extracting carrier signal

[gary36]

I have a signal with relation A sinw1t cos(phi-w2t). w1 = 50Hz and w2=1khz. I need to extract carrier from the product. I tried using Band pass filter with cut off 750 hz and 1500 hz. I am unable to get the second term properly. Tried different combinations, but with little success. Help required to get the right method of extraction.

 

[gary36]

Intention is to recover the value of Phi. The possible method is to compare a reference 1 KHz signal to the cos(phi-w2*t) signal and infer the value of phi implicitly.
Since carrier recovery was not so easy as mentioned in previous posts (for AM signal), I preferred some simple solution such as using tank circuit that could selectively pass on frequency and can have high Q compared to other implementations.

 

[BradtheRad]

A couple more methods to emphasize the carrier only...

Clip the original signal using anti-parallel diodes,
or
Apply unhampered gain to the original signal.

Such methods work better if modulation percentage isn't total.

 

[gary36]

Can you please elaborate your point mentioned in #22. I am not quite clear. Also I see that square wave generated from comparator is symmetrical whereas the source is not (discontinuity at zero crossings). Please clarify.

 

[FvM]

Intention is to recover the value of Phi. The possible method is to compare a reference 1 KHz signal to the cos(phi-w2*t) signal and infer the value of phi implicitly.
Since carrier recovery was not so easy as mentioned in previous posts (for AM signal), I preferred some simple solution such as using tank circuit that could selectively pass on frequency and can have high Q compared to other implementations.

You understand that in the modulated signal carrier phase is "switched" by 180 degree for each half cycle of modulation signal? This has two consequences:
1. The carrier can't be extracted by a simple filter, e.g. "tank circuit" because the filter output of one half cycle cancels the other half cycle.
2. The output of a suitable detector circuit, e.g. above presented Costas loop PLL will be ambiguous by an increment of 180°

A possible simple solution could be to frequency double the modulated signal with a full wave rectifier and filter 2*f2.

--- Updated Oct 6, 2023 ---

 

[BradtheRad]

Can you please elaborate your point mentioned in #22. I am not quite clear. Also I see that square wave generated from comparator is symmetrical whereas the source is not (discontinuity at zero crossings). Please clarify.

I suppose there is just a tiny amplitude of carrier present where we can't see it. And the op amp amplifies it. To me it's unexpected.

The simulator menu has a component called 'AM source'. Modulation percentage might only be 99 percent (though it appears to be 100 percent). We can edit the carrier frequency and fundamental frequency and voltage. It does not have an option to edit percentage of amplitude modulation.

It is normal for AM radio stations to modulate their carrier just partially. On my oscilloscope the broadcast carrier frequency is very prominent. It's obvious that the frequency is unchanging although the amplitude changes.

 

[D.A.(Tony)Stewart]

Intention is to recover the value of Phi. The possible method is to compare a reference 1 KHz signal to the cos(phi-w2*t) signal and infer the value of phi implicitly.

Phi, ϕ is an initial condition when t=0 and is also a constant like DC. The ONLY way to recover ϕ is with a stable 1kHz reference and a phase detector that is calibrated to ϕ=0 and phase locked to the carrier center frequency, f2.

Any other attempts will be erroneous. Any phase detector will do e.g. XOR gate as long as it is calibrated for ϕ=0 and obviously the frequency error is also 0, i.e. a phase-locked loop or an ultra-stable synthesized clock.

 

[D.A.(Tony)Stewart]

Here is my 15-minute design using the built-in Type-I phase detectors of a PLL (XOR) found in CD4046 IC's using the same VCO to modulate and demodulate. Just to give an idea of a cheap and dirty design.

 

[FvM]

I don't see how the 4046 detector is related to the discussed DSB AM signal with suppressed carrier.

 

[D.A.(Tony)Stewart]

I don't see how the 4046 detector is related to the discussed DSB AM signal with suppressed carrier.

forgive my oversight, I was thinking FM

--- Updated Oct 6, 2023 ---

I don't see how the 4046 detector is related to the discussed DSB AM signal with suppressed carrier.

I believe KarimF has the best design for AM-SC demodulator which he has used for many years.

Prj01 - A Simple Reliable Double Sideband Suppressed Carrier (DSB-SC) Demodulator

To those who are interested in AM communications, I attached below LTspice files which introduce the simplest reliable analogue AM demodulator. It works for all modulation indexes; from m=0 (no modulating signal) to infinity (no carrier). It uses one PLL (not two PLLs in quadrature as in Costas...

www.edaboard.com

 

[FvM]

I didn't attempt to present a solution for a design problem because no complete problem was yet given. I'm just suggesting principle solutions and fighting various misunderstandings.

Thanks for remembering the DSB SC thread. Carrier recovery is only part of DSB demodulation function, thus you can hopefully save part of the effort. The principle idea of Prj01 carrier detection (frequency doubling and filtering of 2*fc) is also implemented in post #24. Using a basic PLL in place of the filter can be step to a practical solution.

 

[gary36]

My exact problem is to extract the value of Phi. And perhaps I am trying to evaluate several options after reading posts here. Therefore my primary question was to extract carrier thinking that it could be one of the plausible solution. I was also wondering about the response time of DSB-SC demodulator?

 

[FvM]

There's a trade-off between response time and frequency range (in case of PLL solution, separate pull and lock range) as well as phase jitter/resolution.

 

[gary36]

Finally I got to the root. Originally this problem was shared to me by a young researcher and wanted all analog solution. After a detailed discussion, he came up with the patent document that specified this particular problem and possible solution. I have attached the document. Please refer to Fig.3 and page 5. I did not quite understand the implementation though and still searching. Require a final call from the forum experts.

 

[FvM]

Sounds like reversal of the original problem specification "extracting the carrier".
Carrier is known because it's generated by the circuit itself. All you want to do is to determine the carrier phase in a derived signal. May be you had this in mind, but (at least in my perception) didn't tell clearly.

Other than suggested by post #1, signal of interest (e.g. summed chopper output in US3866214) is assumed stationary during measurement cycle. Thus it can be bandpass filtered and passed to regular phase measurement.

US3866214 doesn't look really exciting to me. I doubt that it has been competitive at the time of origin, now we have resolver-to-digital chips off the shelf.

If the problem is phase measurement relative to known stationary carrier, I would look at quadrature detector methods in the first place, starting will simple XOR dector.

 

[gary36]

I was not clear with the intention until discussed with the researcher in detail. But you had indicated earlier that bandpass filter will not work effectively for AM signal and secondly you are suggesting a XOR detector. Little bit on confusing. Could you please explain with the implementation circuit?

 

[D.A.(Tony)Stewart]

A sinw1t cos(phi-w2t).

Without a universal time reference to measure "phi" when t=0, both at the transmitter and receiver, there is no possible way to recover it. Only changes in phase which is w2t can be demodulated. There are many static delays which affect the static constant of phi, including propagation delay & filter group delay.

This question cannot be answered to measure the exact phase when the signal started at some arbitrary t=0 without an external time reference.

--- Updated Oct 11, 2023 ---

An XOR gate as a common phase detector to compare two coherent clock frequencies with a resulting f1-f2 and f2&f2 where the duty cycle is phi assuming f1 is the input equals f2 as a VCO controlled by the negative feedback of the XOR phase voltage filtered to some bandwidth. (This is >50 yr old technology)

Phase Detector: Digital Analogue Linear Mixer » Electronics Notes

Check out this this phase sensitive detector tutorial: circuits; operation; types; analogue; digital; linear; mixer . . .

www.electronics-notes.com

https://web.ece.ucsb.edu/~long/ece594a/PLL_intro_594a_s05.pdf

This is just a primitive 4046 type II PLL and showing the XOR mixer. (Type I) with sliders for signal f, and phase.

https://tinyurl.com/yttwzblf

 

[gary36]

A very silly proposal. What if I divide the modulated signal by the message ( If I have one). Will I not get back the carrier.