ALTA 6 and MLE

[Steve]

Does ALTA 6 allow for regression analysis on data sets or is MLE the only option? I am using cumulative damage with a Weibull life distribution and IPL for cumulative damage relation.

[David]

Hi Steve,

MLE is the only analysis method available in ALTA. Regression can run into problems when trying to analyze the data. Further detail is presented at http://www.weibull.com/hotwire/issue30/tooltips30.htm under the second tool tip.

I hope this helps.

[Steve]

Ok, thanks.

Another question: I am finding that while my stress is voltage (and nothing else) the analysis indicates that as voltage increases, my reliability goes up. What I am testing is simply a heating element. This makes no sense to me. Is it because I have only 5 failure points? I am using a simple step stress/cumulative damage model.

[David]

With only five failures while applying the Cumulative Damage model, there is not much information to go on. In terms of life increasing with stress, you might even be able to deduce this by simply looking at the data. With only five failures I would be a bit hesitant regarding the results when applying the Cumulative Damage model. An accelerated test with a time-varying stress generally requires more units than one with a constant stress to obtain the same level of confidence. This is due to the increased complexity of incorporating a time-varying stress. Even though analysis can be conducted using a single stress profile, we also recommend applying multiple stress profiles, similar to the application of multiple stress levels when applying a constant stress. This will allow you to collect more detailed information as to how stress is affecting life.

[Steve]

Does AltoPro 6 allow for using interval data? I can't seem to find that functionality.

[David]

Hi Steve,

Unfortunately, ALTA 6 does not have the ability to analyze interval data. However, this was added to ALTA 7. An evaluation version of ALTA 7 can be downloaded for free from our Download Store at http://www.reliasoft.info/download/p...9&cat=6&page=1.