Dynamic Signal Acquisition

cancel
Showing results for 
Search instead for 
Did you mean: 

Wandering DC level 9234 and microphone

Hi there,

 

We are using a NI 9234 module with accelerometers, impact hammers and IEPE microphones (GRAS).

 

I'm fairly new to using NI hardware interfaces, and have versions of data acquisition working in both Matlab (via session-based interface) and directly in labview/signal express.

 

I've noticed this strage phenomena of a wandering DC level for all measurement channels, especially when connected to the IEPE microphones. See the attached figures. This requires extra steps during data processing, which is pretty frustrating. Is this normal behaviour for these data acquisition devices?

 

There are 2 attachments, which show the behaviour during silence, and during speaking. In each case you can see the DC level wandering around all over the place, like there's a grounding problem in the device, or something similar.

 

Note that I have tried multiple different BNC cables, and I've run the tests in different physical locations (home, office, lab).

 

Many thanks,

Mike

Download All
0 Kudos
Message 1 of 10
(8,312 Views)

Hi Mike,

 

Just a couple of questions to start with,

 

1) What is the scale on the graph? I am assuming it is just volts.

2) Would you be able to let me know what microphone you are using?

3) Can you remove the microphone from the input and see what noise we get on the device with nothing connected? 

4) Do you have another device that you can measure the signal that we get from the microphone instead of the NI 9234?

 

Cheers,

 

Luke

0 Kudos
Message 2 of 10
(8,256 Views)

Hi Luke,

 

Many thanks for your message.

 

1) The scale is in Pascals (i.e. it's converted to Pa via the TEDS chip/interface interaction)

2) The Mic is a 46AE (we have numerous other models too from GRAS, all of which seem to behave like this). Our model produces about 54mV/Pa when powered over IEPE/ICP. It is about 4 months old.

3) Yes, I have done this using SignalExpress for speed. The first image attachment shows the result of using a 'Sound Pressure' input type with IEPE powering on, but the transducer (46AE) not plugged in. The second image shows an example with the same setup, but the microphone IS plugged in, in a very quiet office (i.e. not much going on acoustically). This result is typical of what we seem to see when using these transducers plugged straight into the 9234 (as they're supposed to be).

4) Yes, I have tried inputting the microphone via a PCB ICP power supply (battery powered), and taking the input as a straight voltage into SignalExpress. The third example image is the voltage signal I obtain when using a 'voltage' input with the Mic/PCB supply connected, but the supply switched off. The PCB power supply is unit gain (i.e. no analog gain applied). The fourth image is what happens with the supply switched on, with the room almost silent. The fifth image is the same as the fourth, but I clicked my fingers twice.

 

Finally, I've added a 6th image, which show the mic plugged directly into the 9234, switch on as usual, and with 2 finger clicks.

 

Many thanks,

Mike

Download All
0 Kudos
Message 3 of 10
(8,242 Views)

...And here are images 4 and 5.

 

Thanks,

Mike

Download All
0 Kudos
Message 4 of 10
(8,238 Views)

And once again... this time the result of using the onboard IEPE powering with our 46AE mic, recording silence, and with/without highpass filtering (5th order Butterworth, cutoff set to 20Hz).

 

Could this just be the sensitivity of the microphone down to those infrasonic regions? Perhaps it's louder down there than I realised in our building... 

 

Mike

0 Kudos
Message 5 of 10
(8,229 Views)

Hi Mike,

 

From the last image it definitely looks like you are getting some low frequency background noise or interference in your signal, would you be able to try measuring the signal in different environments other than your office and see if we still get the same noise. From what I can tell this is expected behaviour and just using a filter you are able to remove the unwanted signal. These signals are not audible anyway so it shouldn't have an impact on the audio you will be hearing.

 

Cheers,

 

Luke

0 Kudos
Message 6 of 10
(8,168 Views)

Wandering DC levels is most often the result of taking measurements before IEPE, sensor, and AC coupling filters have all settled. The 9234 has an AC coupling filter cutoff at 0.5 Hz. It can take a while (easily 10 seconds) fot that filter to settle after a change in DC level. Sensors also take a moment to stabilize when first powered. After enabling IEPE, your measurement procedure should allow for settling before measuring SPL.

 

The BNC connectors make nasty little antennae, so if you disconnect the sensor, put a terminator (such as 50 Ohm) on the end of the cable.

 

As for the low-frequency noise you observed (especially in Screens 4 and 7) - this noise is likely real (coupling in from the environment). Do you see any difference between channels? Run them all at the same time to even see if the same noise appears on multiple channels.

Doug
NI Sound and Vibration
0 Kudos
Message 7 of 10
(8,127 Views)

It may also be interesting to drop a Power Spectrum Block in your Signal Express Project if you continue to investigate the low-frequency noise.

Doug
NI Sound and Vibration
0 Kudos
Message 8 of 10
(8,123 Views)

Your Mic nearly goes down to DC ... well , nearly but according to the spec only a 2dB drop @ 4Hz! 

So migth be wind outside,   or any other low frequency pressure change ....you can't hear, but the mics will report 🙂

  if you have two mics put them close together and have a look at both ....

 

Greetings from Germany
Henrik

LV since v3.1

“ground” is a convenient fantasy

'˙˙˙˙uıɐƃɐ lɐıp puɐ °06 ǝuoɥd ɹnoʎ uɹnʇ ǝsɐǝld 'ʎɹɐuıƃɐɯı sı pǝlɐıp ǝʌɐɥ noʎ ɹǝqɯnu ǝɥʇ'


0 Kudos
Message 9 of 10
(8,100 Views)

Hi Doug and Henrik,

 

Many thanks for posting your comments, much appreciated. 

 

I do seem to have narrowed this down as likely caused by the environment. Channel-channel I see similar behaviour when measuring with a pair of 46AEs. I have repeated this in other environments, and see something broadly similar. I'll try taking the setup into a large, empty field, next, just to see... It seems that our physics building is full of low frequency rumble.

 

The only real issues this implies are:

 

1) That low frequency wobble might push the overall signal level into clipping in some cases (i.e. when measuring close to the upper limit of the mic's dynamic range)

2) Having to filter out the low frequency stuff adds a processing step, and messes up the phase if using a low order IIR (which may be a problem for some of our measurements where we're examining transients). I guess I'll look into designing some kind of compensating allpass section to try to push the phase back into line, unless anyone suggests a better method...

 

I guess this is what you get for using such wideband, high quality mics...!

 

Many thanks,

Mike

0 Kudos
Message 10 of 10
(8,086 Views)