How to reduce or eliminate the phase noise hump?

[saulbit]

Dear everyone,
I am dealing with an active loop filter PLL. The output designed is 4050MHz~7050MHz per 25MHz step. The synthesic IC is HMC700 from Hittite Microwave, and the Kv for VCO is 2pi*130MHz typically, while the op in the fiter is OP376, with +/- power supplies.
The PLL can be locked easily. However, I found the spectrum of the phase noise of the output is like"^|^". In other words, there is a noise hump @ about several ten Kilohertz offset, and the height of the hump is about 15dB.
How can I reduce of eliminate the phase noise hump? Or did anyone met such problem in your active loop filter PLL before? Then how you overcame it? Thanks in advance.

 

[BigBoss]

It might be spurs which are coming from power supply.Or, the filter element might be bad calculated.
There may be other such causes, the design should be carefully studied.
Even a poor ground connection may cause this or Vtune line can be victim of a noisy circuit around..
The possibilities are so much..

 

[dinesh agarwal]

I think this is because of coupling of control voltage line with some signal of the offset frequency, resulting in the spurs.

 

[biff44]

What is an op376??? Not familiar with that one.

And what do you have hooked between the op amp and the VCO tune line?

 

[biff44]

No answers?

I would assume that the phase noise bump was due to control loop instability. In other words, not having enough phase margin where the open loop gain crosses zero dB. You need to make some minor changes to loop filter resistor and capacitor values for more phase margin. A computer program helps (analog devices has an ok one for free), although I would not trust one too much--what you see on the bench is more important.

There could be other issues. Some op amps do have noise bumps, so read the data sheet carefully. Some op amps can not drive capacitive loads, so lessen the capacitance at the output, or buffer the op amp output with a 220 ohm series resistor. Could be power supply noise, but that is trivial to check--just temporarily replace the power supply with lab quality LINEAR power supplies and see if the problem goes away. Could be a problem in your clock reference, which is probably so low in phase noise it is hard to test to see the bump in reference noise.

Rich


Added after 3 minutes:

A good way for a neophyte to see what is going on is to add a square wave perturbation (like capacitively coupling in a 10 mV square wave at the phase detector output, and using an oscilloscope to see the transient response on the VCO tuning line. If you see the square wave rise/fall, followed by all sorts of ringing and taking forever to settle out....that means your control loop is almost unstable. Control loops that are almost unstable ALWAYS have phase noise bumps.