filter

[desham]

thank you Lynn

[desham]

I tried to compare input sine wave samples to the output samples using digital comparision. I am getting an error. Please help me out 

[desham]

I tried to compare input sine wave to the filtered sine wave.To get number of samples that they differ. 

In analog signal we can take the percentage error, but in digital its looking complex

[johnsold]

1. The full scale signal at the input to your digitizer is Sum of the two signals.  The sine signal has an amplitude of 5 so its peak to peak range is 10. The sine plus noise signal has a sine amplitude of 5 or peak to peak of 10 plus a noise amplitude of 3 or peak to peak amplitude of 6.  So the full scale range at the input to the digitizer (which is specified in peak to peak) is 10 + 10 + 6 = 26.  You have the range set to 10.  Again you will be doing a substantial amount of clipping.

2. All filters have a transient at startup.  You can see some of the effect of this as the long flat region in the filtered sine.  Most filters also have a phase shift, meaning that the output will be delayed.  For many filters the phase shift varies with frequency.

3. Given 1 and 2, why would you expect the digitized outputs to match?

4. Is there a reason for putting the sine in twice?

Lynn

[desham]

Thank you Lynn

if i take sine wave of amplitude 5

noise of amp 1         total  peak to peak range is 12.  after filtering it will be it is peak to peak range 12

but now when i want to compare with original sine wave it is having peak to peak amp of 10 and output is 12. when i am clipping filter output to 10 to match the input sine wave its not working. Please help me out

[johnsold]

For consistency let's set all A/D and D/A conversions to the same full scale value (12).  Create one control and wire it to all the converters.  That will simplify comparisons.  If some of the converters have different full scale values the "weight" of a bit will be different and make comparisons quite complicated.

Use labels on the controls for the DWDT Digital to Digital comparison VI.  It looks like you copied some controls and they got confused.  The context help for the controls you had set to values of 16 suggested that they were resolution controls.  The context help for the comparison VI shows that they are the Start Sample index controls.  The other control is for comparison mode and its possible values are 0,1,2.  You had it set to 16.  To get diff data you need to set it to 2.

After fixing those things the result is still that all the samples are different.  This is what I would expect.  Some of the noise energy is within the filter passband so the filter output will have a random component.  The transient and phase shift characteristics will also change data enough that samples in general will never line up exactly and thus will show up as differences in the diff data.

Lynn

[desham]

what is the impact of noise and frequency on the system.

[johnsold]

It does not look like you changed any of the things I suggested in my previous post.  

Did you read my reply to Drew in your other thread?

It is better to keep all this in one thread.  Then anyone who is interested in the subject can keep up with everything which has been posted.

Please define what "system" you are talking about.  If you are trying to model a noisy digital transmission channel, the noise would typically be added after the A/D conversion.  If you are trying to model an analog transmission system, you typically would try to clean up the signal before digitizing it. You still have not explained the purpose of the hash and rsa subVIs in your test VIs.  They essentially do nothing.

Filters can reduce noise by a function of the ratio of the noise equivalent bandwidths (which is different from the 3 dB bandwidth) of the system up to the filter and the filter.  You need to be careful about units and impedances because the noise may be defined in terms of spectral density.  I will not attempt to explain this until you have given a good definition of what you are trying to model, because it is dependent on the model.

Lynn

[desham]

I am  trying to model an analog transmission system. In real time hash and rsa purpose is different but in this model i have been told like this. Is there any impact of noise and frequency on the system.

I think( noise will degrade the performance of the system.FIlter should be efficient to eliminate the noise.)

[desham]

hash and rsa, i have told to do like this.