lowpass filter in modulation

[autumn2014]

hello
i'm using LABVIEW to implement LSB-SSB modulation, this requires a VSB signal passed through a lowpass filter, the problem i couldn't get my required spectrum, i wonder if the problem is in the filter or elsewhere, i hope you can help me in that , i only need to know how the filter in labview works , thanks

 here is the right attachment (the previous one was wrong sorry

[autumn2014]

hello
i'm using LABVIEW to implement LSB-SSB modulation, this requires a VSB signal passed through a lowpass filter, the problem i couldn't get my required spectrum, i wonder if the problem is in the filter or elsewhere, i hope you can help me in that , i only need to know how the filter in labview works , thanks

 here is the right attachment (the previous one was wrong sorry

[johnsold · Accepted solution]

As I mentioned in the other thread (and thank you for starting a new one), you need to post the actual VI with typical control values saved as default. We cannot tell what is inside the Express VIs from a picture.

Modulation is not done by adding two signals. Try multiplication.

It is difficult to get a sharp enough filter. SSB transmitters often use crystal filters.

Lynn

[autumn2014]

hi

that wasn't the right VI file sorry , i know it's multiplication that generates a DSB signal, you talked earlier about crystal filters i didn't find this kind of filters in labview, how can i implement SSB modulation? here is my VI

[johnsold]

A crystal filter is a hardware device which uses multiple quartz crystals to implement either narrow pass band filters or filters with very sharp passband edges. There is no close equivalent in software.

An elliptic filter has a very sharp rolloff near the cutoff but considerably less attenuation farther out in the stopband.

I put an elliptic filter (low pass) in your VI. The carrier frequency and the filter cutoff frequency are set to 500 Hz. The modulation frequency is se to 20 Hz. With the filter order set to 1 I adjusted the carrier amplitude for +30 dB on the frequency domain graph. Then I increased the filter order until the upper sideband is 60 dB lower than the lower sideband. The order to reach that point is 10. 

The carrier frequency component has started to come back as the filter order increases. For the10th order filter it is about 80 dB below the lower sideband. I am not sure why this is happening.

The filter transient lasts about 50 ms. This may result in some distortion of your modulating signal.

Lynn

[autumn2014]

thanks for your help, i really appreciate tht

but i don't understand how you designed that equivalent filter and the output in frequency domain was distorted, i mean it should be an impulse after or before the carrier frequency, i hope i could find an ideal lowpass filter to get the required spectrum

ps: in labview there is a template but it couldn't work with my software (2012) that's why i'm this lost

thanks, may you explain how u designed that filter? , if u do i'd be grateful

[johnsold]

The realities of real filters!

1. It is not possible to implement an ideal filter in the real world. *

2. All filters, hardware or software, have a transient when the signal is first applied (or changed).

3. The narrower the bandwidth or the sharper the cutoff, the more significant the transient.

4. All software filters have errors due to the finite size of the numbers used in the calculations.

5. Many filters and all spectral measurements are affected by the finite size of the data set provided as inputs. The mathematics of the Fourier transform assume the signal behaves the same for all time from -infiinty to +infinity. Since people are unwilling to wait infinite time to get results, errors occur.

How did I design the filter in the VI I posted?

- I selected a filter type -elliptic- based on my knowledge that elliptic filters can provide a very deep null close to the cutoff frequency.  I used the default passband ripple of 1 dB because the passband ripple is somewhat irrelevant in this situation and 1 dB is a small change for an audio signal. Then I experimented with the filter order until I got what I thought seemed like a reasonable upper sideband suppression. If I had detailed specifications for a filter, I wold have looked up the characteristics in a filter handbook to see if the selected parameters met all the requirements.

* You can sometimes approximate one by calculating the Fourier transform, setting all the undesired frequency bin values to zero, and then doing an inverse transform. However, if there is any spectral leakage, this technique does not work. In any case it is limited by the same numeric concerns in item 4 above.

Lynn

[autumn2014]

thank's , i'll study filters and design an almost ideal one for my design, uknow i didn't study these types of filters before, we only learned about analog modulation techniques  theoritically ,my project is to implement all types of modulation in labview.

thanks lynn

[johnsold]

The study of filters can be a large task because there are so many types.  Your library should have several books on filter theory and design. You might start by looking at a filter handbook. The handbooks will not have as much theory but will show examples of the most common typse of filters. They also typically have graphs showing the frequency response of different types of filters and different orders within each type. These graphs can be useful for getting an idea of the beahvior of different filter types without getting lost in the equations.

In software it is feasible to design filters of very high order, although you still have transient effects. In hardware high order filters are often impractical because of the amount of circuitry required and the effects fo the tolerance of available components.

Lynn