Filtering RF from a DC singal?

[LiquidOrb24]

Hello All,

I'm having a hard time trying to solve a problem of mine. Basically I'm picking off an RF signal at a frequency of 13.56MHz and imputing the sinal into an IC that analyzes the data and shoots out a 1-1.8V (DC) signal corresponding to phase information on the RF line. However I want to input this (DC) signal into a microcontroller to analyze it further but I'm still getting a 13.56MHz signal tagged onto the DC signal.

I tried using a low pass filter before the microcontroller to filter everything out above 2KHz just to make sure but I'm still getting a small amount of noise on it. All of my grounding is good and am not radiating anything onto the DC side of the board where the microcontroller is but still nothing.

My next attemp would be to use an op-amp with either a low pass filter configureation or using an op-amp as a buffer because of the high input impedance.

Any suggestions on filtering an RF signal from a DC line...it would be most greatly appreciated.

 

[bananasiong]

Hi,
How low do you want the ac signal to be attenuated?

 

[Sceadwian]

You didn't mention what components you used to make the low pass circuit. How much of the AC signal is actually being imposed on your DC signal and how are you measuring it? Is it really a problem? meaning is the AC component enough to interfere with the micro controller?

 

[Speakerguy]

Ferrite bead and ceramic cap right at the input pin, after an active lowpass, and then do a quasi-averaging filter in the pic:

value = (new_val + 9 * value) / 10

done repetitively will de-noise the signal nicely. It should execute fairly fast if you use 8 bits and a PIC18F.

Pick the ferrite bead according to its Z curves. Same thing goes for the caps. Check out Wurth Electronics for the beads and Kemet's K-Spice to help pick out the cap so you notch out the problem frequency.

 

[LiquidOrb24]

I'm using ceramic caps and inductors to crate my low pass filter...the inductor as series and the capacitor as a shunt. C = 22uF and L= 220uH. I want the singal to be attenuated as short as possible before being filtered because it is going from the RF side of the board to the DC almost imidiately after the DC signal is goling out of the processing chip...I guess its not that big of a deal just hte microcontroller is taking this signal in and going through an A/D converter and first of all the microcontroller doesnt have the resolution to process that signal and I think it will just effect the processing of the DC voltage in order to get a nice reading.

Would it be possible to send it through a low pass filter first then an active low pass filter then a buffer or is that just being redundant?

 

[Speakerguy]

Are you using a 2 or 4 layer PCB? The C should be much smaller and the L would probably be better replaced by a ferrite bead, it might not be inductive at frequencies that high.

 

[audioguru]

An electrolytic capacitor attenuates frequencies up to about 800kHz. Above 1MHz it is not a capacitor but it is an inductor.

I have never seen a huge 22uF ceramic capacitor. It probably is also an inductor at the frequency you want to attenuate.

An active lowpass works up until the opamp runs out of gain. Most opamps have a gain of 1 (no gain) at 1MHz. Then it doesn't attenuate higher frequencies.

 

[Speakerguy]

Excellent point on the GBW AG. He could use a current feedback opamp to get around that, but components behaving like they should that high up is still dicey.

 

[LiquidOrb24]

Thanks for your help guys...I found some op-amps that may be able to work at that frequency and as far as the low pass filter I think I'll try using the active low pass using the op-amp. I appreciate the help

 

[crutschow]

Thanks for your help guys...I found some op-amps that may be able to work at that frequency and as far as the low pass filter I think I'll try using the active low pass using the op-amp. I appreciate the help

An active filter seems overkill for this task and may not work that well. I vote for ferrite beads and ceramic caps, 0.01uF or so (surface mount types directly a ground plane). Run the signal trace through the pad on the cap to minimize the inductance to the capacitor. Even a fraction of an inch of trace can have a significant inductive effect at that frequency.

 

[Sceadwian]

You still haven't mentioned exactly how much signal you're getting at the input to the micro controller, are you absolutely sure this is going to interfere with what you're doing? The input to most micro controller analog inputs have lowpass filters above the audio range built in. The 13mhz signal as long as it's amplitude is low enough shouldn't even be detected. Can you do scope screen captures or even just take a photo of scope so we know what you're looking at? This seems like a whole lot of analysis and solution finding for something that may not even be a real problem, just a perceived one.

 

[Nigel Goodwin]

The input to most micro controller analog inputs have lowpass filters above the audio range built in.

Sorry, never heard of any such thing - many will have a sample and hold, but it's not the same thing as a low pass filter.

 

[Sceadwian]

True enough, I thought it was a lowpass, apparently it's just a Nyquist limit of the sampling clock (at least on AVR, I think PICs are roughly the same) On a slow ADC clock in the MHZ range the sample and hold signal will average out to zero depending on the signal impedance and the sample and hold capacitor, effectively it will act as a lowpass with aliasing effects, this can be increased by adding more resistance to the I/O pin. Again it all depends on the amplitude of the imposed AC signal compared to the DC range required, and the sampling rate desired.