Hi, My idea is simple but is it working in noisy channel ? I first send my IF signal to an ADC, then multiply it with COS and SIN from a CORDIC algorithm, then send theses two outputs to low pass filter to obtain I and Q signals. I send I an Q signals to another CORDIC algoritm to compute the angle information, average this angle information ? (for better perfromance in noisy cases) and send this angle (inverted in sign) to the CORDIC which generates the previous COS and SIN. So the previous multiplication is in phase with the IF signal. So I need to have the angle information faster than the rotating I and Q signals (due to phase incoherence beetween the transmitter oscillator and the receiver clocks that generates COS and SIN) Do you think it's a case used in digital demodulation ? And does it works fine ? Thanks... Best regards.
BPSK demodulation with CORDIC
Started by ●December 15, 2006
Reply by ●December 15, 20062006-12-15
patrick.melet@dmradiocom.fr wrote:> Hi, > > My idea is simple but is it working in noisy channel ? > > I first send my IF signal to an ADC, then multiply it with COS and SIN > from a CORDIC algorithm, then send theses two outputs to low pass > filter to obtain I and Q signals.Why is the source of sine and cosine important?> I send I an Q signals to another CORDIC algoritm to compute the angle > information, average this angle information ? (for better perfromance > in noisy cases) and send this angle (inverted in sign) to the CORDIC > which generates the previous COS and SIN. So the previous > multiplication is in phase with the IF signal.A low-pass filter averages inherently.> So I need to have the angle information faster than the rotating I and > Q signals (due to phase incoherence beetween the transmitter oscillator > and the receiver clocks that generates COS and SIN)I don't understand.> Do you think it's a case used in digital demodulation ? And does it > works fine ? > > Thanks... > Best regards. >-- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●December 15, 20062006-12-15
patrick.melet@dmradiocom.fr wrote:> Hi, > > My idea is simple but is it working in noisy channel ? > I first send my IF signal to an ADC, then multiply it with COS and SIN > from a CORDIC algorithm, then send theses two outputs to low pass > filter to obtain I and Q signals.You need neither sin no cos to demodulate BPSK.> > I send I an Q signals to another CORDIC algoritm to compute the angle > information,You don't need to compute the angle information. average this angle information ? (for better perfromance> in noisy cases)You don't need to average the angle information because it is already averaged by the lowpass filters. and send this angle (inverted in sign) to the CORDIC> which generates the previous COS and SIN. So the previous > multiplication is in phase with the IF signal.I don't quite understand. Are you trying to reinvent the differential detection?> So I need to have the angle information faster than the rotating I and > Q signals (due to phase incoherence beetween the transmitter oscillator > and the receiver clocks that generates COS and SIN)Neither sin no cos are required.> Do you think it's a case used in digital demodulation ? And does it > works fine ?I think this is a case when you need to get a textbook by Proakis or by Sklar. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Reply by ●December 17, 20062006-12-17
Why I don't need cosine and sine to COHERENTLY demodulate a IF BPSK signal ? I don't understand why ? If I want to coherently demodulate it I need to have sin and cos in phase with my IF, isn't it ? So I would extract the phase with a CORDIC and send the inverted sign angle to a COS and SIN to muliply them with my IF BPSK signal... thanks
Reply by ●December 17, 20062006-12-17
patrick.melet@dmradiocom.fr wrote:> Why I don't need cosine and sine to COHERENTLY demodulate a IF BPSK > signal ? > > I don't understand why ? > > If I want to coherently demodulate it I need to have sin and cos in > phase with my IF, isn't it ? > > So I would extract the phase with a CORDIC and send the inverted sign > angle to a COS and SIN to muliply them with my IF BPSK signal...How can the sine and cosine *both* be in phase with anything? They're 90 degrees apart. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●December 18, 20062006-12-18
OK I see what you want to say... Because it's BPSK I only need the I (In-phase) signal. Compute the angle with CORDIC (from I signal) and send a new COS to multiply with the IF BPSK signal. In fact, I want to use a BPSK pilot to demodulate a QAM signal... So when I have my BPSK phase information from the CORDIC, I'm going to send a COS and a SIN (with the COS in phase with my QAM signal) to obtain I and Q from the QAM signal... Thanks.
Reply by ●December 18, 20062006-12-18
patrick.melet@dmradiocom.fr wrote:> OK I see what you want to say... > > Because it's BPSK I only need the I (In-phase) signal. > Compute the angle with CORDIC (from I signal) and send a new COS to > multiply with the IF BPSK signal. > > In fact, I want to use a BPSK pilot to demodulate a QAM signal... > > So when I have my BPSK phase information from the CORDIC, I'm going to > send a COS and a SIN (with the COS in phase with my QAM signal) to > obtain I and Q from the QAM signal... >Do you know of a slow guy who was trying to build the basic FM discriminator for a half a year or so? VLV
Reply by ●December 18, 20062006-12-18
patrick.melet@dmradiocom.fr wrote:> OK I see what you want to say...Not exactly. "In phase" means zero phase angle. Sine and cosine can't both be in phase; they're pi/2 apart. I suspect that all you need to have is the oscillator *frequency locked* to the carrier. (Note the word "suspect". I'm not familiar with the details.) That means that any phase offset is OK, so long as it's constant. What I do know is that while CORDIC is a fine way to calculate trig functions, it's overkill for generating continuous quadrature signals. There are simpler quadrature oscillators. Clay Turner wrote a short paper on them. Google for it.> Because it's BPSK I only need the I (In-phase) signal. > Compute the angle with CORDIC (from I signal) and send a new COS to > multiply with the IF BPSK signal. > > In fact, I want to use a BPSK pilot to demodulate a QAM signal... > > So when I have my BPSK phase information from the CORDIC, I'm going to > send a COS and a SIN (with the COS in phase with my QAM signal) to > obtain I and Q from the QAM signal...Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●December 18, 20062006-12-18
yes i know this slow guy... now the fm demodulateur works fine with very good performance compared to an analog version... i'm beginner in digital radiocommunications... so I think I have stupid question, isn't it the aim of this forum to help or show the way to build components with very good performance... It's perhaps only your job to only helps beginner guys and not qualified guys... best regards
Reply by ●December 18, 20062006-12-18
patrick.melet@dmradiocom.fr wrote:> yes i know this slow guy... > > now the fm demodulateur works fine with very good performance compared > to an analog version... > > i'm beginner in digital radiocommunications... > > so I think I have stupid question, isn't it the aim of this forum to > help or show the way to build components with very good performance...Helping beginners is probably not the central focus of this group, but it certainly is a frequent activity.> It's perhaps only your job to only helps beginner guys and not > qualified guys...Nobody has a job here. It's all for pleasure, edification, and enlightenment. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
