Running into issue with what I believe is the Open Thermocouple Detection on a 9211 using the Scan Engine

[JaRay200]

Due to the scale of this project, I am using a multiplexer to sample up to 200 T-Type thermocouples instead of having individual channels on modules to support up to that number. 

The multiplexer is non-addressable and operates only on the pulses from a clocking signal.  When the enable line is high, everytime it gets a pulse from the clock it will step to the next input.  Between relay steps the minimum time for a valid reading is 5 msec accounting for any bounce and the operate time of the relay.  I plan to double this amount of time just to make sure everything is ready.

The issue I am runnining into is that I am currently using a 9211 to get the initial code written to step through the multiplexer and building my array for each set of samples.  I have a 9212 on order that has a faster sampling rate, but I'd like to get this 9211 working, even if it means sacrificing my sample time for now.  

What I have noticed is that the first few samples are either at the rail or close to it.  I've verified this by reading in raw voltage instead of temperature along with having a power supply connected.   I have the supply setup for 40 mV.  The longer I setup my delay before I begin reading the less railed values I get.  If I setup my delay for 1 second then all of my readings are valid however that delay is just too long.  I am only using the first channel of the multiplexer for now.  I have logic in place to step through the channels without issue, but it is not included here.  

The Scan Engine is set for 1 mSec and the dt of my timed loop is set for 100 scans.  Once again, this is slower than I plan for the final product, but I want to get it working here first.  

[yea_likethecity]

JaRay,

Could you post a screenshot of your results? I think I understand what you're doing but I don't think I quite understand the behavior you are seeing. It sounds like the first few samples are showing railed voltage if the delay isn't long enough. If that is the cause, could you verify the voltage externally somehow to see if it's an issue with sensors or your equipment? I was thinking maybe check it with a voltmeter in addition to reading from the scan engine.

[JaRay200]

Sure, it might be tomorrow morning before I get to it, but I can definitely do that.  Is there a known time that it takes for the OTD to update?

[johnsold]

The 9211 uses a multiplexer to switch the input channels (and CJC and autozero) to a single A/D converter. The specifications do not indicate the type of multiplexer but many NI devices use switched capacitor multiplexers. The time to clear overcharge from the multiplexer capacitor may depend on the source impedance.

Are you using all the channels or does your external multiplexer connect only to one channel? As a test, try connecting to only one channel. Turn off CJC and autozero. Then run your multiplexer and see if you get different results. This should tell you whether the problem is in the external multiplexer or the internal one.

Also note that the conversion time for the 9211 is specified as 70 ms per channel.  So you will need that much time plus any settling time for your external mutiplexer. Even though you may be able to run the scan engine faster than that, you will not be getting meaningful data.

Lynn

[JaRay200]

I am aware of the multiplexed input and am currently only using one channel.  In the VI I posted, the initial delays account for the multiplexing switching.  The first 50 msec delay is the time required from turning on the enable line to sending the first clocking pulse.  The specified time in the multiplexer manual states that ~4 msec is needed before taking a reading, I am doubling that to be on the safe side.  From there I enter a loop that checks the input of the single 9211 channel for a "valid reading".  This is being done crudely for now but just checks for any value < .07 V where .08 V is the rail.  When I get a valid reading it then starts to acquire data.  What I am experiencing is that it is taking around 800 msec from the time my first loop starts to poll the channel to when a valid reading has been received.  It seems much longer than it should be.   

I'll try the suggestions you listed tomorrow.