I just finished some RF testing of my Flex-1500s millihertz VFO accuracy, using the following equipment:
* HP 3325A Signal Generator (millihertz resolution) locked to a ThunderBolt GPSDO 10MHz reference * Flex-1500 with IF= 0 Hz, Spur Reduction = Off, CWU Pitch = 600 Hz, Clock Offset = 0, and locked to the same 10 MHz reference * Fldigi used to measure 600± tone via a Virtual Audio Cable. (Fldigi was set for +74 RX ppm for sound card sample rate correction, which may have affected these tests?) I injected 21 frequencies in 1 Hz steps, from 3,557,990 Hz to 3,578,010 Hz into the Flex-1500 at about 80 dBm, and incremented the PowerSDR VFO to match. I collected data with the fldigi Frequency Analysis tool into a csv file with sub millihertz resolution. I removed all of the transient data between the 1 Hz steps where it takes fldigi 5 or 10 seconds to start tracking the new frequency. The VFO RF frequency error was then represented by the audio frequency above or below the 600 Hz CWU Pitch frequency. I found a median shift / error of 0.068 Hz, which could, in part, be due to the fldigi sound card correction factor mentioned above. So, after subtracting the 0.068 Hz error, the frequency errors bounced back and forth between ±0.022 Hz as shown in the graph below. This organized zig-zag pattern is due to the DDS oscillator chips 32 bit binary resolutions never exactly matching the VFOs integer dial reading. [I posted a graph along with this text in the Files section of FMT-Nuts Yahoo Group: http://tech.groups.yahoo.com/group/FMT-nuts/ ] So I have some question about how to eliminate the 0.068 Hz median error I measured: 1. Since I am using a VAC between PowerSDR and fldigi, does the fldigi Sound Card Sample Rate Correction (+74 RX ppm) have any affect? 2. Could there be another (software?) source of the 0.068 Hz median error? (I did confirm my signal generators accuracy by using a GPSDO locked counter with sub millihertz resolution) 3. Should I try adjusting the PowerSDR DDS Clock Offset? I have found that it is still active even though I have checked the box that I am using an external 10 MHz clock reference. 4. Are there any warm-up frequency drifting affects that arent eliminated by using the GPSDO 10MHz reference and digital (virtual) audio signal paths? I plan to use this empirical dial correction method to adjust my readings following the FMT this Wednesday. It would be better to have a calculation utility like the Analog Devices DDS Simulator, but not until we know for sure what the PowerSDR and firmware are really sending to the DDS c hip. John Stuart, KM6QX Lafayette, CA _______________________________________________ FlexRadio Systems Mailing List FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archives: http://www.mail-archive.com/flexradio%40flex-radio.biz/ Knowledge Base: http://kc.flexradio.com/ Homepage: http://www.flexradio.com/