2
$\begingroup$

As the order of filter needed for digital information gets higher, it takes longer to process and requires more memory. Maybe analog filtering could be of use here. I know that the words analog and high precision dont often appear in the same room together, but lets assume they do for the sake of this post.

Ok, so I suggest that when a higher order filter is needed the following linkage of devices could be implemented:

DAC > Analog Filter > ADC

My question regarding this are:

  • Has this linkage of devices been used before?
  • If so, where can I find literature on it?
  • If not, why do you think it hasn't been used?

(Assume highly precise analog filters are being used)

$\endgroup$
3
  • $\begingroup$ Note that many high precision DAC and ADCs do a lot of digital processing internally (for linearity compensation, successive approximation, or noise filtering, etc.), requiring both time and memory to do this processing. Thus defeating the purpose of your analog filter chain. $\endgroup$
    – hotpaw2
    Commented Mar 8, 2016 at 18:21
  • $\begingroup$ Hmm...I guess this would almost defeat the purpose of briefly hopping into the analog domain...maybe a opti $\endgroup$ Commented Mar 8, 2016 at 18:46
  • $\begingroup$ Hmm...I guess this would almost defeat the purpose of briefly hopping into the analog domain...maybe a better bet would be to leave digital as digital, but rather analog-ize the operations needed for digital filtering $\endgroup$ Commented Mar 8, 2016 at 18:52

2 Answers 2

3
$\begingroup$

You wouldn't normally do that because it isn't worth the cost (hardware.)

FIR filters get computationally very expensive if you need sharp cutoffs.

Digital IIR filters that provide the same cutoff as a long FIR are computationally very cheap.

An analog filter will be equivalent to a digital IIR.

If an analog filter can provide the performance you need, then you are better off implementing a digital IIR filter. It saves you the DAC/ADC parts and it saves you the mess of an additional analog section with its inherent instabilities and potential to pickup noise.

If the digital IIR that can provide the performance you need is unstable, then it is very likely that the equivalent analog filter will also be very difficult to make work correctly.

$\endgroup$
1
  • $\begingroup$ Ok, well put. Informative response $\endgroup$ Commented Mar 8, 2016 at 14:39
1
$\begingroup$

There is an analog filtering technique that can be effective for narrow band filtering. It's call switched capacitor filtering. If you can switch at the center frequency then you can modulate the center frequency down to DC and then impose as narrow a low pass filter as you want; subsequently you switch it back up to the desired frequency; which doesn't have to be the same although it usually would be. This allows very narrow filtering in the analog domain at some cost in cash and additional noise. Incidentally the straight single pole filter at DC can be enhanced at will. The extra noise comes from the switching so you have to read the application notes and make sure you don't contaminate your signal beyond it's original noise level. As I recall Maxim has a line and application notes.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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