uncertainty of 9211 thermocouple module

[oilyfingers]

On pg 16 onwards of the 9211 module manual (http://www.ni.com/pdf/manuals/371566e.pdf) there are a number of graphs showing the error introduced by the DAQ system.

My question is, if there 'error' on the graph is say 1degC, does this mean the uncertainty of the measurement is +/- 1degC, or +/- 0.5degC [obviously, +0.5degC - (-0.5degC) = 1degC].

I presume this is for a 95% confidence interval.

Dave

[gt3000]

Hello Dave,

I have looked at the USB 9211 manual so that I can try and answer your question.  The graphs you are seeing from page 16 onwards have provided modeled results, using an intricate equation taking into account the specific factors of the device, as well as the typical error range and accuracy of each type of Thermocouple that can be purchased and used. 

From my research, it is my understanding that, using your example question, that 1 degree C of error implies that you will be +/-1 degree C from the temperature you are trying to acquire. 

In your application, how will this information be used?  Are you currently already using thermocouples and a 9211?  Do you require a different specification?

Regards,

[oilyfingers]

Thanks - I think that answers my question. Really I was just confused by the graph as to whether the 'error' quoted on the y axis is the 'peak to peak' error (if you like) or HALF the peak to peak error, in other words if it says 1degC 'error', is that plus 0.5, minus 0.5degC, or plus 1degC minus 1degC presumably with some confidence interval eg 95%. I think the clarity of the y axis label could be improved.

Dave

[gt3000]

Hello Dave,

I am glad to see that your questions have been answered.  I have found a resource on the National Instruments website that demonstrates the permitted error bands from the IEC Standards organization.  As this is in text, and not a graph, it may clarify the possible error just from the nature of each thermocouple. 

Knowledgebase article: http://digital.ni.com/public.nsf/allkb/776AB03E065228408625727B00034E20?OpenDocument 

A previous forum post for a different user describes the type of testing that is included in producing the graphs in the 9211 documentation.  One user posts a method for also finding the possible error under the environment conditions for you application, by plotting several test measurements.

Forum post: https://forums.ni.com/t5/Multifunction-DAQ/NI-9211-Measurement-Accuracy/m-p/752976?requireLogin=Fals...

Regards,

[oilyfingers]

Thanks George, but the first link you sent me is a link to thermocouple uncertainties, not uncertainties in the DAQ system - they are two separate things. 

If I could get in contact with whoever wrote the 9211 operating manual that would be good! Can you find out?

Dave

[gt3000]

Hello Dave,

You are correct in your observation.  I was not supplying the table in place of a description of the NI 9211 manual data.  I wanted to demonstrate that measurement error and accuracy of your readings is dependent on the physical nature of the thermocouples you will be using.  The plots in the manual however, provide an error band based on the technology employed by National Instruments, such as the gain, noise filters, and the 24-bit resolution of the Analog to Digital Converter.

It seems that you are heavily interested in the accuracy of the temperature measurements.  In an earlier post I asked about your application.  This is because you can begin using other temperature transducers for a higher accuracy, such as an RTD, and acquire data with an NI 9217, seen below.  RTD's also have greater linearity, which, as you saw from the 9211 manual, is not the case for thermocouples.

https://www.ni.com/en-us/shop/model/ni-9217.html (You can find the manual from the "Documentation" heading).

What further information would you like from the operating manuals?

Regards,

[oilyfingers]

George - I am doing PhD research and I need to account for the uncertainties in my measurements. I know that RTDs are more accurate and I'm actually using an RTD that we've developed (with our own instrumentation) as well, but I'm also using some thermocouples for other parts of an experiment and I need to quantify the uncertainty explicitly. So it's just a simple question about what the error bands in the graphs in the manual actually show! I'm happy with my choice of thermocouples at the moment, thanks...

🙂

Dave

[gt3000]

Hi Dave,

As mentioned earlier, the graphs plot the error as, for example, +/- 1 degree C, from the temperature you are trying to read.  The IEC link from earlier is the best source of internaitonally acceptable bands of error for thermocouples, that the NI hardware would be subject to, and would be a great reference.  However, to produce similar graph models, you can use the available circuit diagram on page 14 to produce a block diagram noise model.  This is a procedure covered in fundamental Noise and Signal Integrity modules at university.  Once you have the model, the filter and impedence characteristics of the components uses is on page 20.

This will allow you traceable results for your own graphs, which will be appropriate for your PhD Research.  You can then cite the manual as well.

The 9211 manual has been available since the product's release several years ago.  The groups working on it could have been dispersed and working on new projects entirely in this time, and cannot be contacted. 

Best Regards,