Can we control the frequency of signal acquired by by USB-DAQ ?

[awais.h]

Hi,

I am acquiring two signals from my USB-DAQ 6008. First signal is a sine voltage with a frequency of 48.5<f<51.5 and the other signal is a pulse (square wave) voltage with 48.5<f<51.5

Now the problem is that I need to have both of the frequencies locked at one particular frequency between 48.5 and 51.5 (preferably 50Hz) in order to perform my calculations on the signals. The frequencies of both signals keep changing (within the limit) over time. I know it sounds very silly but is there any way that I can make the DAQ assume that the frequency for each signal is 50Hz?

Any help would be greatly appreciated

Regards,

Awais 

[PauloM]

Hi,

I don't know if that's what you mean, but the DAQCard cannot change the frequency ofyour acquired signals.

Why do you want to do that?

Is there any programming issue that has to do with the frequency of the signals that you want to do?

Please detail a bit more.

Paulo

[awais.h]

Hi Paulo,

My application is for an electrical generator. I am feeding the terminal voltage (AC) to the DAQ (after stepping down of course). The generator is synchronized onto a grid with a freq of 50Hz but the actual frequency changes upto + or -3% i.e (48.5-51.5 Hz).

My sensor is attached to a fixed point at stator and it sends a pulse every time a fixed point of rotor pole passes the sensor with the same frequency problem.

My labview program measures the time difference between the rise time of pulse and zero crossing of the sine wave and that time difference tells me how much load there is on the generator (with the help of some equations and formulas)

Ideally, if the frequencies of both signals are 100% exact, there would be one complete cycle of sine wave and one pule in one time period (i.e 1/50 = 0.02 sec) but if the frequency changes even a little, there could be more than one pulse in one complete cycle of sine wave that would give me 2 time differences hence two calculations for load.

I was wondering if I could somehow add another filter of some sort that collects the signals from daq and then converts them into a fixed 50 Hz signal. Is that possible or am I just going crazy due to only 2 hours sleep since last two days.

Regards,

Awais 

[PauloM]

Hi Awais,

So, the problem is, you are acquiring 20 ms of data, that would fit perfectly in one cycle only if the frequency is 50Hz, right?

When the frequency is higher, you may have two pulses and that may confuse your processing, right?

When the frequency is lower, you may not have any pulse there, isn't this a problem?

Do you have to measure in every cycle?

Of not, if you adjust your sampling time to mach 1 cycle of the higher frequency you may have(1/51.5), in some cycles you would not be able to measure, but you would never get two pulses in one sampling period.

Other option would be to adjust your number of samples to the last acquired frequency measured. If the frequency doesn't change very fast you would have a good idea of your actual frequency, by the last measured.

Another option would be to adapt your processing code to not get confused with two pulses on the same sampling info.

Hope this helps,

Paulo

[JÞB]

I think I see a way out of this.  Waveforms  of exactly 0.02 seconds duration may have 0,1 or 2 counts if you are counting from line Max to line Max and line Freq is 50Hz+/-3%

We need to attack the 0 case first (not enough data to calculate) before working on the 2 case (too much data).  There is only one way to do this.  Extend your waveform duration to cover the worst case Always take TOO MUCH data (We'll through out the excess later ).  So your acquisition period should cover the worst case PLUS a bit of engineering overhead 50Hz - 10% = 47.5Hz so take 0.021 seconds of data.  At this point you are garunteed to have at least 3 zero crossings in your waveform (unless the line freq is sevely out of spec).  Find the times of the zero crossings and truncate both waveforms by taking only the subsets from the first to the third zero crossing.  You have exactly one period of data and you can calculate freq, phase, load etc.  

Note that I said this works only if the line is not severely out of spec.  A "special case" if the 3 zero crossings cannot be found  an error exists.  Generate an error "-42 Low Line Frequncy"  Somebody needs to know the system has a fault!