How to control the Reference clock to interference RF receiver?

[poorren]

Hi guys,
I'm trying to build a homebrew VHF receiver. When I build the first prototype, I divide the system into three board, power supply, baseband receiver, and RF receiver. The power supply is legacy board, which is designed for another circuit. The analog power supply and digital power supply shared GND plane. And it will offer power to both baseband and RF board.

I used a TCXO(10Mhz) as reference clock and placed on baseband board. Now, I used BJT amplifer plus 74AC04 to amplify and distribute this clock. One path is sent to a PLL on baseband board for ADC sampling rate. Another path will be sent thorough a SMA connector to RF board. With some experiment, I found there is 10Mhz noise on GND of RF board.

One of reason is the noise coming thorough the power supply. Because, they shared the same GND, I didn't place any inductor or bead between the GND of TCXO power/distribution circuit and power supply board's GND. So, the noise will directly follow the routine of power GND to RF board. Even I disconnect the SMA clock to RF board, the noise is still there. So, I guess this is true. Now, I'm considering to use seperate GND plane for different circuit(i.e., TCXO will has its own GND), and short with a bead.

But another concern is whether the noise may coming via the SMA connector. The SMA connector may connect the RF board GND and TCXO GND directly. I'm wondering if that will induce clock noise on RF board. The dilemma here is I must provide a reflux path for clock signal. How could I control that noise?

Do anyone have similar issue like me? Any suggestion or comment is welcome.
Thanks advance!

Jeff

 

[biff44]

How do you know there is 10 mhz actually on the board? Just because you touch some measuring probe to a board does NOT mean you are actually measuring signals only on that board.

You can distribute your 10 mhz using fiberoptic if you really do not want any clock signal floating on the receiver board. You want really good RF bypassing right at the crystal oscillator and any buffer stages, to keep 10 MHz off of their ground planes. You can try lowpass filtering the 10 Mhz so higher harmonics do not propagate. You can send the signal differentially with line driver/line receivers. You can add metal shield cans.

Those are the 1st set of steps I would look at.

Rich

 

[tony_lth]

For power supply of RF board, you should use at least 10uF to bypass 10MHz.