can we adjust gains of PXI4472B?

[benbol]

Hi,

I have bought the NI PXI4472B. I have 2 identical sensors which have a frequency range of 0.03-50Hz. With these sensors, I measure the velocity of the ground. Unfortunately, I have a good coherence only from 1 to 50Hz.
The problem is that the resolution of my ADC is not good enough : even if I have 24 bits of resolution, the full scale is -10V/+10V, and the signals of my sensors are in the maximum range of -50mV/50mV. So I would like to know if it is possible to adjust ADC gains to have a full scale of the maximum amplitude of my signal for example.

Please help me,

ben

[SanaaT]

---

benbol a écrit:
Hi,

I have bought the NI PXI4472B. I have 2 identical sensors which have a frequency range of 0.03-50Hz. With these sensors, I measure the velocity of the ground. Unfortunately, I have a good coherence only from 1 to 50Hz.
The problem is that the resolution of my ADC is not good enough : even if I have 24 bits of resolution, the full scale is -10V/+10V, and the signals of my sensors are in the maximum range of -50mV/50mV. So I would like to know if it is possible to adjust ADC gains to have a full scale of the maximum amplitude of my signal for example.

Please help me,

ben

---

Hello Benbol,

The link below explain in detail how to adjust gain if DSA device like PXI-4472 :
http://digital.ni.com/public.nsf/websearch/5DBCDB11906354018625676F0083CC32?OpenDocument

Hope this help,

Regards,

Sanaa T
National Instruments France

[Jack_A.]

Ben -

The PXI-4472B has a single fixed input range (+/-10V). You could externally amplify your signal before it gets to the 4472B if you need more resolution.

-Jack

[benbol]

Ok, thank you very much for your answer.

ben

[schi]

... that is the fine thing about 24 bit ADC: You typically do not get a higher sensitivity, you just get rid of choosing the input range 😞

The limiting factor often is the analog design: cross-talk, noise, etc. This really makes a high quality (expensive) measurement system. Today, you get 24 bit converters, giving you a theoretical dynamic of nearly 150 dB. But you will not get that in practice, because you cannot do that easily in analog. And you cannot do it in anti-aliasing, too. So, what you typically get at the specs is something about 120 dB, for real world conditions it is often worse.

If you go down 120 dB, starting at 10 V, you end up at 10 muV. If you have an old, traditional, simple low dynamics 16 bit system giving a 10 mV input range (and many did so) and 80 dB dynamic range (even at 10 mV!), you end up with 1 muV, i.e. a 20 dB higher sensitivity for low-level signals. And you can be sure that there is visable aliasing coming up in your measurement data, etc...

These traditional systems have only these drawbacks:

1. they are more expensive to build as they need high quality internal amplifiers
2. as a user, you have to care about signal dynamics
3. 24 bit sounds more modern, more sexy on the sales side.

So, as in many likewise situations: Remember the old instruments and have a close look on performance degradations of "new, simple" solutions (likewise reasoning applies for high precision, low signal measurements comparing charge type accelerometers with ICP / IEPE type ones)!

just my 2 cent,

schi

[Jack_A.]

There are also devices which have both a 24-bit ADC and gain which provides even higher sensitivity.  An example of this is the PXI-4462.

-Jack

[schi]

thanks, Jack, for pointing to that device. Yes, you are true, there are fine devices which deliver good performance. Nevertheless, the device fits exactly in what I said:

• Channel price about 1000 €, i.e. twice as much as the popular 4472
• Signal dynamics limited to less than 120 dB
• Signal dynamics limited by the analog side for high sensitivities
• Anti aliasing is a critical issue.

So, the device is really fine (according to the spec, I did not test one by now), but confirms my earlier remark.

schi

[benbol]

Thanks Jack and Schi for your answer. Maybe the DAQ PXI-4462 will be an answer to our question because we want to measure displacement of the ground motion which is a nanometer scale (so the full scale is approximatively +-50mV) And do you know what are the characteristics of DAQ specialized for low frequencies measurements (0.1 Hz until 200Hz)?Maybe you know some DAQ which will be fit for our application?

thanks a lot,

ben

Message Edited by benbol on 07-19-2005 09:42 AM

Message Edited by benbol on 07-19-2005 09:42 AM

[schi]

Hello ben,

that is really a good question. As I had a short look on the 4462 spec, I remember it had a 3 Hz highpass filter, so you will not be able to measure there (with AC coupling i.e. highpass filter switched on).

First question: Do you need IEPE (aka ICP) supply? If so, it will get difficult, as you will always have to use AC coupling some way. If not, you might try DC coupling and perhaps use a separate high pass filter (analogue, -3 dB well below 0,1 Hz).

In addition, you might have a close look to your sensors and consider additional signal conditioning (high pass filter, low pass filter, high quality amplifier) to make the overall measurement work. In addition, as you might also be interested in phase as well, mind the phase response of all the filters / signal conditioning. You might get additional issues there. 

In general, take care of the time constants of the filters and that is especially true for setteling of the sensor responses when using IEPE powered sensors. They might easily come up to be a few 10 seconds! Take care of signal switching, supply switching, etc.

I know of some special measurement systems for vibration measurements, but I do actually not know for sure which of the standard DAQ cards will give the most simple and cheap solution for your overall measurement problem (might also depend on channel count, software considerations, etc.).

Although it is more questions than answers, it will hopefully help

schi

[benbol]

Thanks for your answer. I am  working with accelerometers, so I need IEPE supply, and so I have to use AC coupling, if not my DAQ input is saturated. I don't understand the reason, because the output impedance of my sensors are less than 10 ohm, do you know why? Do you know if I can find a DAQ, when used on AC coupling, which have a high pass filter whith a cutoff frequency less than 0.1Hz? The other solution is to use an external supply for my accelerometers, so I will be able to use DC coupling, do you think it's possible?

 I am also working with sensors which measure velocity and which have an external supply. So we don't need IEPE supply in this case and I use DC coupling. I don't understand in this case why the DAQ input is not saturated, do you know why? Can't I use a digital high pass filter (for instance in Labview) instead of an analog high pass filter? Because I'm afraid of the noise I can add.

For the time constants of the filters, you are true : it's approximatively 10 seconds when using IEPE powered sensors with the PXI4472B. So I wait before recording my data. Is it because of high pass filter (with a cutoff frequency of 0.5Hz)? Or because of the antialiasing filter? 

In general, do you think the PXI4472B is suited for low frequencies mesurements (0.1Hz to 100Hz for ground displacement measurement with nanometer scale) if I use DC coupling? I don't know if +-10V  full scale with 24 bits of resolution is enough.

We want to have a DAQ with several channels simultaneously sampled. Eight channels will be perfect, if not we can synchronize multiple devices. It will be also good to have a sampling rate which will be variable from around 50Hz to 1000Hz or more, which is not the case of the PXI4472B.We would like to have a very good resolution, because we measure very small velocity or acceleration(50mV is the maximum amplitude) For the moment, we are working with Labview (DAQmx and Sound and Vibration Toolkit) with Windows 2000 professionnel, but I am open minded to every better solution.

Thanks a lot for your answer,

ben