# Microphones disturbed by capacitance

I've used several KE 4-211-2 Sennheiser microphones to assemble a device for acoustic signal analysis. They are connected to a National Instruments data acquisition device with ~1 m long cable and sampling rate is set to 50kHz.

I've added a resistor and a capacitor as described here (2nd page). However after a start of data collection instead of clear acoustic signal I get strong disturbance looking like a step function in low pass filter. It obviously distorts my frequency spectrum.

After investigation I found out that when NI device is not collecting data, capacitors charge slowly and start of collection probably results in dramatic drop of impedance on inputs. What might be a reason for this behavior and what should I do to make it work properly?

I think there is everything on this picture

• Please add a picture of your setup that clearly shows how things are connected. A circuit diagram is welcome too (hit Ctrl-M). – jippie Apr 27 '13 at 6:21
• It would be very helpful if you could give more information about the data collector, maybe a link to the datasheet. – clabacchio Apr 27 '13 at 8:27
• Are you switching the operating voltage (Vo in your schematic) when you start acquisition, or is it continuous? – Brian Drummond Apr 27 '13 at 9:32

Perhaps your circuit does not have a DC return path for the bias current that needs to into/out of the data collector's input:

simulate this circuit – Schematic created using CircuitLab

One side or the other has the responsibility of implementing the DC return. Without this, the current flowing from the input (which is arbitrarily represented here as an op-amp's + input), no matter how small (could be nanoamperes) will gradually charge the capacitor until the DC level at the + input is out of range.

Another way to look at it is that the input naturally has some arbitrary voltage bias, and by connecting a capacitor, the circuit will eventually reach that bias point, which is not in range. The added resistor conveys the 0V voltage ground reference to the input, forcing it to be biased at 0. Because the impedance of the input is high, that resistor can also be high, so as not to degrade the input impedance to an unacceptable level, yet it appears "stiff" enough to the input that it conveys the 0V impedance without an offset. If a microampere of bias current flows through 100K, it's not a big voltage, in other words.

In your circuit, since you already have a 5K2 impedance and this is your wiring outside of the microphone, what you can do is double the resistor value and put two of them in parallel, but one on the other side of the capacitor.

simulate this circuit

11K is a resistor value available in the E24 series.

I've used multi-channel NI analogue acquisition cards and they have a single ADC which can be programmed to connect to one of the mulitple-inputs by a multiplexer chip. Channels not un-connected to the ADC will float high and, when that channel "connects" it will glitch.

You need a pull-down resistor on the inputs so that if there is leakage current from the mulitplexer, it has a path to ground. Probably something in the order of 3k3. It's easy to try and hopefully this advise works.

Data acquisition boards often do not have high input impedances when they are powered down. Your board needs to be powered up. A preamp between the mic and tha A/D might help, depending on the preamp.