Accuracy / error / uncertainty: what 'typical' and 'maximum' means in terms of 'level of confidence' or 'sigma'?

[bidouille404]

Hello everybody.

Question 1)

In some of NI measurement equipment, the specifications regarding accuracy make use of the terms "typical" and "maximum".

For instance in the specifications of cRIO module NI9229, the "typical gain error" is 0.03% of the reading while the "maximum gain error" is 0.13% of the reading.

I could not find anywhere the exact definition of "typical" and "maximum" in terms of "level of confidence" or "number of sigma".

The expected answer is for instance something like:

- "typical" corresponds to a level of confidence of 95%" or 2-sigma,

- "maximum" corresponds to a level of confidence of 99.7%" or 3-sigma.

What are the levels of confidence associated with the terms "typical" and "maximum", when applied to accuracies in the NI specifications?

Question 2)

For other NI equipements, the specifications make use of only one type of "accuracy". They do not use the terms "typical" or "maximum". For instance, in the specifications of the PXI6225 card.

When nothing is precised, what it the level of confidence associated with the given accuracy?

Thanks for the answers!

[Audrey_P]

Hello bidouille404,

about your two questions:

Question 1) in the NI 9229 Operation Instructions and Specifications manuel the terms "typical" and "maximum" as on page 24 (Gain Error) refer to measurement conditions. "Typical" refers to measurement carried out at 23 °C (the calibration temperature) +/- 5 °C. Maximal refer to measurement carried out at any temperature between -40 °C and 70 °C.

It simply means that for a measurment carried out at 23 °C if measurement is equal to 5VDC, the accuracy is -/+ (0.03% of 5V + 0.008% of 62.64V).

Question 2) in the NI 622x specifications manual, it is stated that Specifications listed are typical at 25 °C unless otherwise noted.

Typical specification are used most of the time, but it is allways specified.

Hope this responds to your questions.

Regards

[bidouille404]

---

Audrey P a écrit :

Hello bidouille404,

about your two questions:

Question 1) in the NI 9229 Operation Instructions and Specifications manuel the terms "typical" and "maximum" as on page 24 (Gain Error) refer to measurement conditions. "Typical" refers to measurement carried out at 23 °C (the calibration temperature) +/- 5 °C. Maximal refer to measurement carried out at any temperature between -40 °C and 70 °C.

It simply means that for a measurment carried out at 23 °C if measurement is equal to 5VDC, the accuracy is -/+ (0.03% of 5V + 0.008% of 62.64V).

Question 2) in the NI 622x specifications manual, it is stated that Specifications listed are typical at 25 °C unless otherwise noted.

Typical specification are used most of the time, but it is allways specified.

Hope this responds to your questions.

Regards

---

Thank you for your answer. You are correct concerning the difference between "maximum" and "typical". It simply refers to the ambiant temperature.

I have an extra question then. What is the level of confidence (or number of "sigma") associated with the given accuracies? Is it a 95% level of confidence value (2 sigma)? Or something different?

This point is quite important when expressing uncertainties. In my lab, we usually give "2-sigma expanded uncertainties" but I know this is not always the case. I could not find this information in the NI documentation.

All the best.

[Hense]

Hello,

I have the same question as Audrey_P: Are the uncertainties presented on NI-manuals based on a 2 sigma (95%) or 3 sigma (99%) confidence level?

Kind Regards,

Hense

[John_P1]

Which manual?

The 622x specifications (which were linked earlier in this thread) include a chart that lets you calculate the accuracy specification at any given operating condition and confidence level.  The "absolute accuracy at full scale" in the chart is derived using the parameters specified underneath the chart itself (in this case, 100 points of averaging and a 3 sigma coverage factor).

Some other products don't give nearly as much information in the manuals though.

Best Regards,

[Hense]

Thanks John! THis was the answer I was looking for: (in this case, 100 points of averaging and a 3 sigma coverage factor).