Hello Flexers, I did a freq cal four days ago and have checked it everyday since. Here is what I see on my SDR, -1Hz to +2Hz.
I'm using the 1 Watt SDR and I leave it and the PC powered on at all times. I cal by setting the mode to AM and the filter to 25Hz using the Spectrum display this yields five displayed data points or a resolution of ~13Hz. Since the display uses a 'connect the dots approach' you get a linear interpolation. Adjust the clock offset while centering the peak and getting equal amplitude at the one above(+13Hz) display point and on the one below(-13Hz) data point. Caution! When I first did this, I cycled through the various WWV freq's and it appeared 20MHz was off by 6Hz. What I discovered is that the Filter BW has to be the same for each frequency. It appears that when switching BW to 50Hz and maybe 500Hz it shifts the center freq by -6.3Hz. Maybe there's a quick fix for this? Comments? You can easily observe this by looking at a CW signal and changing the BWs. Also a comment about the Rubidium, Cesium and Crystal oscillators. Oscillators are characterized by short term(phase noise) and long term(drift) freq stability. The phase noise it what concerns us most when it comes to receiving. We want it low enough so that it doesn't swamp weak signals. Ideally it should be >10db below the noise floor of the receiver so as to not reduce sensitivity and hence dynamic range. It is the 'Q' of the resonant component(s) that determine the phase noise. The quartz crystal is still the leader when it comes phase noise. Rubidium and Cesium are superior in time keeping (long term stability), but have awful phase noise. These are the 'real' atomic clocks. The navy is the largest user of these. No GPS under water! Regarding GPS, its timing is based on the Cesium clocks. Back to the SDR. Since the crystal has the best phase noise we want to keep it. Long term drift, measured from hours to years, is dominated by temperature. Additionally, aging can cause 20Hz of drift per year at 10 MHz. It's caused by a change in physical properties. Trapped gasses escaping, contaminants, etc. Temperature varies physical properties such as thickness and area of the quartz, which in turn varies the frequency. A 5ppm oscillator can drift 50Hz at 10MHz with a 10 degree F change in temperature! RTXO's (room temp xtal oscillators) range from 2-10ppm. The largest temperature change occurs at turn on, therefore you should wait an hour before performing a freq cal. That's why I leave mine on all the time and have only seen a 2 Hz maximum change so far. I'm sure those using high duty cycle modes with 100W SDRs will notice higher shifts. Don't keep chasing your tail. During transmission the SDR will warm up and the freq will drift in one direction and when you go back to receive the unit will cool and return to it's original frequency. So don't keep recalibrating. The sound card clock is hopefully derived from a crystal. 5ppm at 48KHz yields 0.24Hz. Not much concern. Other Options: TXCO The temperature compensated oscillator uses components external to the crystal to offset drift. These parts (caps, inductors, thermistors) have opposite temperature coefficients. The TCXO typically gives a 5X performance increase over the RTXO or 1 ppm. 10Hz drift at 10MHz for a 10 degree change. Oven Controlled Crystal Oscillator. In this case the oscillator is placed in a temperature controlled box. The temperature is maintained at the crystal turnover point. This is where the freq change vs. temp is smallest. The best performance is about 20 degrees F above the highest ambient. Typically 150 deg F. The oven should be under power at all times, if not the crystal temp will rise from room temp to 150F at turn on. During warm up the freq could drift nearly 1KHz at 10MHz. After warm up you can expect <5ppb. Or 0.05 Hz at 10 MHz for a 100F change! Another idea is to take a 10MHz crystal and mount it onto a large piece of metal and place it in an insulated box. The metal will slow the temp variations as will the insulation. If you mount it external to the SDR the PA temp variations won't effect freq stability. As mentioned in other posts you can derive the reference from GPS. Don't overlook your friendly cell phone system. CDMA cell phones have very critical timing requirements and are synchronized to Cesium sources. I try to keep it simple. I'm always battling RF in this, RF in that. I don't plan to get the reference through another receiver. Sources of 10MHz oscillators are abundant, 99 percent of the RF test equipment manufactured in the last 30 years use 10MHz reference oscillators. Many are ovenized, almost all are in MW equipment. Surplus telecom equipment is another resource. I have picked up three ovenized oscillators on EBay for less than ten dollars each. Who said amateurs can't make a better radio? Well I guess this was a little more than a comment on atomic time standards. :) 73, John k2ox
