Dear Analog Engineering Community,
I am using the ADF4113HV to generate a phase locked signal on 7.784-7.786GHz from a 48MHz crystal reference.
The simplified loop circuit is
The settings are as shown:
The current setting resistor is 4.7Kohm.
The resulting signal is contaminated with a travelling spur much like that seen in the fractional N PLL synthesizers. I tried adding a 470k-ohm negative bleed resistor, but the spurs were still observed with no chance. I notice that the spurs are often strongest at power up and move away with time. However, if I am unlucky, the spurs move close to the carrier and stay there, causing severe phase disturbance. Shifting the reference frequency slightly causes them to move about. What is going on here?
Is there something about my settings that is causing this or is it an intrinsic problem with the phase detector of the 4113HV chip. I tried running on battery power, and the results were the same. Help!!!!!!! I am doing some sensitive precision timing measurements and this is really leaving me in a bind!
Thanks and best regards,
Thanks for your response.
I found the source of the wandering spur. It is a subtle temperature dependent interaction between a voltage regulator and an 3rd overtone voltage variable crystal oscillator…
I found the source of the wandering spur. It is a subtle temperature dependent interaction between a voltage regulator and an 3rd overtone voltage variable crystal oscillator in a phase-locked 48MHz reference. Apparently, slight shifts in supply voltage would intermittently induce the excitation of a spurious mode of oscillation that would interfere with the desired mode. Since the frequency multiplication is so high, the phenomenon was not visible on the reference signal spectrum. The problem was only confirmed by using a known clean reference for the PLL. By changing a couple of capacitors in the crystal oscillator feedback network, the problem went away.
At the end of that day, the ADF4113HV was working perfectly. I only initially suspected the digital dividers because of having seen the exact same effect with the fractional-N PLLs. Since this is an integer N PLL, the frequency dividers cannot have this behavior.