Calculate the time out of sample points with set frequency

[Kurzma]

Hi all,

at the moment I am doing my Master Thesis and have the following problem.

I am trying to measure Data and asign it a proper timestamp. My problem is, that the data is coming in bursts and the timestamps I assign with the software are wrong.
The controller I am using for monitoring the data should run at 100 Hz, but it doesn't. I checked the Data while applying a sinudial 100 Hz signal and there appeared constant peaks every 10 datapoints and a sinudial slow drift, in contrary to the expected horizontal line.

Now I am trying to determinate the delay in aquiring the data (peaks in pointdata) So we got the equation:
f(t)=Asin(wt) + C ?
f ~= 100 Hz
w = 2

f
A= data acquired
C= could be calculated

and somehow just calculate t ? But it doesn't feel right...

Also I came up with another approach:
f(t_x) = x = Asin(wt + O)
f(t_y) = y = Asin(wt + d + O)
x - y = Asin(wt + O) - Asin(wt + d + O)

I would really appreciate any help with this. I am a real doofus when it comes to calculus especially with harmonic oscillators. I browsed the whole day trigonometrical identities and didn't really progress far.

I hope I explained everything properly, If anything is unclear please feel free to ask me.
I also uploaded an sample from my data.

Best Regards

 

[.Scott]

I'm making some guesses here, but this seems to be what you are describing:
You are collecting data from some source - apparently a signal that can be replace with a sine wave generators.
The data you are receiving at the software end is not only being received in clumps, but is apparently being collected in clumps. So that sine wave isn't looking as smooth as it should.

My guess is that the data collection rate is being controlled through software timing - rather than a regular series of hardware-generated pulses.

For example, if you are collecting data through an analog-to-digital converter, the sampling signal to the ADC should not be driven directly by the software. It is okay for the software to have some executive control over the signal (start, stop, frequency/period selection), but it should not be generating the timing pulses themselves.

 

[.Scott]

Let me add something: The experiment you're doing with the sine wave is good - assuming that the period of the sine wave is slower that the sampling period - but not a whole lot slower. Slower by a factor of 3 to 12 would be fine. I would collect 16384 samples and apply an FFT (using, for example, MatLab) and then check for a very strong spike.

But I am also presuming that you have already concluded that the data is not being collected at regular intervals. That should be all you need to know. Determining the delay might provide a useful diagnostic for finding a problem in a system that should work.

You haven't told us what data collection system you are using and why you think that system should work. So the first place I am looking is at the basic design of you data collection system.

 

[.Scott]

The controller I am using for monitoring the data should run at 100 Hz, but it doesn't. I checked the Data while applying a sinudial 100 Hz signal and there appeared constant peaks every 10 datapoints and a sinudial slow drift, in contrary to the expected horizontal line.

I looked at your data and reread you post.
I am not sure what you want. Have you concluded that the controller is not collecting the data correctly, but you want to continue using that controller anyway?
Normally you would fix your data collection system and recollect the data. Perhaps you cannot recollect the data?