Will the NI 9232 perform with an IEPE shock accelerometer?

[GLopp]

I've been tasked to design a test to measure vibrations from shock excitation and and I'd just like to get some confirmation of the NI 9232 module working with this PCB shock accelerometer with specs at the following link: http://www.pcb.com/products.aspx?m=350C03.

I'm familiar with using the NI 9234 for use with IEPE accelerometers, however, this module has a sampling rate of 51.2 kHz per channel whereas I'm looking for 100 kHz per channel. I saw the NI 9232 has a 102.4 kHz per channel sampling rate which is a better fit for my application but also has a +-30V input range whereas the accel I'm looking at getting only reaches +-5 V, so it looks like I'll have to sacrifice in resolution by using this module. Additionally, I see that the 9232 has either 0 or 4 mA IEPE excitation, whereas the 9234 has either 0 or 2 mA, The specs I linked for the accelerometer show a constant current excitation of 2-20 mA, so because the 4 mA falls in this range, this means the 9232 module will work with this accelerometer correct?

I apologize if this question seems a bit basic, I'm pretty new when it comes to designing a data acquisition system on my own whereas my previous experience in testing has come from utilizing previously implemented data acquisition set-ups.

[JustJohn]

Hello GLopp!

The card that you are looking at does seem to meet the specifications that you require, however with this 9232 device you will only be using 4-bits of the ADC resolution if you leave the accelerometer as is. The 9232 doesn't look like it has multiple modes for max voltage meaning you would be using the full 24-bit ADC designed for +/- 30V for +/- 5V meaning you only use 1/6th of the bits available to you. This could be avoided if you amplify the signal coming from the accelerometer by 6. That way you use the full range of the ADC. Alternatively you could look into some of the other modules that have a +/-10V range and allow for sampling around 100kS/s/ch. How accurate does the measured accelerometer data need to be?

[GLopp]

Thanks for the response John,

I'm a bit confused as to what you mean by only using 4-bits of the 24-bit resolution of the ADC. I'm aware that 4-bits is 1/6 of 24-bits but I thought the effective resolution was calculated differently. I was using the equations found at this site (https://www.maximintegrated.com/en/app-notes/index.mvp/id/5353) where they say the effective resolution is calculated using the equation  Resolution = log(N)/log(2) where N for this case would be N = (1/6)*(2^24) = 2.8e6. This would give an effective resolution of 21.42 bits. I did look at something along the lines of the NI 9215 or the higher channel NI 9220 which have the +/-10V input range with 16-bit resolution, but the way I was calculating the resolution for my accelerometer signal, the NI 9232 would still perform better. Additionally, since these modules don't have internal signal conditioning, I'd be required to get an external signal conditioner correct to provide the IEPE excitation for the accel correct?

[JustJohn]

Hello GLopp,

You are correct about the effective resolution. The 4 bits comment was me improperly associating the range of your signal with the range of the module. Using this 24 bit ADC, your signal can be represented by 2^24 (16777216) individual numbers between +30 and -30. This gives you a code width of 60V/16777216 = 3.57x10^-6 Volts. Just because you are not using the full range of the ADC, does not mean that the code width will change. Your signal will be represented in increment values of 3.57x10^-6 Volts even when the signal is +/- 5V. This is more than enough resolution for the +/-5V signal from the accelerometer. The only thing that would affect this code width would be the bits of the ADC or the range of the Vin, but since both of those things are static with the 9232, it wouldn't change. Your resolution calculation is correct and would give you better performance than the other mentioned modules. 

It sounds like 9232 module is the one that you would need for the application you are describing assuming this 3.57x10^-6 Volt code width is acceptable.