High Frequency Harmonics after LPF

[sk614]

Hi I am trying to filter a 7MHZ signal using RLC-LPF.  My signal flow are as follows.
I am taking in a 6.78Mhz AC, rectifying it in a full wave rectifier. I then send it to a LPF (around 7Mhz; R = 430, C = 100pF, L = 4.7uH.) My problem is I still have high frequency harmonics around 14Mhz to 140Mhz.  I compared the signal with and without the LPF, and difference are very subtle. 

[suicidaleggroll]

I compared the signal with and without the LPF, and difference are very subtle.

That's not very descriptive.

Filters are not brick walls, they simply attenuate.  An RLC filter is 2nd order, which has a rolloff of 12dB/octave (40dB/decade).  If it has a cutoff of 7M, then it will attenuate any signal at 14M by 12dB, and any signal at 140M by 52dB.  What values are you seeing?

[Kleinstein]

Rectifying a 7 MHz signal ideally gives DC plus 14 MHz plus a few more harmonics of the 14 MHz.

[f5r5e5d]

a sim could be illustrative, free LTspice is popular, powerful

[uncle_bob]

Hi I am trying to filter a 7MHZ signal using RLC-LPF.  My signal flow are as follows.
I am taking in a 6.78Mhz AC, rectifying it in a full wave rectifier. I then send it to a LPF (around 7Mhz; R = 430, C = 100pF, L = 4.7uH.) My problem is I still have high frequency harmonics around 14Mhz to 140Mhz.  I compared the signal with and without the LPF, and difference are very subtle.

Hi

What you have built is a classic frequency doubler. It will have a *lot* of energy at higher frequencies.

R/C filters are rarely seen at HF. Why? because they don't roll off very fast. You don't say how your parts are hooked up, but at 7 MHz:

4.7 uH = 206 j ohms
100 pf = 227 j ohms
R = 430 ohms

Those values don't look a lot like a normal lowpass filter. What is the driving impedance (if not 430 ohms)? What is the load impedance? (if not 430 ohms).

====

So what are you trying to do?

If you want DC, then set your filter to a much lower frequency. 4.7 mHy and 100 nF might be good bets.

If you want the ~7 MHz, don't rectify in the first place.

If you want the ~14 MHz, use a bandpass filter rather than a lowpass.

Bob

[sk614]

Thanks for all the replies! What I am trying to accomplish to rectified a 6.78Mhz AC signal that I am receiving from an coil antenna (via magnetic resonant coupling) and then convert it to DC.
I am connecting my RLC in series. I just want to convert the AC to DC and have been able to do that. However my dc signal is very noisy and after looking at the FTT on the  oscope I have a lot of high frequecy harmonics.

In truth I don't even know if I need to keep the 6.78Mhz. Can I filter at lower frequency?

[danadak]

Grounding and bypassing everything Mhz signals and above.
Not all caps equal in their f vs z curves, take a close look at the
datasheets for the specific parts you are using. Use combination
of ceramics and tants, use polymer tants for order magnitude
better f vs z over regular tantalums.

Look carefully at scope probe ground, where you are picking up
switching noise.

Analog and digital grounds should be split, and meet at board
edge.


https://www.dropbox.com/s/2h96beh1fbvz4e2/noise_notes.zip?dl=0

https://www.dropbox.com/s/ruaf9booe17jk8n/PCB%20Layout.zip?dl=0


Regards, Dana.

[uncle_bob]

Thanks for all the replies! What I am trying to accomplish to rectified a 6.78Mhz AC signal that I am receiving from an coil antenna (via magnetic resonant coupling) and then convert it to DC.
I am connecting my RLC in series. I just want to convert the AC to DC and have been able to do that. However my dc signal is very noisy and after looking at the FTT on the  oscope I have a lot of high frequecy harmonics.

In truth I don't even know if I need to keep the 6.78Mhz. Can I filter at lower frequency?

Hi

If all you are after is the DC, then *no* you do not need to / want / wish to hang on to any of the ~7 MHz. The normal way to do it is to use bypass caps (0.1 uf X7R ceramic caps) as your cap to ground. You may use more than one if you have a lot of energy involved. For your series device over to the rectifier, use a coil and no resistor. Something in the hundreds of uH should do fine. Like the caps, the value is non-critical as long as it is "big enough. With proper grounding and layout, you should see very little AC on the cap(s).

This of course assumes I have understood your application.

Bob

[mikerj]

Thanks for all the replies! What I am trying to accomplish to rectified a 6.78Mhz AC signal that I am receiving from an coil antenna (via magnetic resonant coupling) and then convert it to DC.

Are you trying to do wireless power transfer (like a Qi charger) rather than trying to demodulate an RF signal?

[sk614]

Are you trying to do wireless power transfer (like a Qi charger) rather than trying to demodulate an RF signal?

Yes I am doing wireless power transfer (A4WP standard) , my DC signal is not clean still having depends on distance around 250-400mVpp.  I want to make it clean as possible. 


This is what I am rectifying (ignore the 2.4Ghz section).

[suicidaleggroll]

Two questions

1) If all you want is DC, then why are you setting Fc to 7 MHz?  You want Fc to be as low as you can get it.
2) If you're doing power transfer, then why do you have a 430 ohm series resistor?  You can't transfer any significant power through a resistor that big.

You need to lose the resistor and throw a bunch of capacitance in parallel and possibly some inductance in series.

[uncle_bob]

Are you trying to do wireless power transfer (like a Qi charger) rather than trying to demodulate an RF signal?

Yes I am doing wireless power transfer (A4WP standard) , my DC signal is not clean still having depends on distance around 250-400mVpp.  I want to make it clean as possible. 


This is what I am rectifying (ignore the 2.4Ghz section).

Hi

Consider that you may need to optimize both your transmit and receive  coils. That may involve resonating them at your desired operating frequency. It also could involve changing the number of turns or the spacing.

Bob

[Kleinstein]

If you just want the DC, you can set the filter to a lower frequency. Usually the filter should be more like higher order LC - no intentional R as you don't care about ripple. It might even be useful to have a kind of chauer type (e.g. capacitor in parallel to the inductance if the parasitic capacitance is not enough) filter thus extra strong damping at 14 MHz.

What you measure also depends on the way you couple the scope - not every point that looks like ground is a good HF ground.

[uncle_bob]

Hi

Ok so some very basic filter math:

Each L or C in a simple filter *should* drop the harmonics by 20 db per decade once you are well past the filter cutoff. If you intend to have a carrier that is 2X the filter cutoff, you might need 20 elements in the filter to get 120 db of filtering. That's a lot of L and C. You also need pretty good precision in all those parts to get it to work right.

The other approach is to observe that 120 / 20 = 6. If you pick your filter cutoff frequency to be 4 decades below your carrier, you can use a single element filter. In this case, the filter for your ~ 7 MHz carrier will be at 7 Hz. With two elements, it would be at 120 / 40 = 3 and have a cutoff at 7 KHz.

120 db is a very arbitrary choice of attenuation. It is unlikely that your ground structure / board layout is good to 120 db. It also is going t to be quite hard to find a single element cap or coil that gives you 6 decades of performance. The intent here is not to pick an attenuation number, only to show what you do after you *have* that number.

Bob