Quadrature Demodulation at 90 kHz

[nauman]

Hello all

I have a linear frequency modulated (LFM) signal having 90 KHz center frequency and 20 KHz bandwidth around it. My problem is to demodulate it at base band to get I and Q both (phase information required). I have searched a lot at internet but mostly found quadrature demodulator ICs have starting input frequency range in MHz. AD8333 was what i found having lowest input frequency but even it does not work for 90 KHz (as i discussed with analog devices technical staff).

Can anyone kindly guide me how to solve my problem?.

Thanks in advance

Best Regards
Nauman

 

[Tom.G]

Use an analog multipier IC as a demodulator.

 

[tech99]

My problem is to demodulate it at base band to get I and Q both (phase information required).

Not sure what you mean by "demodulate it at baseband" when you say it is on a 90kHz carrier?

 

[nauman]

Use an analog multipier IC as a demodulator.

Thanks for reply. I require some pure sinusoidal oscillator for generation of sine and cosine signals at 100 KHz and also two identical low pass filters with linear phase.

Can you kindly suggest some ICs for them?

Thanks & Regards
Nauman

 

[nauman]

My problem is to demodulate it at base band to get I and Q both (phase information required).

Not sure what you mean by "demodulate it at baseband" when you say it is on a 90kHz carrier?

Thanks for reply. I have a LFM signal at 90 KHz carrier with 20KHz bandwidth. After quadrature demodulation, both I and Q should have frequency band from 0 to 10 KHz

 

[tech99]

You cannot have quadrature modulation with frequency modulation, only with phase modulation.
Have you worked out how to obtain or extract the 90 kHz reference carriers?

 

[nauman]

You cannot have quadrature modulation with frequency modulation, only with phase modulation.
Have you worked out how to obtain or extract the 90 kHz reference carriers?

Sorry if i am wrong, i think you mean "FM" as used in FM Radio but it is LFM ( linear frequency modulation) where frequency linearly varies with a time. Can you kindly explain further your first point?

Yes, 90 KHz quadrature sinusoidal oscillator (generating sine and cosine signals) is what i am looking for but could not found some ready made IC for it. Do you have some in mind?

Thanks.

 

[Jeff Rosenbury]

Is there a reason you need an IC solution? Discrete components should work at 90kHz. You might even get by with op-amp mixers, etc.

My guess is the reactive elements in the tank circuit are too big to work well with low frequency ICs.

 

[nauman]

Is there a reason you need an IC solution? Discrete components should work at 90kHz. You might even get by with op-amp mixers, etc.

My guess is the reactive elements in the tank circuit are too big to work well with low frequency ICs.

In my opinion, a good quadrature sinusoidal Oscillator IC (if it exist for 90 KHz) will cater for noise and phase errors for two sinusoidal. With discrete components, i will have to cater these issues.

Thanks

 

[jim hardy]

it does not work for 90 KHz (as i discussed with analog devices technical staff).

even though its datasheet says DC to 50 mhz? hmmm.

How about good old AD630 ?

 

[Tom.G]

For quadrature RC, Sine wave oscillator ckts see:

http://www.ti.com/sc/docs/apps/msp/journal/aug2000/aug_07.pdf

If you need higher stability, a crystal oscillator can be used and the quadrature phase obtained with a phase-shift network, as the quadrature RF demodulator chips use.

If you use a 4-quadrant analog multiplier for demodulation you may get better results overdriving the Local Oscillator inputs to saturation with a squarewave. One minor advantage to using the multiplier is you get both the I and Q outputs from the same package.

 

[nauman]

even though its datasheet says DC to 50 mhz? hmmm.

How about good old AD630 ?

Thanks. According to its data sheet minimum sine wave frequency at 4LOP pin is 2 MHz, however square wave frequency can be lower than 2MHz at this Pin. In your opinion, how much degradation (e.g. in terms of harmonics) in I/Q output will occur if i apply square wave instead of sinusoidal wave?

yes, I will study AD630 in detail and will bother you again.

 

[jim hardy]

l and will bother you again.

i'm in over my head already !

 

[sophiecentaur]

You cannot have quadrature modulation with frequency modulation, only with phase modulation.
Have you worked out how to obtain or extract the 90 kHz reference carriers?

I'm not sure what you mean by "cannot" when the only difference between FM and PM is that the PM modulat[Edit ing] signal is the Integral of the FM modulat[ing] signal. I agree that demodulation of a high deviation FM signal may not be convenient with a Phase Demodulator if the demodulator is only looking for phase changes within four quadrants but there is not a fundamental reason. Let's face it, the pre- and de- emphasis that's applied to FM sound signals is actually turning them into wide band PM signals.
Also, I guess that it's easier to use a quadrature modulator than a frequency modulator to obtain 'QM' signals because the reference phase is less likely to drift around. Nonetheless, in principle . . . . .

 

[tech99]

I'm not sure what you mean by "cannot" when the only difference between FM and PM is that the PM modulat[Edit ing] signal is the Integral of the FM modulat[ing] signal. I agree that demodulation of a high deviation FM signal may not be convenient with a Phase Demodulator if the demodulator is only looking for phase changes within four quadrants but there is not a fundamental reason. Let's face it, the pre- and de- emphasis that's applied to FM sound signals is actually turning them into wide band PM signals.
Also, I guess that it's easier to use a quadrature modulator than a frequency modulator to obtain 'QM' signals because the reference phase is less likely to drift around. Nonetheless, in principle . . . . .

Agree that in principle, yes. But you would have to start with the two data streams being applied to I and Q modulators, then regard the combined signal as FM. At the receiver, a pure FM demodulator, where output is proportional to instantaneous frequency, could not separate the two signals - it would require I and Q demodulation.
I am not sure how Nauman is going to phase lock the demodulating carriers to the incoming signal, which is essential. For example, with communication systems using 4-PSK modulation, the coding of the data stream is arranged to facilitate carrier recovery.

 

[sophiecentaur]

In the context of the most present forms of implementation, you are totally right but the only effect of phase error in the demodulator is to cause crosstalk between the channels, which could be eliminated with the right processing. I'm obviously being tongue in cheek here but, sometimes, there are novel methods of achieving results which seem loopy but which sometimes get taken up in mainstream systems. It's always worth while keeping an open mind.

 

[tech99]

In the context of the most present forms of implementation, you are totally right but the only effect of phase error in the demodulator is to cause crosstalk between the channels, which could be eliminated with the right processing. I'm obviously being tongue in cheek here but, sometimes, there are novel methods of achieving results which seem loopy but which sometimes get taken up in mainstream systems. It's always worth while keeping an open mind.

I do agree about the effect of phase error, but if the insertion frequency is not locked and is wrong then the subsequent processor stands little chance.

 

[sophiecentaur]

Yes. Of course, any conventional receiver needs to phase lock 'explicitly' and the lock has to be to the correct quadrant.

 

[jim hardy]

I wouldn't know how to meet that sinewave requirement .

Plain old PLL like LM567 will lock onto a sinewave but you get triangle or squarewave out.

Signetics' famous 1972 PLL application book is available as pdf here
https://archive.org/details/bitsavers_signeticsdcsPLLApplications_5800304

 

[meBigGuy]

Are you convinced or required to do this with analog techniques. By digitally sampling at 4x the carrier you can extract I and Q data streams.

EDIT: here is a link http://accelconf.web.cern.ch/AccelConf/p95/ARTICLES/RPQ/RPQ02.PDF

EDIT AGAIN: Here is an analog solution, maybe? http://datasheet.octopart.com/AD8339ACPZ-Analog-Devices-datasheet-5418633.pdf

 

[sophiecentaur]

I wouldn't know how to meet that sinewave requirement .

I don't think there is one. Harmonics will be away from the data frequency (of course) and the later filtering should take care of them.

 

[nauman]

Hi all

I greatly appreciate all of you for your help. Considering the available ready made solutions, AD8339 seems most suitable. The only confusion is whether we can apply 360 KHz (360/4=90 KHz) sine wave at its 4LOP pin or i have to go for square wave and then cater the harmonics at its I/Q outputs later.

 

[Tom.G]

According to pg. 18 of the data sheet, the internal LO logic is digital:

"Furthermore, the divider is implemented such that the 4LO signal reclocks the final flipflops that generate the internal LO signals..."

So you may as well drive it with a square wave or a pulse to reduce jitter.

 

[meBigGuy]

AD8339 4LOP input is designed for LVDS square wave (see the very first specification and the ref design in fig 62)