Is the sampling rate, 16384 Hz a special number for Signal Express or BNC 2110 and NI USB-6251?

[rico1985]

Hi,

Is the sampling rate, 16384 Hz a special number for Signal Express or BNC 2110 and NI USB-6251? 

The following are my test setup.

Generated Signal: Sine wave

Frequency: 200Hz 

Sampling Rate: 16384Hz

Block size: 4096

The output channel is connected to two input channels directly. And power spectrum steps and  frequency response steps are used to get phase, magnitude, amplitude and coherence. At this specific sampling rate, the coherence is a mess and different from what I got other tests using other signal analyzer. I am wondering what's the reason.

And the frequency steps in the power spectrum are always different what they are supposed to be, which have frequency shifts. Thank you.

[Doug_L._Bear]

Is it just with 16384, or are frequencies above this one also adversely affecting your measurements?  Or is it just in the general region of 16384?

[rico1985]

Hi Doug,

Thank you for your reply. 

It's a small region around 16384 Hz. When the sampling rate was changed to 17 kHz, 15kHz or other values, I got perfect coherence, phase and magnitude.

Thanks, 

[Doug_L._Bear]

I haven't been able to reproduce your issue on this end.  I have a slightly different DAQ card, I will continue to try and update you tomorrow.

[rico1985]

Doug,

I did the test for several times and got the same results. Did you find anything so far? The attached is my Signal Express project. Thank you.

[Doug_L._Bear]

Because I am not seeing the same thing that you have been able to reproduce multiple times, could you attach a graph of your coherence both at a "good" frequency and at this bad 16,384 Hz.

[rico1985]

Doug,

The following are all of the tests I did.

With no averaging (Sometimes signals look bad.)
Channels; Samples;Sampling Rate (Hz); Magnitude(dB);Phase (deg); Coherence;Time shifts(second);
1&0              4096              16384                   See Fig.1       See Fig.2             1                     -
1&0             24000             24000                  0 +/- 5m         0 +/- 50m             1                      -
1&0             50000             50000                  0 +/- 5m         0 +/- 50m             1                      -
With RMS averaging
Channels; Samples; SamplingRate (Hz);Magnitude; Phase; Coherence;Time shifts(second)
1&0             4096              16384               See Fig.3 See Fig.4 See Fig.5         10.15u
5&0             4096              16384               See Fig.6 See Fig.7 See Fig.8         10.1u
1&0            24000             24000               See Fig.9 See Fig.10 See Fig.11     6.69u
5&0            24000             24000               See Fig.12 See Fig.13 See Fig.14   6.6u
1&0            50000             50000               See Fig.15 See Fig.16 See Fig.17   3.3u
5&0            50000             50000               See Fig.18 See Fig.19 See Fig.20   3.3u
As the frequency goes higher, the coherence and magnitude dropped and the phase increased. 16384
Hz is still a special frequency to the AD converter.

The figures are in the attached files. Thanks a lot.

[Doug_L._Bear]

I am currently exploring two options with this.  The first possibility is that 16384 is a harmonic frequemcy of the circuitry and is causing this lack of coherence.  The other is that DAQmx sampling rates need to be a divide down of the 80 MHz sample clock, and when odd frequencies are given they are coerced to a divide down frequency, which may be reflected in the coherence calculations.  If you could run  your tests and let me know if you see this with other "odd" frequencies, not just round numbers.

[rico1985]

Doug,

Thank you for your reply. I ran my tests with other frequencies from hundreds Hz to 50kHz and didn't see lack of coherence. 16384Hz is the only odd frequency in my tests. I am not familiar with the harmonic frequency of the circuitry. Would you please explain more about it to me?

Thanks a lot.

[Doug_L._Bear]

A resonant frequency is a characteristic of any circuit defined by the physical characteristics of its components.  It basically means that at that frequency a very small input frequency will become a very large output frequency.

http://en.wikipedia.org/wiki/LC_circuit#Resonance_effect