GMSK, I, Q sample normalize
in [DSP]
|
Prev: According to their plan, the first 2*3.5TeV collisions will be performedthe 30-th of March. Im scared, but can not do anything, - all ways lead to catastrophe. LHC/CERN.
Next: Identification
From: Pat Z on 12 Mar 2010 13:15 hi, I asked this before and still don't quite make it right. Here are the I, Q samples datas I captured from the receiver. I tried 1)revert to binary offset by deducting 65536 if the value is over32767 2) calucate I^2 + Q^2. and the result is not flat. I was thinking (I/ 32767)^2 + (Q/32767)^2 should be 1. I sample 15943 14344 9326 11092 11853 1985 64725 62120 61503 8337 15911 14463 9602 8110 8347 8501 8399 8488 8388 Q samples 58604 59910 6512 5222 57206 50402 49840 53827 1906 4829 58638 59870 5436 7561 6727 6673 6722 6692 6726 6682 6705
From: Tim Wescott on 12 Mar 2010 13:25 Pat Z wrote: > hi, > > I asked this before and still don't quite make it right. > > Here are the I, Q samples datas I captured from the receiver. I tried > 1)revert to binary offset by deducting 65536 if the value is over32767 How do you know this is the right thing to do? > 2) calucate I^2 + Q^2. and the result is not flat. I was thinking (I/ > 32767)^2 + (Q/32767)^2 should be 1. Why? It's a _receiver_!! Try this simple though experiment: You have a receiver, designed to receive a signal which left the transmitter with a constant envelope. The signal is subject to fading and additive noise. The receiver design is such that the nominal signal is preconditioned to exactly fill up the entire range of the receiver output. But this preconditioning is based on following the signal amplitude, so it cannot follow rapid changes in signal strength. And there's that dang additive noise! What bad thing happens as a consequence of this receiver design feature? > > > > > I sample > > 15943 > 14344 > 9326 > 11092 > 11853 > 1985 > 64725 > 62120 > 61503 > 8337 > 15911 > 14463 > 9602 > 8110 > 8347 > 8501 > 8399 > 8488 > 8388 > > Q samples > > 58604 > 59910 > 6512 > 5222 > 57206 > 50402 > 49840 > 53827 > 1906 > 4829 > 58638 > 59870 > 5436 > 7561 > 6727 > 6673 > 6722 > 6692 > 6726 > 6682 > 6705 This doesn't look like the constantly-changing-in-phase signal that I would expect from GMSK sampled faster than the symbol rate. What is your sampling frequency vs. symbol rate? -- Tim Wescott Control system and signal processing consulting www.wescottdesign.com
From: Jerry Avins on 12 Mar 2010 14:55
Pat Z wrote: > hi, > > I asked this before and still don't quite make it right. > > Here are the I, Q samples datas I captured from the receiver. I tried > 1)revert to binary offset by deducting 65536 if the value is over32767 > 2) calucate I^2 + Q^2. and the result is not flat. I was thinking (I/ > 32767)^2 + (Q/32767)^2 should be 1. ... "Revert" means "go back to". If you want to go back to offset binary (I assume that's what you mean when you write "binary offset"), What is it before you go back? It seems likely from what you write that your ADC provides offset binary, and you want to convert to two's complement binary. Converting between two's complement binary and offset binary is done by inverting the most significant bit (and possibly sign extending), no matter which direction of conversion is wanted. Jerry -- Discovery consists of seeing what everybody has seen, and thinking what nobody has thought. .. Albert Szent-Gyorgi ����������������������������������������������������������������������� |