Why the coherence of two synchronized IEPE microphone signals is not close to one?

Hi, everyone,

I'm working on a sound source localization project. I use compactDAQ and 4 IEPE microphones from NI to set up a microphone array for the localization. When I read microphones data, they should be synchronized right? So I use mscohere function to calculate the coherence of two microphones. The distance between two microphones is like 2cm. However, the coherence I got is really bad not even close to 1. I will share how I record the microphone data code here. Are two microphone data recorded simultaneously? If so, what is the cause of bad coherence?

devices = daq.getDevices;
s = daq.createSession('ni');
addAnalogInputChannel(s, 'cDAQ1Mod1',0,'Microphone');
addAnalogInputChannel(s, 'cDAQ1Mod1',1,'Microphone');

s.Channels(1,1).Sensitivity = 0.04754;
s.Channels(1,2).Sensitivity = 0.04754;
s.Rate = 51200;
s.DurationInSeconds = 5;

[data,time] = s.startForeground();
plot(time,data);

[dpb]

"Are two microphone data recorded simultaneously?"

That's up to the DAQ as to whether it actually does SSH (simultaneous sample and hold) or is just sequentially scanning the channels in order.

The hardware documentation will answer the Q? of whether it is capable of SSH or not; that's first step.

[Lilin]

Thanks for your fast reply. The cDAQ I use is cDAQ9174 which supports the SSH. I'm just wondering maybe I didn't write the code correctly to sample the signal simultaneously.

[dpb]

It's been a long time since I used a NI data acq system so I don't know any of the current product by number but looks like the 9174 is the chassis, not the actual input module. Which actual input module are you using; the doc for the chassis indicates there are those with and those without SSH available.

[Lilin]

The input module is 9232. In its datasheet, the sampling mode is simultaneous.

[dpb]

OK, so now have indeed confirmed the hardware is; so something else would appear to be going on.

Do you have anything but noise as the actual response--what if you provide a fixed single tone to a single input; can you reproduce it in tone and amplitude?

Then try two.

When that works, then try the actual signal...