sample count

Dear andurrahman0234,

> for the 1on clk and 36 bit input, number of sample count> should be  in the  0-100Mhz for power spectral analysis?

> I know that sample frequency 200e6 in this range. According> to this, I run the simulation to look pwer spectral density....‏

>  Can u help me, please?‏

I must apologize as I do not fully understand your question nor its context.

Perhaps a few questions might help..

1. Are you referring to a 10 ns period clock for a 36 bit wide data bus? You typed "1on clk and I am guessing you intended to type 10 ns?

2. While a 10 ns period corresponds to a fundamental frequency of 100 MHz, unless you are referring to a 10 ns period sinusoid, the frequency content may far exceed 100 MHz. What type of signal are you referring to?

3. Please also consider my item [2] with respect to your comment "I know that sample frequency 200e6 in this range.".

4. What is your desired frequency resulting for whatever signal you intend to simulate its power signal density and, as mentioned in [2] again, what is its expected frequency content?

5. Without knowledge of the items in [2] and [4], speculating on the appropriate number of samples required for an accurate estimate of a power spectral density is not possible - hence I can not comment on your statement "I think that minimum 1024 sample count is enough to look in psd".

In short, for me anyway, there is not enough information in your question for me to comment on your statement about sample count. Sorry!

Shawn 

İnput 36 bit digital bus at the 100mhz clkk frequency. İ want to look psd in the range of 0-100M interval. But plot has full of noise in result of Delta sigma modulator while getting output, i think correct circuit

Dear abdurrahman0234,

abdurrahman0234 said:

İnput 36 bit digital bus at the 100mhz clkk frequency. İ want to look psd in the range of 0-100M interval. But plot has full of noise in result of Delta sigma modulator while getting output, i think correct circuit

It appears you are only providing a partial response to my question [1]. My apologies once again. but speaking (writing?) for myself, I really cannot provide an accurate answer that will fully address your overall concern.

Nevertheless, it sounds as if the signal of interest is indeed a digital bus. Hence, it will have spectral content far above the fundamental frequency of  50 MHz (if the data is NRZ) or 100 MHz (if RZ). Hence, sampling it at  200 MHz rate will result in aliasing and will not provide an accurate estimate of its frequency content. You need to study the bandwidth of your 36 bit digital bus through a series of experiments to examine its harmonic content to determine an appropriate sample rate.

Shawn

all  of signals are digital, it is digital circuit. Intended frequency range 100 mhz to look psd. clk is 10ns.Output is as below. I dont understand, why noise is high?

Dear abdurrahman0234,

abdurrahman0234 said:

.Output is as below.

I don't see any attachment or find a figure if you intended to include it. You also still did not answer my remaining questions...I don't think I can help you unless you provide your thoughts on my remaining questions. Even if I see a figure, I will still not know your responses to my questions. In fact, I think if you consider answering them, it will provide you more insight on a solution to your observed PSD result.

Shawn

I referred digital bit signal output. Ut should be 150 mhz interval. İt doesnt matter interval. but psd look noisy. sample frequency 200e6 sample count 8192 transient time 40u

Dear abdurrahman0234.

From the small sample of the time domain response you provided, the frequency content of your signal far exceeds 100 MHz as the transition times have components of frequency much much greater than 100 MHz. Hence, sampling at 200 MHz will not accurately show the signal spectrum as high frequency components will be aliased down to low freqencies. Hence, the Fourier analysis you performed will not accurately represent the Fourier components of the signal.

Shawn

what ı can do?I noticed that it shoıws dB. I think that it should be dBm