Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Bandwidth in GFSK

  1. Feb 14, 2014 #1
    Hello, I have a doubt how data rate is calculated in GFSK ? I am new to RF. so please describe.my second question is why can"t we change frequency faster for high data rate? a typical 433Mhz transceiver hardly gives 100Kbits only why?
     
  2. jcsd
  3. Feb 14, 2014 #2

    Baluncore

    User Avatar
    Science Advisor

    Welcome to PF.
    The FSK data must pass through the very narrow 433MHz channel allocation without causing co-channel interference. The Gaussian filter is probably being used to remove the broadband “key clicks” generated by the bit transitions.
     
  4. Feb 15, 2014 #3
    if i am not wrong, we need at least 2 frequencies in FSK .why can't we use two nearer frequencies for that lets say 433,000,005= 1 & 433,000,006=0 .can a frequency detector detect that much change?
     
  5. Feb 15, 2014 #4

    Baluncore

    User Avatar
    Science Advisor

    Yes it can, but with only a 1Hz difference you would have to sample for one entire second before you knew for sure which it was.
    Data rate is proportional to the bandwidth.
    1Hz BW is about 1 bit per second.
    100 kHz BW is about 100 kbps.
     
  6. Feb 15, 2014 #5
    First of all ,thanks for your immediate response and sharing your wisdom. in the above post you said i have to wait for 1 second to detect the frequency I understand that but is this possible to measure the wavelength so that even a single wave will give the frequency & i will not have to wait for that long.
     
  7. Feb 15, 2014 #6

    Baluncore

    User Avatar
    Science Advisor

    How do you measure the period, (not wavelength), to an accuracy of one part in 433 million and in the presence of system noise?
     
  8. Feb 15, 2014 #7
    433 MHz= 433 waves in 1 us===> 216.5 waves in 0.5 us.what if i take those 216 waves and calculate the time period from them hence frequency ? if it's impossible why? if possible it would give a huge data rate. since 0.5us for each bit then 2bits in 1us=>2Kbits in 1ms=>2Mbits in 1 second.it's still far better than 100kbits. Note these are my theoretical assumptions .
     
    Last edited: Feb 15, 2014
  9. Feb 15, 2014 #8

    Baluncore

    User Avatar
    Science Advisor

    Fundamentally, a radio channel has a bandwidth just like a low-pass filter has a cut-off frequency. The data rate through a low pass filter or radio channel cannot be greater than the actual bandwidth of that channel.

    To measure the frequency of a 1 MHz sinewave to one part in 1 thousand requires a 1 MHz * 1kHz = 1 GHz clock. Now a nice clean sinewave has a measurable frequency, but when you add other noise to that sinewave it is no longer possible to identify where the critical zero crossings are to within 1 nanosecond. To find those zero crossings you must fit a sine wave to the received signal, that will take one whole cycle per measurement. Detection is multiplication which is correlation of the received signal with a reference waveform.

    One bit every 0.5us = 2 M bits per second, which requires a 2 MHz bandwidth. Unfortunately the channels allocated on the 433MHz band are only 25 kHz apart, spread over 1.7 MHz. What happens to the other 68 users when your data interferes with their low data rate signals.
    http://en.wikipedia.org/wiki/LPD433

    If you want higher data rates then you need more bandwidth. That requires a microwave link or an optic fibre. There is a big difference between the low data rate links between distributed microcontrollers and an 11 Mbit/sec datalink on 2.45GHz.
     
  10. Feb 16, 2014 #9
    Thanks a lot for all your posts.will you please recommend me any books where i can read more about RF ,modulation & other RF related stuffs.Thanks again for clearing my doubts.
     
  11. Feb 16, 2014 #10

    Baluncore

    User Avatar
    Science Advisor

    These days, everything to do with modulation seems to have become digital signal processing, so that is now the best approach. Your experience, field of interest and mathematical ability will dictate the best reference.

    For now, start here and follow your interest.
    http://en.wikipedia.org/wiki/Digital_signal_processing
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted
Similar Discussions: Bandwidth in GFSK
  1. Bandwidth Requirement (Replies: 10)

  2. Capacity vs bandwidth (Replies: 29)

Loading...