Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Filter bias (Read 8147 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Filter bias

Reply #25 –
> The only problem with that is that the waveform gets somewhat grainy.

The picture shows a crude connect-the-dots version of the samples, which has very little correlation with the actual waveform represented by the samples.

Filter bias

Reply #26 –
Just a suggestion, but how about transmitting multiple lower frequencies instead of one modulated frequency. You could implement one filter per frequency, tuned to pass just that frequency. Since you are transmitting over multiple frequencies simultaneously you could afford to use multiple cycles of each frequency per bit transmitted.

Filter bias

Reply #27 –
Quote from: lvqcl on
You can upsample your signals by 8x or 16x (with a good quality resampler) to see their wave forms.
Also, http://xiph.org/video/vid2.shtml (or http://www.youtube.com/watch?v=cIQ9IXSUzuM#t=358)


Thanks! But what would be the use? I know the waveforms of the continuous function are sines, I created the signal. I saw the video, I know there is only one sine function in that bandwidth going through those points, I just wondered how more difficult would it be to recognize a given phase inversion in such grainy resolution.

Well, maybe not much. Maybe for each cycle of points I could just test what the minimum squares is for a sin x and for a -sin x to decide the winner.

Filter bias

Reply #28 –
Quote from: pdq on
Just a suggestion, but how about transmitting multiple lower frequencies instead of one modulated frequency. You could implement one filter per frequency, tuned to pass just that frequency. Since you are transmitting over multiple frequencies simultaneously you could afford to use multiple cycles of each frequency per bit transmitted.



Now you are getting close to the book solution for this problem. What the OP wants is really just a voice telephone network modem such as was used with voice telephone networks to transport data between computers not that long ago. Everybody used them  before there were high speed (usually cable system based) data networks.  There are probably a jillion of these stowed away in attics and basements.

Filter bias

Reply #29 –
Quote from: pdq on
Just a suggestion, but how about transmitting multiple lower frequencies instead of one modulated frequency. You could implement one filter per frequency, tuned to pass just that frequency. Since you are transmitting over multiple frequencies simultaneously you could afford to use multiple cycles of each frequency per bit transmitted.


Yes, good point. I thought of that but at lower frequencies there is added environment noise. Ideally, it should work even when there is not very loud music playing in the background.

Filter bias

Reply #30 –
Quote from: ceticosp on
Quote from: pdq on
Just a suggestion, but how about transmitting multiple lower frequencies instead of one modulated frequency. You could implement one filter per frequency, tuned to pass just that frequency. Since you are transmitting over multiple frequencies simultaneously you could afford to use multiple cycles of each frequency per bit transmitted.


Yes, good point. I thought of that but at lower frequencies there is added environment noise. Ideally, it should work even when there is not very loud music playing in the background.

You would be implementing bandpass filters with a very narrow bandwidth, instead of a filter with enough bandwidth to accurately render your modulation. My guess is that would more than make up for working in a noisier frequency range. In any case you should transmit enough redundant information to correct for transmission errors.

BTW, I come from an era when 300 baud transmission over phone lines was the norm, and only the very lucky could afford 1200 baud.

Filter bias

Reply #31 –
Quote from: pdq on
BTW, I come from an era when 300 baud transmission over phone lines was the norm, and only the very lucky could afford 1200 baud.


I seem to recall hearing about people who made 300 baud modems work over relatively long distances using microphones and speakers as long range acoustic couplers.

Filter bias

Reply #32 –
Quote from: Arnold B. Krueger on
Quote from: pdq on
BTW, I come from an era when 300 baud transmission over phone lines was the norm, and only the very lucky could afford 1200 baud.


I seem to recall hearing about people who made 300 baud modems work over relatively long distances using microphones and speakers as long range acoustic couplers.

Not very pleasant to listen to - https://www.youtube.com/watch?v=3I2Q5-15Mic
 

Filter bias

Reply #33 –
Quote from: ceticosp on
Thanks! But what would be the use?


It would show you why the center frequency is irrelevant by making the lower and higher frequency images look equally good. 

Quote from: ceticosp on
I saw the video, I know there is only one sine function in that bandwidth going through those points, I just wondered how more difficult would it be to recognize a given phase inversion in such grainy resolution.


The "resolution" here is just an artifact of how you have chosen to plot the images.  What determines how easy it is to measure phase is the number of cycles you transmit (the inverse of symbol/data rate)  and the SNR.  If you signal is not being resolved, either increase the SNR or decrease the rate at which you transmit information.