saturation-like problem

[Paulo_br]

Hello Folks,

I am having trouble when acquiring acoustic emission signal using a National Instrument Board, model PCI 6111 (12 bits resolution, 2 channels of 5 MS/s rate each).  I am using a specific VI in LabVIEW to acquire the signal and then a simple code in Matlab to read and process the data. The data format recorded is int16. Figure 1 shows the raw acoustic emission signal acquired at 2.5MS/s. This is a pencil test, which consists of breaking a pencil lead on the sensor and record the signal (Inpulse-like input). Apparently it is OK but when you zoon it in, for instance, the region marked in Figure 1, you get the Figure 2. Observing Figure 2, you can see some flat regions in the signal but not quite observable yet. Zooming in other region, now in Figure 2, we get the waveform shown in Figure 3, which we can clearly see the flat regions in the waveform. I thought it could be a kind of saturation but I am not sure. I tried to decrease the amplifier gain of the sensor circuitry unit but even though it did not work. Even if when I zoon in the noise region in the waveform I can clearly see the flat regions as well. That is upsetting me because I know my signal is not good to be analyzed yet. I have checked the DAQ board functioning with a function generator by inputting a known waveform. The data was very nice acquired. I would appreciate any suggestion and advice you may have.

[Laurent-04]

Hi,

I don't think that your problem is due to saturation since you have samples higher than the "bad" samples. You have two consecutive measurements at the same value, but your signal can change between these values. Check your useful pass band regarding your sample frequency. If the missing values between two identical consecutive measurements are significant, increase your sample measurement.

Of course, if you have to filter the signal with a low pass, you don’t need to do anything in my opinion.

[chrisger]

hi there,

your data has I16 - format (+-32767), but in the screenshots only a range of about +-200 is used. if the data shown in the pictures is typical for your application i'd recommend to INcrease the amplifier gain of your mic or to adjust the voltage range of your DAQ-card to the signal.

the flat regions in the data are just consecutive points with the same value (use a display mode which shows the datapoints without a line between them) because of the poor relative resolution due to the weak signal. there's nothing special with that.

[Paulo_br]

Thank you very much Laurent, you helped me. The problem was not regarded to saturation indeed, it was about scale range I was using instead. Cheers

[Paulo_br]

Thank you so much Chrisger. You are right about the range I was using. I increased the gain of the sensor amplifier and got much better results. I should have seen this before. Cheers