-2
\$\begingroup\$

I'm having trouble understanding the concept of a Band-pass filter.

Not, that I have the following signal (in the frequency domain):

enter image description here

What I essentially need to do is remove the noise and only keep the highest frequency points (i.e. between 60-80 on the x axis) whilst removing the low frequencies. For this I need to use a band-pass filter, but, in order to implement such (coding project) should I:

1) use a low-pass filter AND 2) use a high-pass filter

If so, how would I therefore determine the cut off points to cut at in the signal?

\$\endgroup\$
  • \$\begingroup\$ Is this a software question or a hardware question? \$\endgroup\$ – Adam Head Jan 23 '14 at 16:43
  • \$\begingroup\$ "60", "80", and other dimensionless values are not measures of frequency. \$\endgroup\$ – Olin Lathrop Jan 24 '14 at 12:39
1
\$\begingroup\$

If you want a head start on trying various filters, this site is excellent for that. I picked the RLC band-pass filter and tweaked the values to give 70kHz centre frequency: -

enter image description here

I chose 70kHz because your graph was centred at "70" and I took a wild guess that you meant 70kHz but it won't make a difference if it's 70MHz BUT 70Hz requires something different entirely because the inductance would be too big - you'd need to use op-amp filters.

Anyway, at about 60k and 80k the response to an input signal will be about 6dB (half amplitude) down. There are plenty of filters to choose from so good luck in your hunting.

\$\endgroup\$
  • \$\begingroup\$ Thanks for the reply. Probably a stupid question, R, L and C you say you chose at random, is there a way/formula to be able to obtain these values from the signal? \$\endgroup\$ – Phorce Jan 24 '14 at 13:30
  • 1
    \$\begingroup\$ L and C define centre frequency at \$\dfrac{1}{2\pi\sqrt{LC}}\$ \$\endgroup\$ – Andy aka Jan 24 '14 at 14:30
  • \$\begingroup\$ smaller R makes bigger peak in response. \$\endgroup\$ – Andy aka Jan 24 '14 at 14:35
1
\$\begingroup\$

Yes, you want a bandpass filter. You can make a bandpass filter by combining a high pass and low pass filter.

However, your frequency range is quite narrow. That means something simple like a R-C high pass and low pass, or its digital equivalent, won't be selective enough. You are going to need a higher order filter.

Before you even think about the filter implementation, you need to define some clear specs. You have stated the passband, but left out everything else. For example, how much should the filter attenuate how far outside the passband? How much ripple can you tolerate in the passband?

\$\endgroup\$
  • \$\begingroup\$ Thanks for the reply. I was simply going to use the highest / lowest frequencies to determine where to filter.. Am I missing the point? Ok, I will start researching on higher order filters.. Note how this is just a test signal, the filter MUST work on other signals. \$\endgroup\$ – Phorce Jan 23 '14 at 16:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.