ANSI Z540-3 Risk Analysis - How to do the math for risk analysis

Stijloor

Stijloor

Leader
Super Moderator
dv8shane said:
I am currently compliant to Z540-1 but I wish to go to the new -3 standard
Does any one understand how to do the math for risk analysis as in http://www.a2la.org/guidance/Risk_Analysis_Z540-3.pdf

I am unsure if my process that involves guardbanding measurements that have a TUR<4>1.5 to 80% to reduce false acceptance/rejection based on Dave Deaver's http://assets.fluke.com/appnotes/Calibration/ddncsl94.pdf meets the new requirements.
Click to expand...

Can someone help?

Thank you!!

Stijloor.
 
G

George Weiss

Saturday morning food for thought only:

The formulas within the linked page discuss means of controlling false-accept, and in at least one case, managing it to a maximum of 2% for best fit to Z540.3
In the A2LA paper by Howard Castrup, Ph.D. the first step is that:
Z540.3 clause 5.3 allows meeting a TUR requirement of 4:1 in cases where estimating false accept risk is not practical. SO: if TUR is >=4:1 then no PFA <2% analysis is needed for Z540.3
I have a focus of getting the item tested with <2% false accept as a requirement limit. The papers have varied methods for your confusion, to allow options.
If getting through the Z540.3 door is the quest, and the path of least resistance is desired, then set a minimum confusion course. Selecting an accepted method such as NCSL’s Recommended Practice RP-10. The formula is simple to apply to each test parameter measured with the listed formula:
K=1.25-(1/TUR), or K=SQRT(1-(1/TUR*TUR)) So test limit = specification limit in first formula for TUR = 4:1. EASY ! This will allow for acceptable Z540.3 practice, and no need for re-inventing the wheel until you decide you want to.

SO:
  1. Pick a managed risk course, so you have a course. This means pick a formula.
  2. If the NCSL’s method, then you have a supporting document, and need not spend nights doing further analysis.
  3. If TURs are over 4:1 then as in Z540.1 just test the item without guard bands.
  4. Use the formula to add guard bands in cases where TURs <4:1
  5. Report on your datasheets: DUT spec, total k2 uncertainties, guard banded test limits.
You have arrived at the finish line of race #1. Now improve on the managed risk method if desired.
The math is only: test limits = DUT spec * K, which is 1 for TUR 4:1 easy right !
I am open to comment, and expect to be shot down/up. Have a nice day !
 
You must log in or register to reply here.

Similar threads

Ajit Basrur
Facts and Myths about Generic Drugs
Replies
2
Views
4K
Ajit Basrur
Ajit Basrur
Top Bottom