PXI-6281 lowpass filter limitations

[xl600]

I purchased a PXI-6281 to act as an ADC for a 6 axis accelerometer I have.  When I try to configure it as follows

• Differential mode
• 3 or more channels
• 10kHz sampling rate
• 40kHz Lowpass filter enabled

I get an error from DAQmx that the settling time violates the period between two conversions (Error 200038).

According to the datasheet, the settling time for multi-channel acquisitions with lowpass enabled is 50us.  10kHz is 100us.  Why is this error happening then?

If I reduce the number of channels to 2 or less, it doesn't complain.

If this is a limit of the card, what spec would have clued me into this?  I need 6 channels with lowpass and nothing I read seems to have indicated the 6281 wasn't up to the task..

Thanks,

XL600

[SeaSick]

Hey XL600,

The settling time of 50us that is specified in the manual is per channel. The device is multiplexing through the channels, and each channel has a settling time of 50 us. So in order to read from multiple channels you will need to sample at a rate of 20k/n where n is the number of channels.

Regards

[xl600]

That makes sense.  So, here's my issue...

I need to sample at least 6 channels at 10kHz (For my DEI MotionPak accelerometer).  There's no doubt I will have to incorporate anti-aliasing filters as the 40kHz filter won't be adequate.  I don't have the option of spending any more $ on something fancier like a dynamic signal analyzer so I need to try to get the 6281 to work for my DEI MotionPak accelerometer.

I have an SCB-68A so my plan is to put the filters there using the built in RC filter pads.

With the lowpass filter disabled, the settling time is stated to be 8us per channel.

10kHz * 6 channels equates to about 20us of potential settling time per challel.  So it seems like this should be more than adequate.

But, the 6281 manual claims the source impedance should be 1kOhm or less.  My external filters will be either Fc = 1kHz or Fc = 2.5kHz.  The chart in the 6281 specs doesn't really match the table which claims 8us for no-filter settling time.  But if I read the chart correctly to hit +/-4ppm of step at 20us, my filter should't use a resistor greater than about 4k?  Am I interpreting that properly?

Thanks!

[SeaSick]

Hey XL600,

What chart in the specs are you looking at to match which table? Do you mind clarifying what specs you are specifically looking at.

Regards

[xl600]

371022K.pdf, Page 4-7 "Multi channel scanning considerations" under bullet 1 "Use low impedance sources) is where the 1kOhm reference is located.

371292g.pdf, page 2 "Typical performance graphs", "AI<0..31> Settling Error vs Time for Different Source Impedances"

[SeaSick]

Hi XL600,

Thank you for pointing me to the specs that you were referencing.

I haven't looked specifically into the math that you did to calculate the values in your previous post, but in general yes your thinking is correct. By adding the low pass filter you will be increasing the impedance and as a result increasing the settling time. 

An option that may allow you to decrease the impedance is to implement an active low pass filter. By doing this you will implement a voltage follower that will lower the impedance and provide you with a lower settling time than if you were to use a passive low pass filter.

Regards

[xl600]

After discussing this internally, we've desided to go towards the active filter approach.  Frequency Devices, Dataforth, and Precision Filters are all candidates with a number of applicable products.

Thanks for the help!

[SeaSick]

You're welcome! Good luck implementing the active low pass filter!