I can successfully make a working reference clock in this
> manner and fudge it to take into account transmission distances,
> processing by the receiver circuitry, and OS handling of interrupts, how
> accurate a time signal could one reasonably expect to get? I can get +/-
> 10ms accuracy from internet time servers, and would hope to get ~1ms
> accuracy from a radio clock. This would be suitably accurate for my
> purposes, and hopefully accurate enough for internet users of my time
WWVB is a stratum 0 reference, so you would have a stratum 1 reference
clock, and 1ms is easily possible.
> I realize that WWVB is not quite as precise as GPS, but according to
> NIST WWVB can provide an uncertainty of less than 100 microseconds. I
> would be satisfied to get within 1ms, as it'd be a 10x improvement over
> my current setup, and I wouldn't have to pester internet time servers
> anymore. Tucson is 1006km from Boulder, so according to my back of the
> envelope calculations, there's about a 3.353ms delay for the radio
> signals to travel this distance. Does this math seem right? Assuming I
> can get consistent results from the radio, I can fudge it relative to
> better time standards (the local university has an NTP server with PPS
> input from a Trimble Palisade GPS receiver, so if I lug it over to
> campus I can sync it with almost no network delay) to get it more
> precise if necessary.
There are other delays as well, so if you can compare it to another
reference clock such as GPS you can add a fudge line to ntp.conf for
better results. However ntpd takes into account network delays.
If you do get it working let me know, and anything extra you needed to do
and I will put it up on my website for others to follow.
Jonathan A. Buzzard Email: jonathan (at) buzzard.me.uk
Northumberland, United Kingdom. Tel: +44 1661-832195
More information about the questions