Aperture time of PCI 6251

[him727]

Could you please tell me the aperture time of PCI6251? Thank you

[amezam]

Hello him727,

You can find the settling time for multichannel measurements on page 2 for the PCI-6251. It is the first table on that page.

Best Regards,

Alina M

[him727]

Thank you for your reply!

Actually I am a beginner and I am curious about the time that the signal is integrated. Is it called aperture time or settling time? As I am using 1MHz sampling rate, I expect the time should be several nanoseconds. Thank you!

[amezam]

Hello him727,

The setting time is the time it takes the signal to reach a certain value and remain within a specified error. You will notice that the settling time is given in a +/- ppm range of step in the table. Check the following links on information on how the convert clock (which specifies the delay between channels in a multi-channel acquisition) and the settling time relate. The higher the convert clock, the longer the settling time, hence a more accurate reading. By default, there’s a 10 us pad for settling time, but depending on the sample rate, this will change as explained on the link below.

KB How is the Convert (Channel) Clock Rate Determined in NI-DAQmx and Traditional NI-DAQ?

Is Your Data Inaccurate Because of Instrumentation Amplifier Settling Time?

KB How does VI Logger determine interchannel delay?

KB Minimum and Maximum Values for the Interchannel Delay Setting

KB How is the Channel Clock Rate Determined in my Data Acquisition VI Using Traditional NI-DAQ?

Best Regards,

Alina M 

[him727]

Thank you very much. I have learnt a lot.

Actually I am still curious about the aperture time. I have read a book saying that

'A sample of a continuous signal cannot be acquired instantaneously. Instead the signal must be observed for a finite length of time called the aperture time. The signal is effectively smoothed or averaged over the aperture time'

Therefore, I would like to know about the time in which the signal is integrated and averaged. Thank you!

[him727]

http://www.ni.com/white-paper/6234/en

I think it is also called 'signal means'. Thank you

[johnsold]

That was "signal measure", not "means."

It appears that NI does not routinely specify the aperture time for their A/D converters.  I think the M series boards use SAR type A/D converters. Again, it is not clear whether they use a track and hold or sample and hold architecture. With track and hold the aperture time is close to the time between samples minus settling time.  For large steps the settling time is larger than the reciprocal of the maximum sampling rate.  With sample and hold the aperture time could be as small as about 1/16 of the time to convert a sample.  To make that work the internal amplifier would need to have sufficient drive to fully charge the sampling capacitor in that short time.  They use a custom designed programmable gain amplifier in these devices, so there is no way to know how fast it can drive the S/H capacitor.

It is also possible that they use the multiplexing circuits to serve as the S/H circuit.  If that is the case, the aperture time is about the inter-sample interval.

Lynn

[him727]

Thank you. I still have a problem on settling time. If I use only one channel, is it true that only the first data point requires settling time of 10us? There is no change in input voltage if I use only one channel so I think that the settling time is not required after the first data point. I am not sure... Thank you.

[johnsold]

There are two things which determine when a settling time needs to be considered.  Obviously one is channel switching. The other is the single channel case where the signal changes rapidly.  Settling time specifications are often based on a full scale change in input value without channel switching.

You can model the input as a capacitance to ground and a resistance in series.  Any change in the input signal will get to the actual A/D converter after a delay through the R-C circuit.  It does not matter whether the change is due to channel switching or the variation of the signal.

If the change in your signal is small from one sample to the next, or, equivalently, has no frequency components close to the Nyquist limit, then you do not need to worry about settling time.

Lynn