Opamps are your friend.... 1000x amplifier isn't a big issue.. would worry a bit about noise, but.... Good manual to get you started: http://focus.ti.com.cn/cn/lit/an/sboa092a/sboa092a.pdf The above appears to be devoid of filters... the following one has some filters included including a wide range of bandpass filters... http://courses.cit.cornell.edu/bionb440/datasheets/SingleSupply.pdf Searching for "opamp applications" or "opamp circuits" or "opamp filters" will get you going. In theory, you should be able to do this with 4 opamps... one to pre-amplify the signal and one for each frequency. You may also want to put a voltage follower on the front of each filter. I'd recommend you look at the LF147 if you want a cheap, but decent garden-variety quad opamp (aka 4 opamps in one package). I also really like the MCP6024 which is a more modern version with rail-to-rail ins and outs, but can't vouch for the slew rate figure (which affects the maximum usable frequency) since I mainly have been using it in a DC application. -forrest Bruno L. Albrecht wrote: > I know this can be a pretty beginner question, but I manage to design a good > low pass filter (3rd order, 2kHz) that has a good frequency response. The > major problem is that the signal has a ~100uV amplitude and still has to > pass through a highpass to lose all the DC. What should I do? I guess a > 1000x amplifier wouldn't be a nice idea for a low budget project... > > On Thu, Dec 3, 2009 at 10:07 AM, Vasile Surducan wrote: > > >> In analog will be quite complicated because you need 3 bandpass >> filter. The good news is that frequency are far away one from each >> other so you will not need a very sharpen filter (meaning higest >> order). One solution might be the variable transconductance amplifier >> where you can move the central frequency of the filter with a DC >> signal. >> >> What probably is not cristal clear, is that even with FFT you need an >> antialising filter in front of your A2D, else you'll see a lot of >> frequencies generated by Nyquist. >> >> Vasile >> >> On 12/2/09, Bruno L. Albrecht wrote: >> >>> Yeah, I thought a lot about doing it all in some microcontroller, with >>> >> FFTs >> >>> and Goertzel et al, but I'm trying to do it all in analog electronics. >>> >>> On Wed, Dec 2, 2009 at 6:05 PM, Marcel Birthelmer >>> wrote: >>> >>> >>>> On Wed, Dec 2, 2009 at 9:01 PM, andrew kelley >>>> wrote: >>>> >>>>> Use a discrete fourier transform. >>>>> >>>>> http://www.dspguide.com >>>>> http://www.dattalo.com/technical/theory/dtmf.html >>>>> http://en.wikipedia.org/wiki/Discrete_Fourier_transform >>>>> http://en.wikipedia.org/wiki/Discrete-time_Fourier_transform >>>>> >>>>> Andrew >>>>> >>>> If you're only looking for a single frequency, you don't have to do >>>> the full spectrum DFT... you can just calculate the frequency >>>> components in question. >>>> -- >>>> http://www.piclist.com PIC/SX FAQ & list archive >>>> View/change your membership options at >>>> http://mailman.mit.edu/mailman/listinfo/piclist >>>> >>>> >>> -- >>> http://www.piclist.com PIC/SX FAQ & list archive >>> View/change your membership options at >>> http://mailman.mit.edu/mailman/listinfo/piclist >>> >>> >> -- >> http://www.piclist.com PIC/SX FAQ & list archive >> View/change your membership options at >> http://mailman.mit.edu/mailman/listinfo/piclist >> >> -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist