Ferrites before or after voltage regulator?

[mrkenneth]

I am working with some RF circuits (up to 2 GHz) and need a clean power rail. I need to reduce 5 V to 3.3 V and have everything as clean as possible. The 5 V input is powered from two Li-ion batteries and is regulated by a normal 7805.

Is it better to place a ferrite before the 3.3 V LDO regulator, or after the regulator? The current will be under 20 mA.

Thanks in advance!

http://www.mrkenneth.com

 

[itsmoked]

mrkenneth,

You aren't going to get any "noise" out of the batteries followed by a linear regulator. So don't bother with a choke if this is the concern. Now if you are hooking this regulated supply to RF circuitry the general feeling is to prevent the RF from leaking back from the RF section, so you would put the choke in the power lead from the regulator to the circuit.

 

[mrkenneth]

Thank you for the response.

The battery is connected to a PIC microcontroller, but the RF circuit is connected to the PIC board through a long, unshielded cable. I am worried about the cable acting as an antenna (because it will be in a noisy environment, and I will be measuring RF). The PIC board may also be producing noise on the 5 V rail.

I may also be powering the circuit from a noisy power adapter (in cases where I do not have a battery handy).

The 3.3 V LDO regulator will be on the same board as the RF circuit that it is powering. Just as a precaution (a inexpensive one at that), should the ferrite go between the cable and the input of the regulator, or between the regulator and the circuit?

Thanks in advance!

http://www.mrkenneth.com

 

[Comcokid]

I would use the filter on the cable at the point it attaches to your RF board. What you want to do is keep the RF on your RF board, and keep the cable from acting like an antenna. Use small capacitors (say 33pf or less SMT types) on each signal line to ground at the point that the cable attaches to your RF board.

On the PIC end, make sure you have more than one bypass on the 3.3 V and use several capacitors about two decades apart in value to keep switching noise, etc from traveling on your VCC. Example: if you have a 10 uF as your primary filter, make sure you also have at least one 0.1uF and one 1000 pF.

If you are working with stronger RF signals, you may want to filter/decouple each end of your long cable otherwise RF will directly radiate into your cable as well.

 

[itsmoked]

mrkenneth, You have not stated what kind of RF you are interested in and why. With more info we can probably give you some pointers that could help head of a lot of hassles with your design.

 

[mrkenneth]

The RF signals should not be too high; I am just measuring cellphone radiation. The more important thing is that the measurements are as accurate as possible.

I do not think I can use SMD capacitors on the PIC board, the entire RF board will be surface-mounted. The power rails will be like this:
7.2V -> 7805 -> 5.0V -> PIC
(separate board): -> 3.3V LDO -> RF

The 3.3V LDO regulator will be on the RF board.

I think I may have a 5.5V supercapacitor on the PIC board, in case the batteries lose contact for a short period of time.

Thanks for the responses!

http://www.mrkenneth.com

 

[itsmoked]

mrkenneth,
Ohhhh kay.... Cellphone freqs are high...real high.

Generally when doing RF circuits like this you would put the power supply and the PIC in a box and then the RF section in another box. Chokes would separate the two boxes. There would then be an antenna exiting the RF section. Often this is all on one flat substrate like bare circuit board, with other chunks of circuit board on edge and soldered between sections. This makes a bunch of little walls with circuits confined in their proper areas. This would look like a giants view of a single storey home with the roof removed. This is how you make a prototype or three. The production form would, these days, be a plastic molded box with an RF shielding material sprayed on the walls to provide the electrical shielding. Alternatives include a standard board with RF sections under little sheet metal boxes that are soldered down on all four sides. There are whole companies that make these little boxes. The very best example of how to make your gizmo would be to tear apart one or two different makes of cellphones. Cell phones have exactly the same requirements you have and are now 8th generation designs that have cost $$$$$ to design, so you should see what works.