What is the real voltage? (pci-6229)

[tannerd]

I am using an M series PCI-6229 to read some thermocouples and voltages through a CB-68LP Connector Block.

I have created a DAQmx task to read these temperatures and voltages.

My issue is that when I include multiple channels in a task one of my voltage readings changes.

When I just read the channel in question (physical channel Dev1/ai15 terminally connected RSE) I read 2.01V (which is what I get when I use a voltmeter to measure the voltage).

When I add other channels to the task (both RSE and Differential) the reading drops to 1.95V.

When I add some thermocouple channels to the task, the reading drops to 1.72V.

I have another voltage (physical channel Dev1/ai14 terminally connected RSE) that reads correctly at 5.16V regardless of the number of channels I include in the task.

The 5.16 volts is from a power supply, and the voltage in question is from a temperature sensor with signal conditioning (analog devices AD22100).

Is this a sensor issue? and if so, why?  Or should I change the way I read this voltage?

Thanks fo your help,

David

[devchander]

May be a grounding related issue?? Have you tried connection all channels in differential mode and still experience the same readings??

Also are you usnig DAQ assistant?? or are you building an acquisition vi?? ( assuming that you are using LabVIEW)

do attach the code that you are using and we could suggest better

[tannerd]

Hi devchander,

Thanks for the reply.

I tried connecting all channels in differential mode but am still experiencing the same issue. (although the problem has been reduced slightly - instead of a difference of ~0.3 V between the actual and what is read, now it is only a difference of ~0.2 V)

I am using NI-DAQmx to create tasks that will be used in LabVIEW.  I don't believe this to be a LabVIEW issue since the problem can be seen using the DAQ assistant test panel.  So, I don't have any code to show, just the settings in the NI-DAQmx tasks.

Again, thanks for the help,

David

[tannerd]

I've solved the problem.  This was a sensor wiring error.  As can be seen on the sensor datasheet:

http://www.analog.com/UploadedFiles/Data_Sheets/AD22100.pdf

The temperature sensor voltage output should be wired through a RC filter.  But why?

From reading through the sensor datasheet I learned:

"A single resistor and capacitor are recommended to provide immunity to the high speed charge dump glitches seen at many microprocessor ADC inputs."

What is a "high speed charge dump glitch" and why is this an issue with my NI multifunction DAQ board?

If anyone knows, I'd love to learn.

Thanks,

David

[Sandia_Mike]

Surprising that the addition of an RC filter fixed everything.  One problem I've had with the PXI-6229 is the relatively low source impedance -- only 2k ohms.  Was expecting something more in line with a voltmeter, like greater than 10k ohms.  So I've had to add buffers to high-impedance sources to compensate.  Adding a capacitor could make the signals you are measuring look like a lower impendance (briefly anyway), and so might compensate for the low measurement impedance.

Don't know what the "glitch" problem was from the application note, but maybe charge output from the microprocessor's multiplexer as it switches among channels? 

[Neil_S.]

David,

Thanks for posting to the NI forums.  I am also surprised that adding a low pass RC filter fixed what seemed to be a ghosting problem.  Our multifunction DAQ boards such as the NI PCI-6229 have a single ADC with a front end multiplexer to provide a large channel count without dramatically increasing cost.  However, any multiplexed system is susceptible to ghosting.  Ghosting is caused by a small parasitic capacitance that exists on the ADC input that can allow a charge build-up.  For example, if the first channel of an acquisition is at 5 V it charges the parasitic capacitance to 5V.  When the multiplexer switches to the next channel (say at 3 V) the charge on the capacitor must be discharged to bring its voltage to 3 V.  If the second channel has a high input impedance the time it takes for the capacitor to discharge may become significant and the reading may be higher than it should be.    Adding a low pass filter will actually increase the input impedance and capacitance of the input channel and will contribute to the problem.  However, it may be that the behavior you noticed was caused by some unknown factor such as a grounding issue that was fixed when you added the filter.  Since the data sheet for the IC you are using explicitly suggests using the low pass filter it is probably a good idea in this case.  By the way, the input impedance of the 6229 is like 10G Ohms not 2k Ohms.

As far as the glitches are concerned, I don't believe the NI boards suffer from the glitching mentioned in the data sheet.  There are many types of ADCs, some of which, are more susceptible to glitching.  Our M-Series boards such as the NI PCI-6229 use a successive approximation ADC which is less susceptible to glitches than a direct conversion ADC.  For more information take a look at the following knowledgebase that discusses different ADC technologies:

http://digital.ni.com/public.nsf/websearch/32FD9AA817D0EBE68625708C005E1B26?OpenDocument

I am glad that you were able to get your sensors working with our M-Series DAQ device.  Good luck with your application.  Let me know if you have any additional questions.

Regards,

Neil S.
Applications Engineer
National Instruments

[Sandia_Mike]

Thanks for commenting on my misinterpretation of the 6229 input impedance.  Saw a reference in the DAQ M Series User Manual mentioning really high input impedance, but also a suggestion that the signal source impedance should be less than 1k ohm for good response in sampling. 

I disagree, however, on your comment that an input filter will degrade the subsequent sampling stage.  Assuming the sampling multiplexer “breaks-before-make”, having a capacitor on the input signal should only improve the sampling accuracy of a high impedance (but of course low frequency) source.  It should swamp whatever capacitance is in the sample & hold circuit of the 6229 or parasitic capacitance.  With a higher-frequency, but also high-impedance, source signal probably the only alternative is to insert a follower amplifier as the DAQ User Manual also mentions.

[Neil_S.]

Sandia_Mike,

Thanks for the clarification.  I see how the filter could improve the accuracy since it is on the source side of the multiplexer.  The capacitor of the filter gets charged with the input signal which swamps the input capacitance of the board.  It makes sense, I am not sure how I missed it.  Too much thinking, perhaps .  The suggestion in the User Manual to use a source impedance of less than 1 k Ohm, however, definitely has to do with ghosting.  Higher impedances increase the settling time to more than the max sampling rate of the board.  You can use higher impedances as long as you use longer settling times (which in turn means slower acqusitions).

I appreciate your correction.  It is why I enjoy the forums.  It keeps me honest and I learn all the time.

Regards,

Neil S.
Applications Engineer
National Instruments