Frequency Measurements Accuracy

[Moh]

Good day everybody,

I am acquiring a 220V AC signal to a LabVIEW application after stepping it down to a 2V signal ( Simple resistance voltage divider ), and am using built-in VIs to measure frequency, amplitude, THD, and harmonics.

A question pops up here , what is the accuracy of my measurements ? specially the frequency ? this turns to be very important in my research ..

Can anybody help?

Thanks in advance

Moh

[iwanttofly014]

What hardware did you use? Pc sound card?In my opinion ,Accuracy lie on your hardware.

[Moh]

Actually i am using an NI DAQ Card PCI 6013

it can be found here https://www.ni.com/en-us/support/model.pci-6013.html

but i really think it has nothing to do about the hardware  .. i mean that i am acquiring the signal with enough samples and that might be the important thing.

I think that the built-in VIs are in question here.

[gwd]

What you are doing here is pretty scary because if you pick the wrong wire to use as your low you could potentially raise the reference terminal of your DAQ to 110 V.

We're talking about danger to the board, the computer, AND people nearby.

[Moh]

well , am not that ignorant

the cct is well implemented and there are enough " safety " steps before i take the signal . isolators exist to protect the DAQ card too.

Now , the main question is about the ACCURACY of my measurements . any help please?!

[johnsold]

The frequency accuracy is primarily determined by the accuracy of the timebase used to set the conversion clock. I did not look up the specs on your DAQ board, but that is where the information will be. It is often specified in parts per million (ppm) or percentage. The resolution is determined by the sampling rate. This may be limited by the hardware or by the rate chosen by the programmer. If you only take four samples per cycle, it will be difficult to determine the period or frequency well. The resolution and accuracy both may be limited by noise or other distorting factors, but that probably will not be a major issue in your application.

Lynn

[DFGray]

Your analysis method can also easily effect your accuracy.  For example, the tone detection VIs need at least 10 cycles to be very accurate (they are fourier transform based).  If you have a relatively clean waveform, you can probably get hardware limited accuracy by taking a waveform with about 100 cycles, finding the zero crossings by linear interpolation, then doing a line fit to the points to find the frequency.  The tone detection VIs should be as accurate, as well.  But don't trust anything.  Test your methods with known values, if you can.  Use two or three methods and see if they give the same result on the same data.  If it's important, make sure it works the way you think it does.

[Moh]

Lynn, DFGray

Thank you for your valuable inputs . I will take them into consideration.

I am trying to calculate it by various methods. I have made a very good idea about the issue

thanks one more time,

moh 

[chrisger]

hi there

first you have to know whats the INPUT of your measurement. So take another device you can trust in (calibrated scope or frequency analyzer or something like that..) and take some measurements directly at the signal source. repeat the measurements with same parameters with your setup.

then there are some values that can specify the accuracy of your measurements:

NEB (number of effective bits). this is the actual (amplitude) resolution in bits of your whole system including cables, A/D converters, software... by multiplying with the range you get a resolution in V or whatever. 

THD/THDN aka SINAD gives you information about the linearity of your complete system (cables, A/D converters, software..) search LV examples for SINAD or THD/THDN. non-linearity results in overtones of a sinus input signal. an ideal system has no overtones.

or search the NI knowledge base for "frequency accuracy" (https://knowledge.ni.com/KnowledgeArticleDetails?id=kA00Z000000P80PSAS&l=en-US)

[SSAC-Smith]

The LabVIEW VI that you need is "Buneman Frequency Estimator" found in the Signal Processing/Frequency Domain pallet. It provides a sine-wave frequency estimate that is good to three or four places under most conditions. There are better algorithms but this one is good enough for most applications and you already have it.

Of course, it does require accurate clocking as discussed in the other replies but it does a good job of handling non-integer waves/buffer and small numbers of waves/buffer.

..ss