HiWe use the HMC733 VCO.
I have a question.
1. The frequency was 9.7 GHz. If we watched it for 1 hour, we could see that it was moved to the left little by little.At 18GHz, the frequency will continue to move to the right.This is the case with the Frequency Drift Rate parameter.The question here is that the frequency stops and shifts little by little. Why does the frequency continue to shift after 2 hours and 3 hours? Could you help me? Thank you.
The frequency of any free running (e.g. not phase locked) signal on a microwave VCO will drift; this is mostly due to temperature. Many of the HMC series VCO's are mounted to an aluminum backer plate. This backer plate not only adds a measure of ruggedness but also increases the thermal mass and reduces the instantaneous frequency drift due to for instance, convection currents from an HVAC supply vent in the lab. Ignoring external environmental factors, the part will also undergo self heating during normal operation and will reach thermal equilibrium after some period of time. I would guess that thermal equilibrium would be reached in less than 2 to 3 hours on the HMC733 but perhaps not. There are also other components in the thermal path that can complicate the thermal path resistance vs time. This includes the solder between the device and board which is likely not a consistent thickness, nor void free across the bond line; the printed circuit board also has a solder interface to the backer plate and whatever the backer plate is in contact with and the quality of that contact is important. The temperature of the device will have a much more difficult time reaching thermal equilibrium if these interfaces are not free of voids and are not very flat. In systems where a low drop out voltage regulator (LDO) is providing the supply voltage to the VCO the entire system board temperature will increase resulting in a small change in the output voltage which translates to a corresponding change in frequency based on the pushing value shown in the datasheet. Some Analog devices regulators like the HMC1060, HMC860 and HMC976 and perhaps others have a temperature proportional function that offsets this effect.
Finally, for the best thermal path, the device should be securely mounted to an infinite heat sink that is making intimate contact across the interface. A part that is just resting on a bench to chassis will have a difficult time stabilizing completely due to varying points of contact because of lack of flatness and intimate contact which can allow for air currents to accumulate within the interface and circulate. Once thermal equilibrium is reached (assuming no internal or external factors) I would expect the signal to drift up and down around some frequency. There may be some small aging term that might come into play as well. We don't specify aging on these VCO's as if it exists it is small enough that it doesn't matter for the applications for which they are intended which is the frequency input for phase locked loop (PLL) synthesizer applications. Any frequency drift or aging will be absorbed by the PLL charge pump and VCO tuning voltage margin available on the VCO and never noticed.