https://www.febo.com/pipermail/time-nuts/2017-May/105298.html
The first time that we worked with this HP10514A mixer we had an HP105 oscillator to provide the 5mc comparison against our #1 cesium clock. Stability was not good enough to get closer than a few hundred picoseconds. We expected some diode drift but not at that level. We could not see a hot finger on a diode in the mixer. One millivolt out of the mixer is about 31 picoseconds phase change near quadrature at the drive levels used then. (±1 V peak dc) The theoretical dc output is sin θ where θ is the phase angle between inputs. Two volts pp of 10 mc rides on that dc. We were operating unterminated on the output of the mixer. As long as the diode drops are stable, we can still see phase changes to a fraction of a picosecond. A picosecond is the delay in an 8 mil length of RG58 cable. (1,500 ps per foot) We have one DVM with a resolution of 1 μ V dc. The data sheet on the mixer is at http://gonascent.com/papers/hp/hp10514a.pdf . We would appreciate anyone who has experience on diode drift on these types of mixers to post accordingly. We built a 400 kc five pole low pass Butterworth filter yesterday that removes the 10 mc from the mixer output. (CLCLC) Attenuation is about 10 million to one at 10 mc which reduces the rf to about 0.1 μV peak to peak. We can now see phase changes on the oscilloscope of a few picoseconds at 1 mV per division. The modulation deviation is hard to see at 5 mc and there is no handy output there to drive a mixer. We therefore drove our mixer with the 90 mc outputs of each HP5061B. We used 6 db 50 Ω attenuators ahead of the L and R inputs to avoid mixer overload. In phase dc out was about 0.5 V and 180° out of phase was about -0.5 V dc and was terminated in 50 by the low pass filter and the scope. We ran both HP5061B's in open loop with modulation on one of them on and the other one off. See http://gonascent.com/papers/hp/hp5061/waveform/90mcmod.jpg . This shows the 137 cps modulation. Note that there is a fair amount of noise present. This contributes to noise in the final lock. Modulation from cesium #2 was 16.9 mV pp and #1 was 13.9 mV pp. The #2 is about ±0.968° lead and lag or ±29.8 ps at 90 mc. Modulation is at 137 cps. Phase change in a half cycle of the modulation is 2x 29.8 ps in 1/274 sec or 16.3 parts per billion average rate of phase change. Average phase change for a half cycle of a sine wave is 2/π (0.636) times peak phase change. Peak phase change is therefore ±25.7 ppb or ± 2.31 cps at 90 mc. The 90 mc is multiplied in the A4 mixer by 102 to 9180 mc. At that frequency the deviation becomes ± 236 cps or 472 cps peak to peak. This is roughly the 550 cps distance between the cesium atomic resonance at 9,192,631,770 cps and the valley either side of it. We have not disturbed the modulation adjustment on our #1 machine which is 82% of the deviation of #2. We will correlate the HP modulation adjustment method with our measurements. It seems to me that over deviation of the modulator will increase the noise of the system. Has anyone experimented with lower modulation and higher gain on A7 board? We can easily see any improvement with our mixer. We are familiar with the 2nd harmonic gain pot but have decided that the instrument needs a complete alignment after the beam tube change. We do not know how many knobs were turned trying to get it to work with a bad beam tube. It needs an alignment like it would have gotten after it was built in the factory. πθ°μΩω±√·Γλ WB0KVV ---------- Forwarded message ---------- From: Bob kb8tq <kb...@n1k.org> Date: Wed, May 17, 2017 at 4:05 AM Subject: Re: [time-nuts] HP5061B Modulation Adjustment To: Discussion of precise time and frequency measurement <time-nuts@febo.com> Cc: "rwa...@aol.com" <rwa...@aol.com>, "Donald E. Pauly" <trojancow...@gmail.com> Hi Keep in mind that the mixer has millivolt level offsets and millivolt level drift (over a wide range). There is a gain pot that “sets” the second harmonic. It can be low and the device will still lock. Bo _______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.