SPC Data presentation - The best way to present histograms in a dashboard format

Z

zibsan

What is the best way to present histograms in a dashboard format? I come from the land of electronics. Pareto and pass fail. I now have a PCB with a bracket from .... on it and want to know how best to portray the variance from 8 key locations on each sample.
 
Last edited by a moderator:

Stijloor

Leader
Super Moderator
Re: SPC Data presentation.

What is the best way to present histograms in a dashboard format? I come from the land of electronics. Pareto and pass fail. I now have a PCB with a bracket from .... on it and want to know how best to portray the variance from 8 key locations on each sample.

Can you clarify a little bit more? Are you concerned about the location variation of 8 brackets? Do you already have eight data sets that you want to portray? :confused:

Stijloor.
 
Last edited:

Steve Prevette

Deming Disciple
Leader
Super Moderator
What is the best way to present histograms in a dashboard format? I come from the land of electronics. Pareto and pass fail. I now have a PCB with a bracket from .... on it and want to know how best to portray the variance from 8 key locations on each sample.

I'm a little confused as your post title refers to "SPC Data Presentation" then in the text you are asking about histograms. If you are indeed interested in SPC Data Presentation in a Dashboard method, I do have a method documented at http://www.hanford.gov/rl/uploadfiles/VPP_21_Dashboards.ppt
 
Z

zibsan

Maybe I am reading way to much into this. I have a machined part with 8 key dimensions. The recieving inspector will measure all 8 points on each piece. Lot size is about 25 pc. They will input the data and now I want to show varience from part to part for each key (8) then lot to lot. I want something simple that I can quickly OK or reject based on the data. :(
 

Steve Prevette

Deming Disciple
Leader
Super Moderator
Maybe I am reading way to much into this. I have a machined part with 8 key dimensions. The recieving inspector will measure all 8 points on each piece. Lot size is about 25 pc. They will input the data and now I want to show varience from part to part for each key (8) then lot to lot. I want something simple that I can quickly OK or reject based on the data. :(

Since you mentioned machining, you may want to take a look at Bob Doering's posts here on the Cove on the subject.
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Maybe I am reading way to much into this. I have a machined part with 8 key dimensions. The receiving inspector will measure all 8 points on each piece. Lot size is about 25 pc. They will input the data and now I want to show variance from part to part for each key (8) then lot to lot. I want something simple that I can quickly OK or reject based on the data. :(

Do you have some sample data that we can use to fab up a chart? I have some ideas, but the actual specifications of the 8 characteristics and some typical data would be handy to determine how many charts you will ultimately need.

If you are doing incoming inspection sampling, are you using c=0 sampling plan?
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Yes I am using c=0 at a 1.0 Aql.

I roughed in a chart that will color code individual data for out of control or out of specification, and plots the high/low values of the dimensions to better illustrate the variation of the dimension within the lot. This is a better view of the data than the 'mean' for what you are doing. You can delete any extra sample lines in the chart that you are not using. You will have to set the chart axis up for your spcifications.

The real problem is the distribution of the dimensions you will be receiving - especially the lengths and diameters. That depends on how the process is controlled, which I doubt is being properly controlled, and the measurement error of only taking one measurement of the characteristic - you will like see a normal distribution. If you attempt to determine any "trends" from such erroneous data, it will likely not be very accurate. But, the high/low give a good visual of how much of the tolerance is being used up in the process.

You may want to suggest that your vendor review the correct way to control a machining process to help the consistency of your incoming materials. It will also allow you to set up your incoming control limits to match their process control limits.
 

Attachments

  • SPC Autoplot Special.xls
    81 KB · Views: 468
Z

zibsan

Bob,

Thanks , it must have taken you sometime to do this and I am thankful. However, I have 8 different dimensions per part that I need to check.
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Are all eight dimensions the same specification - or different? If different, what are the specification ranges and type of characteristic (e.g. length, diameter, location)?

Are you looking for an overall "dashboard" of all 8 dimensions for the lot, for a series of lots (if so, how many) or period of time (if so, how long)?
 
Last edited:
Top Bottom