I'm currently working on a PLL design that's to generate frequencies from 12 GHz to 20 GHz.To do so, I'm using a HMC733 VCO with a HMC704 PLL (with a HMC447 div-by-4 to feed the VCO reference back to the PLL). After tweaking the loop filter, I finally got the PLL to lock throughout the frequency range I wanted, with one drawback: I'm noticing that the noise floor rises on both sides of my carrier signal as shown below:
The SpecAn capture I show is for a 16 GHz VCO output. This behavior is present throughout most of the frequency range of operation of my design (12 GHz up to ~18 GHz). Any thoughts on what might be causing this? I've tried increasing/decreasing the loop filter BW but this behavior persists. I'm currently using an 3rd order active C loop filter, with a 500 KHz loop filter bandwidth and a 80 degree phase margin. Also, both the VCO and the div-by-4 are being powered by a HMC976 (low noise linear voltage regulator).
Are the spurs present in integer mode? It might be that with the wide loop bandwidth the DSM quantization noise is becoming dominant.
Thanks for the quick reply. The plot I showed was actually with the PLL running in integer mode.
Okay, if you put your finger on the loop filter op-amp circuitry does the peak move around? I am suspicious that the op-amp loop filter may be unstable/peaky.
I tried doing so but the spectral output didn't change. I decided to
implement a loop filter for fractional mode to see if I saw similar
behavior as I did in integer mode. After tweaking the PLL register
settings, I managed to get the PLL to lock from 12 to 20 GHz.
What's happening now is that I'm getting single tone spurs at the same
frequency offsets where the noise-like spurs were originally appearing. I
tried playing with the different fractional modes and with different charge
pump offsets. The loop filter I'm currently implementing has a bandwidth of
50 kHz and a phase margin of 75 degrees
If it helps, I can upload a screen capture of the spectral output later.
On Wed, Feb 22, 2017 at 9:26 AM, DonY <
Not sure if you're still having issues. Another idea is check the open loop behaviour of the VCO. Set Reg 0x0B=1 to force the opamp output low. The VCO spectrum should not have any spurs unless there is a power or grounding problem.