For robust signal processing which is better to use 18 bit or 24 bit.

[infra444]

which is more advisiable to use :18 bit or 24 bit for robust signal processing. Basícally i want to write an algorithm for filtering the noises from an ECG signal.

[Ray.R]

The number of bits is related to the resolution of the sampling and may not impact robustness. 

There is a calculation that you can do to determine the one more appropriate for your measurement.  It is based on the voltage range (after signal conditioning).  You may do a search in the Knowledge Base, since I do not remember it (although fairly simple)...

Can you define what you mean by "robust"? 

Message Edited by JoeLabView on 04-16-2008 08:31 AM

[infra444]

Thanks for your reply. By robust i mean 'advance signal processing' , i already have an ecg signal which has lot of noise, so i want to remove those noises for which i am looking for DAQ device which can filter these noises.

[johnsold]

If you have a lot of noise, adding resolution will not help much. Mostly you would get six random bits appended to your 18-bit data. If you have a periodic signal and a good phase reference or trigger, averaging large numbers of datasets can reduce noise and higher resolution might be helpful. ECG signals are not periodic and not tied to a reliable trigger so averaging will not be very helpful.

It is much better to improve the recording hardware to eliminate noise before processing if possible. This usually means using high quality electrodes and skin preparation, differential amplifiers with good common mode rejection, careful avoidance of power mains frequency interference, and proper grounding and isolation techniques.

If you must analyze data which has already been recorded with poor signal to noise ratio, the best type of processing may depend on what types of analyses you want to do.

Lynn

[Ray.R]

I agree with Lynn.

The best place to start (if you are acquirinf signals) is to improve the signal conditioning.  It is best to work with a clean signal at the source than to try and clean it up using filters.  Using filters may (will) also filter out desired signals. 

So get rid of the noise at the sourse.  Noise = any unwanted signals.  Lynn provided very good recommendations in his 2nd paragraph.

RayR