What is the most appropriate MSA (Measurement System Analysis) Study

[Calibrator24]

We have a weighing scale in our company that was being used to count devices. This weighing scale was calibrated thru comparison with standard weights. The actual application required the user to configure this scale to convert it's function from weighing to counting. Since the devices naturally have variations on weight, error on the actual counts cropped up. The customers were complaining if there were shortage and the company doesn,t want to have excesses in the shipment either.

Can we correct this problem thru the use of an MSA study? what would be the best study to undertake if we want to have an accurate counting instrument? and how it should be done? Any insight? Thanks.

 

[Stijloor]

We have a weighing scale in our company that was being used to count devices. This weighing scale was calibrated through comparison with standard weights. The actual application required the user to configure this scale to convert it's function from weighing to counting. Since the devices naturally have variations on weight, error on the actual counts cropped up. The customers were complaining if there were shortage and the company doesn't want to have excesses in the shipment either.

Can we correct this problem through the use of an MSA study? what would be the best study to undertake if we want to have an accurate counting instrument? and how it should be done? Any insight? Thanks.

Any MSA Experts who can help?

Thank you!!

Stijloor.

 

[The Specialist]

Whilst not a specific method for MSA, I do have a couple of suggestions…

Firstly,

Whilst conducting your calibration trials on the weigh-scale, have you also conducted eccentricity and repeatability trials?

Secondly,

Calculate the max. no. of items you can weigh collectively, before the accumulated potential tolerance is greater than that of the minimum accepted weight of an individual unit.

e.g. you will increase the accuracy if you weigh fewer units in your bulk load... by preventing the possibility of the accumulated total upper and lower tolerances (weight) from being greater or lower than that of a single unit.

 

[keithljelp]

Hi,

A MSA is not going to help you. The MSA will only tell you if your measurement system is capable and you all ready know it is not since you are getting inaccurate counts. It is simply a matter of the resolution of the scale you are using. The scale must have a resolution 10X smaller that the variation in the weight of the parts you are trying to count. You can sometimes use 5X smaller.

Example:

If the part's weight varies by plus or minus .5 gram then a scale with a .5 gram resolution will not always give you an accurate count. You do not know how the microprocessor was programmed. Does it round up or down or truncate??? You would need a scale that can resolve to at least a .1 grams and .05 grams would be better. The scale must also have a maximum load that is greater than the total weight of parts you wish to count by at least 20%.

If the scale is located in shipping be careful of wind currents. They will affect a scale.

Hope this helps,

Keith

 

[bobdoering]

We have a weighing scale in our company that was being used to count devices. This weighing scale was calibrated thru comparison with standard weights. The actual application required the user to configure this scale to convert it's function from weighing to counting. Since the devices naturally have variations on weight, error on the actual counts cropped up. The customers were complaining if there were shortage and the company doesn't want to have excesses in the shipment either.

Can we correct this problem thru the use of an MSA study? what would be the best study to undertake if we want to have an accurate counting instrument? and how it should be done?

Here's the problem, it's not the scale's fault the product varies. Typically, a Gage R&R would use the same parts, which means it will not collect the variation you are experiencing and the scale will look great - but you will still have rejections. Calibration will have little to do with your problem (although it needs to be calibrated.)

You need to look at the problem as a product capability issue. If your carton has 1000 parts and each part is supposed to weigh 20 grams, then the scale should weigh out 20,000 grams per lot. To have an error of 1 part - based on part variation - the variation of the individual parts would only have to be .02 grams each. So, you need to look at the per piece weight capability over several lots (because that is where the variation will be the most) to see if weigh counting will even work. Also, for lighter parts, even dunnage weight variation can make a huge error in part counts!