cRIO 9229 9239 - the accuracy given in the specifications does not mach the one of the calibration procedure

[bidouille404]

Hello,

I am using cRIO 9229 modules measuring +-10V signals.

The"NI 9229/9239 Calibration Procedure" indicates page 10 in Table 3 that the gain error is 0.04% and the offset error is 1.35mV for a temperature of 25°C+-5°C.

The "OPERATING INSTRUCTIONS AND SPECIFICATIONS NI 9229/9239" document gives a different information. The typical accurary given in the same temperature conditions (25°C+-5°C) is 0.03% for the gain and 0.84 mV for the offset (it is written 0.008% of a 10.54V range).

It seems that the accuracy given in the calibration procedure is not as good as the one given in the specifications. For instance for the gain accuracy, I have an error of 0.04% in the calibration procedure that must be compared to the lower 0.03% given in the module specifications.

Can someone explain me the difference?

Thanks.

[NIF_Vincent.O]

Hi Bidouille404

Welcome to National Instruments' forums

The gain error for this device is 0.04%.

There has been several releases, that is why you can find different values for the same parameter.

Regards,

[bidouille404]

Thank you for your answer.

I still do not understand where this 0,04% of gain error comes from.

Have a look at the specifications given by NI for the cRIO 9229 module on this page :

http://sine.ni.com/ds/app/doc/p/id/ds-199/lang/fr

The given value of gain error is 0,03% in section "Accuracy, NI 9229" / "Calibrated typ (25 °C, ±5 °C)".

Can you explain me this difference 0,03% versus 0,04%? Where can I find this 0,04% specification?

Thank you.

[dousley]

The calibration procedure does contain a table at the end called "Accuracy under calibration conditions." That table gives several constraints on the environment that are specific to (and only reasonable to maintain in) a calibration scenario. That table represents the maximum error that a module should have during a calibration verification.

The specifications for the module (in the user manual) also provide a maximum operating specification as well as a typical operating specification. The typical specification is only given as a description of normal performance for a module, but does not reflect worst case/ maximum error, where as the maximum spec shows the worst case performance. In the 9229 manual, the typical and max specs are also given for different operating ambient temperatures.

So you have uncovered three different specifications for the module,

1) Maximum operating specs

2) Typical operating specs

3) Maximum calibration specs

They are all different because they are communicating different aspects of performance. It is not surprising that the maximum calibration specs may (in some cases) be wider than the typical operating specs. However, the maximum calibration specs will never be wider than the maximum operating specs.

Does that help answer your question?

DOusley

R&D Hardware, cSeries Modules

[bidouille404]

dousley thank you for your answer. I understand now the difference between "typical" and "maximum" accuracy.

Actually, I thought that "typical" and "maximum" was a direct reference to the temperature environment (typical = 23°C+-5°C ; maximum=-40°C-+70°C) but this was a mistake.

It seems that you know quite well these equipments. I would have then a few other questions.

1) Where can I find detailed information about cRIO module accuracy? For instance, the documentation of the cRIO 9229/9239 gives only the following information:

- typical accuracy for a temperature of 23°C+-5°C

- maximum accuracy for a temperature of -40°C/+70°C.

I would be interest in detailed accuracy such as the maximum accuracy for a temperature of 23°C+-5°C. If I understand well, this is what is given in the calibration procedure (0.04%) but is there another document that gives clearly this information rather than finding it in a calibration procedure.

Generally speaking, the accuracy information given for the cRIO modules (9214, 9217 etc.) are much less detailed that the ones of PCI/PXI modules (NI 6225). I am sure that this information is available somewhere.

2) How should I understand, the "accuracy stability section", for instance the grain drift (5ppm/°C). Is it a "typical" gain drift or a "maximum" gain drift? Generaly speaking, I do not understand how we obtain the maximum gain error on the -40°C/+70°C (0.13% given in the specs) range using the maximum accuracy at 23°C+-5°C (0.04% given in the calibration procedure) and the gain drift 5ppm/°C (given in the specs). The numbers are not coherent.

Actually, my goal is to determine a typical as well as a maximum error within a temperature range that is 3°C/+43°C. I can not sort out this problem right now!

Thank you.

[dousley]

Hello bidouille404,

1) The specifications in the calibration procedure are only valid under very significant restrictions (as indicated in the notes above Table 12 on page 12). A) The ambient temperature must be restricted to the given range, B) The module must be in a particular slot of a particular chassis, C) All the other slots in the chassis must be empty. Those specifications are therefore not generally useful, but are only specifically useful when calibrating the module. For that reason they are not found in the user manual.

In the user manual we provide specifications that apply to the broad range of configurations possible within the platform. These are the specifications that you should apply to calculating accuracy in a given deployment.

2) The stability specifications are provided as typical drift performance. As the note at the beginning of the specs points out, the specifications are typical unless otherwise noted. However these drift numbers can give a good idea of what kind of performance to expect typically. To find the typical accuracy, you can take the typical specification at the limited 23+/-5C and add additional drift according to the drift specifications.

However, for calculating the max specs, it's necessary to resort to the full temperature max specs provided in the accuracy table since we do not provide any additional specifications to give a more granular breakdown of the maximum errors at all possible operating conditions in the platform.

[bidouille404]

Ok thank you for your answer.

This is a pitty that there is not a detailled information regarding the accuracy of cRIO modules. I am also working with NI PCI-6225 cards and the "accuracy section" of the specs is really great, probably the best I have found so far for daq material from many manufacturers.

I would have a last question. I am working on calculating the uncertainty of different measurement channels using cRIO modules. I follow the "Guide of Uncertainty Measurement" as reference.

The uncertainty expressed following this "Guide of Uncertainty Measurement" must be quantified as multiples of "standard deviations" or "sigma", corresponding each to a given level of confidence. An uncertainty can be expressed as "1 sigma" (66% level of confidence), 2 sigmas (95% of level of confidence) etc. 

My question is the following: what it the level of confidence associated with the word "typical". Is it a "1 sigma" level of confidence, "2 sigmas"? Is there any precise (quantified) concept associated with the word "typical"? What really means "typical" for NI?

Thank you.

[dousley]

Unless explicitly stated otherwise, I would not assume that "typical" has a specific confidence interval associated with it. Typical specifications are not warranted like the maximum specifications. Generally typical specifications represent expected behavior, but might not be guaranteed by design or through production tests.