philip@beeblebrox.dle.dg.com (Philip Gladstone) (10/05/90)
In article <59758@bbn.BBN.COM> clements@bbn.com (Bob Clements) writes: >It can hear the WWV signal for at least two hours out of 24 (in >Massachusettts). It listens ONLY to the 100 Hz subcarrier, not >the 1000/1200 Hz tones. It is certainly not in the "Stratum 1" >category. It is somewhat worse than the Heathkit clock, since it >listens only to one frequency (10 MHz) and one audio tone (100 >Hz). I would be really suprised if a board placed anywhere near a PC worked even two hours out of 24. I have built a couple of 60 kHz clocks (synced to MSF Rugby) and they are both stunningly sensitive to VDUs and monitors! However, once positioned reasonably, they give very good results. I seem to get about 700us variance on the 1 second pulses. I think that with a bit more fiddling around with the RF stage and the filters, I could probably get it down to somewhat under 500us. Bear in mind that this is only 30 cycles of the carrier, and you must agree that it is not bad! My stratum 1 clocks just consist of a radio demodulator that gives me a 1 pps pulse train, and then I decode that in software -- taking an interrupt off the leading edge (which corresponds to the second marker). The leading edge is timed using the computer clock (it is the variance in this number discussed above). The pulse data can easily be sampled from the computer and then decoded to give the actual time. I actually use the parallel printer port on my machine and take the data stream into the /ACK and ONLINE pins. My hardware takes an interrupt on rising edge of /ACK, and then I can sample the ONLINE pin! I would have thought that the same approach could be taken on a PC -- the transition of /ACK certainly can cause an interrupt. [I seem to recall that which edge the PC triggers off depends on which adapter card you use!]. The clock board that I used cost around #20 (pounds UK) which is under $40 (US). I would have thought that you could do a nice stratum 1 server with enough effort and a spare PC! Some considerable care might be needed to get the precision up to a decent level -- especially if it is a slow PC. Anyway, enough ramblings. Philip Gladstone philip@dle.dg.com Development Lab Europe C=gb/AD=gold 400/PR=dgc/O=dle Data General, Cambridge /SN=gladstone/GN=philip England. +44 223-67600
Mills@udel.edu (10/06/90)
Philip, Well, remember Rugby is not as far from most places in the UK as Ft. Collins is from most places in the US. The secret to the Spectracom WWVB receiver success is undoubtably the very narrow crystal filter just after the RF stage, which nicely kills the 63-kHz harmonic of the TV sweep frequency used in the US. The result is that 100-microsecond accuracy is not hard to achieve. In spite of that, your 700-microsecond variance still sounds too large to account for in the receiver, suggesting maybe much of that is due to the computer interface and interrupt latencies. I surely can't argue that your interface is anything but inexpensive! Dave
Mills@udel.edu (10/31/90)
Philip, Sorry for the delay in this response. Your clock sure has the virtue of being undear. An accuracy of 700 us is not too shabby, considering the best WWVB clocks known here provide resolution at the 100-us regime. If you are within the groundwave contour of Rugby, you should be able to better that to a few usec and, if not, probably can't better 50 us unless you are very careful in the times of observation. However, I have concluded that, for accuracies down to the usec, that LORAN-C is the target of choice. I have just designed a receiver for that, intended as a PC peripheral, but have not yet begun the build. Dave