Sampling High frequency periodic signal

[dreamcatch16]

I want to sample a periodic sinusoidal signal ranging from 40Hz up to 5Mhz,
I need to sample at least 10 samples per cycle so the sampling frequency should be at least 50Mhz if doing a continuous sampling.
For you information I am using the PIC18f4550 which is a slow microcontroller and I want to prevent from changing it, also I can have a reference signal that is at the same frequency of the sampled signal.
My question is: Is there any techniques for sampling a high frequency periodic signal?
Or is there any high speed ADCs (10bits up to 16bits) with a programmable sampling frequency and an internal buffer memory that I can read it after taking the samples?

 

[aashishsharma]

yes you can find many ADC but sampling rate you need to see(sampling theorum). For 5Mhz you must atleast have sampling rate of atleast 10MHz. PIC crystal freq wont be same as ADC sampling frequency. How that ADC(if external) is getting clock will be deciding your output rate. Processor speed will be determing data rate for interface(say you are using SPI).
be more specific in asking your question
Aashish

 

[FvM]

Reading it thoroughly, the question turns out to be specific. But not asking for a reasonable solution, I fear.

Some questions in return
- if you imagine a 50 MHz ADC with internal buffer memory, which memory depth do you expect? You need more than a million samples to store a full 40 Hz signal period.
- what motivates the idea of having at least 10 samples per cycle of a periodic sinusoidal signal? Which information will you miss when sampling fewer points per cycle?
- why do you think needing to sample a periodical signal above Nyquist rate? A periodical signal can be fully reconstructed with equivalent time sampling (undersampling) methods.

 

[dreamcatch16]

FvM, that is why I specified that if the ADC have a programmable sampling frequency, such that I can sample low and high frequencies, also I want 10 sample because I want to use DFT to reconstruct my signal and get the phase shift from the reference signal, any suggestions?????

 

[FvM]

O.K., you mentioned programmable sampling frequency. That's unusual for a high speed ADC, you would typically select 1 out of N samples or use a decimation filter.

I'm not aware of a solution without implementing some digital signal processing in programmable or hardwired logic. I think that complex digital signal processing (e.g. fourier transform) comes into play if you're either analysing a time varying (modulated) signal or looking for harmonics. Phase and frequency of a periodical sine signal can be derived with much more basic methods.

 

[dreamcatch16]

O.K., you mentioned programmable sampling frequency. That's unusual for a high speed ADC, you would typically select 1 out of N samples or use a decimation filter.

I'm not aware of a solution without implementing some digital signal processing in programmable or hardwired logic. I think that complex digital signal processing (e.g. fourier transform) comes into play if you're either analysing a time varying (modulated) signal or looking for harmonics. Phase and frequency of a periodical sine signal can be derived with much more basic methods.

I know the frequency, what I don't is the amplitude and the phase, do you know any accurate and simple technique to get them?

 

[FvM]

Using a phase sensitive rectifier.

Have quadrature reference signals (square waves), use sign multipliers, average both products. The method is generally used in lock in amplifiers, vector voltmeters and vectorial network analyzers.
https://www.edaboard.com/threads/272103/
https://www.edaboard.com/threads/246986/

 

[dreamcatch16]

Using a phase sensitive rectifier.

Have quadrature reference signals (square waves), use sign multipliers, average both products. The method is generally used in lock in amplifiers, vector voltmeters and vectorial network analyzers.
https://www.edaboard.com/threads/272103/
https://www.edaboard.com/threads/246986/

I think that this technique is sensitive with noise and may be complicated for me to realize, that is why I am trying to do it digitally instead of getting the phase and amplitude analogically.

I read about equivalent time sampling, but my problem is the triggering circuitry.
if you know it or you already used it, what do you suggest as triggering circuitry for the ETS?

 

[FvM]

I think that this technique is sensitive with noise and may be complicated for me to realize, that is why I am trying to do it digitally instead of getting the phase and amplitude analogically.

I read about equivalent time sampling, but my problem is the triggering circuitry.
if you know it or you already used it, what do you suggest as triggering circuitry for the ETS?

The opposite is true. By multiplying the input signal with a +/- 1 function (sign multiplier) and averaging respectively integrating the product, the method is supressing noise effectively. Noise at harmonic frequencies will be attenuated according to the sin(x)/x filter characteristic of the rectangular window.

Any sampling solution, either real- or equivalent time needs accurate timing. Following your prefernce for digital solutions, you would do it in programmable logic.