RF Isolation

Thread Starter

Bill B

Joined Nov 29, 2009
61
Hello,

I have a question about how to select an RF choke for isolating oscillator frequencies from a DC power source. The circuit I am building contains 4 oscillators, one each at 150 kHz, 500 kHz, 1.4 MHz, and 30 MHz respectively. The goal is to sum these frequencies and capacitively couple them onto a household power mains. Ultimately this device will be used to test a CISPR 14 discontinuous interference analyzer. I have a working prototype up to the summing part. The problem I am having is controlling the amplitudes of the individual frequencies. This is due mostly to coupling of the frequencies on the 5 Vdc and GND lines that power the circuit, particularly the higher frequencies. I would like to isolate the 5V and GND lines from the oscillators and ICs in the circuit with RF chokes. I'm not very concerned with the 150 or 500 kHz oscillators although I will still isolate them. I would like to use the same value choke for all connections. I'm thinking something around 100 mH will work. I have looked at a few data sheets for chokes, but I'm confused about how to weigh things like self resonant frequencies and DC resistance along with the inductance. Any suggestions?
 

Papabravo

Joined Feb 24, 2006
21,157
It is a mistake to think that a choke provides isolation. All it provides is an impedance that is a function of frequency. For a fixed value of inductance the impedance will be greatest at 30 MHz., and it will be the smallest at 150 kHz. Whatever impedance it provides to an AC signal it is insignificant for the DC power.

Looking at the datasheet tell me what the DC resistance and the self resonant frequency is for the 100 mH choke. You should also be able to work out the impedances at each of the frequencies.
 

AnalogKid

Joined Aug 1, 2013
10,986
To do what you want, a choke is the series part of an L-C lowpass filter. The shunt capacitor is critical in re-establishing a zero ohm AC impedance from the power rail to gnd for the oscillator to function. For high frequencies it might take two capacitors of different sizes (and possibly construction). The choke needs to have a low enough DC resistance that it doesn't create too much voltage drop at the current level of the circuit being decoupled, and high enough inductance to get the high freq attenuation you need. Both of these values (resistance and impedance) are set by the circuit being isolated. In other words, your call. For both the capacitors and the inductor, it is best to stay on the low side of the SRF (self-resonant frequency), where the components are behaving as expected.

At 30 MHz, a 1 mH inductor has an impedance of 188K. That's a lot. If you have to decrease the inductance to find a part with a SRF above 30 MHz, you've got lotsa room to maneuver. If your oscillator draws 10 mA and it can tolerate a 0.1 V drop from the rail, that's 10 ohms resistance.

10 ohms max. resistance
2 uH - 1 mH inductance
20 mA min. current rating

Digi-Key - 700 parts

ak
 

Thread Starter

Bill B

Joined Nov 29, 2009
61
Analog kid,

I'm not sure I follow what you mean by the need for an LC lowpass filter. Taking what Papabravo said, maybe isolation wasn't the term I should have used? Maybe suppression is a better word. Anyway, the thought is to run a 5V bus and a GND bus across the board. All of the 5V and GND connections will be starred together individually for each oscillator and then connected to the bus through a choke. The idea being that DC passes and AC is blocked or suppressed.I don't have access to the circuit schematic right now. I'll post it later once I have access to the computer where its stored.
 
Top