RF Column 30 - April 1994 Copyright (c) 1994,1995 H. Douglas Lung ALL RIGHTS RESERVED TOPICS: Construct a simple standard frequency receiver - Part II Tricks for obtaining 0.02 PPM calibration accuracy with it --------------------------------------------------------------------- Last month I told you how to build a simple standard frequency receiver and calibrator, but I didn't have room to tell you how to use it. Some purists might take exception with the claim I made last month that it was possible to use the receiver / calibrator to create a frequency standard accurate to 0.02 parts per million (PPM) using a few cheap parts. The conventional wisdom is that shortwave stations can't be used to lock oscillators that closely because propagation delays cause significant variations in the phase of the received signal. Unless you live next door to a shortwave standard frequency station, the signal reaches your receiver after bouncing through the ionosphere - sometimes more than once. The characteristics of the ionosphere change depending on solar conditions and time of day. That's obvious to anyone who listens shortwave or even broadcast band radio. Most commercial standards now use low frequency standard stations - like WWVB in Colorado or MSF in Rugby, Great Britain, both at 60 KHz - for a reference. These low frequencies are not affected nearly as much as high frequencies by ionospheric fluctuations. How can I claim 0.02 PPM accuracy with a shortwave standard? Superior computing power! A phase locked loop can't differentiate between a phase shift caused by a shift in propagation delay and a phase shift caused by a shift in frequency. However, most humans and certainly anyone reading this column can easily see this happening on an oscilloscope and disregard it. I tell you how to do it a bit later. First, here's how to set up the receiver / calibrator. I found the sensitivity of the receiver / calibrator to be similar to that of a portable shortwave radio, perhaps a little better. If you have the room, a half wave dipole cut for the frequency you are receiving and mounted outside works well. At 10 MHz, each side of the dipole should be about 23 feet (or 7 meters) long. You might already have such an antenna if your station uses a time clock system based on WWV. "Direct conversion" receivers sometime suffer from direct detection of strong AM signals. If you use an outdoor antenna and are near an AM station some additional filtering may be needed. I found Radio Shack's amplified shortwave antenna worked well. The whip antenna worked fine when propagation was good, even in south Florida. When the signal was weaker, a piece of wire plugged into the RCA jack on the unit helped, especially when I was able to get the wire outside. With an antenna connected, adjust input tuning capacitor C1 for maximum sound. If you didn't take the time to set the crystal oscillator with a frequency counter, you'll probably hear the beat frequency between the oscillator and WWV or other standard frequency station. It is difficult to adjust the crystal frequency to a precise zero beat (a difference of zero with the standard frequency station) by listening to the audio. Once the frequency difference gets below 50 Hz. or so it is difficult to hear! The best way I've found to set the frequency by ear is to set the crystal as close as possible when the station is sending an unmodulated carrier. Then, wait until the station transmits an audio tone (WWV and WWVB send standard tones at various times throughout the hour). I've found that by listening to the beat on the tone, I can set the frequency within 0.2 PPM -- that's a frequency of 2 Hz, or two fluctuations a second with a 10 MHz. station. This is good enough for most U.S. FCC requirements. Setting it much closer is difficult because the very low frequency fluctuation from the beat is easily confused with signal fading due to propagation. So what's the trick to improving the accuracy? An oscilloscope and an audio tone! If you're an old timer or watch science fiction movies from the 50's you'll recognize Lissajous figures. They're created when two audio frequencies related harmonically are connected to the X and Y (or horizontal and vertical) inputs of an oscilloscope. Most two channel 'scopes have the ability to use one input as horizontal. On the Tektronix 1480 and 1780 waveform monitors there is a BNC connector on the back for this - it's the same one you use for ICPM measurements. Apply the same frequency to both inputs, adjust the amplitude, and you'll see a circle. Double the vertical frequency and you'll have a figure 8 on its side. Change one frequency so that is isn't harmonically related and the pattern will move. Change the phase of two identical frequencies and the circle will rotate. Higher ratios trace complicated orbits on the scope screen. It's fun to play with. What's this have to do with frequency measurement? Simple. By connecting the audio output of the standard frequency receiver to the vertical input of a 'scope and the horizontal input to a known audio frequency, we can watch the scope to see differences in frequency too small to detect with the ear. The AC power line frequency is a handy reference in most places. If it isn't stable in your area, use either a vertical sync output from a sync generator or an audio oscillator calibrated with a counter. It doesn't have to be super accurate - at 100 Hz. a 0.1 Hertz error is 0.1 percent or 1000 PPM - several orders of magnitude higher than the accuracy we're aiming for. It doesn't matter. Since we're comparing the audio beat note from sources at 10,000,000 Hz., the error doesn't multiply - it adds. An error of 0.1 Hertz at 10 megaHertz is equivalent to an error of 0.01 PPM, which is within our tolerance. After zero beating the oscillator as close as possible to the 10 MHz. standard frequency station, carefully increase the oscillator frequency until the audio beat note from the receiver matches that from your power line or reference oscillator. On the 'scope, you see a circle. Every now and then it will flip or collapse. That's due to the propagation delay changing the phase of the received signal. At this point, you should have the frequency within a few tenths PPM. So far, everything I've written was pretty standard practice for frequency checks ten or twenty years ago. Here's the trick to set the oscillator even closer to the reference station. After setting the frequency as close as possible using the Lissajous figure, carefully move the audio reference frequency from the horizontal input to the sweep trigger input on the scope. If you're using the power line frequency for a reference this may be as simple as selecting "Line" as the trigger source. Adjust the horizontal timebase to display three or four cycles of audio. Unless you had amazing luck with the Lissajous figure, the sine waves will be moving left or right (hopefully slowly). Very carefully adjust the oscillator frequency until the waveform stops moving. You'll notice that every now and then the cycles will change polarity or shape without moving. Your eye can filter out (average) the propagation errors from the standard frequency station! The oscillator in the receiver is now locked within 0.02 PPM of the standard frequency station. That's the secret. A few things to remember - the frequency counter hooked to the oscillator on the receiver will no longer be reading 10000.0000 KHz. If you used a 60 Hz. power line frequency as the reference and followed my advice to RAISE the frequency to get the audio beat, the counter should now read 10000.0600 KHz. (10,000,000 Hertz plus 60 Hertz) Don't adjust the counter to read 10.0000000 MHz. using this method! If you are outside the reception range of a shortwave standard frequency station don't give up. Last month's circuit should also work with the low frequency standard frequency stations. You'll have to replace the crystal and input tuned circuits with L/C circuits resonant at the station's frequency. If you do this, let me know how it works. That's it for this month. Next month - my observations from NAB. Not the detailed show news - TV Tech's reporters handle that fine - just my comments on RF equipment at NAB - trends, nifty products and emerging technology. Your comments are always welcome. EMAIL them to me at my CompuServe ID, 70255,460 or route them through the Internet to 70255.460@compuserve.com. Mail is slower, but will also work - 2265 Westwood Blvd., Suite 553, Los Angeles, CA 90064. Fax is faster - 305-884-9661 should work most of the time. I'm busy during the day, but if you want to call, try my Miami office after 6:30 PM Eastern time at 305-884-9664. Copyright (c) 1994,1995 H. Douglas Lung ALL RIGHTS RESERVED