"Historical std deviation" (Standard Deviation) in Gage R&R study

[Berg.Jlle]

Hi.

I have two questions and I'd like any help.

1) What is the difference in use or not of "Historical standard deviation" in a Gage R&R study, since the %SV and %Tolerance are not affected by this value ?

2) I have a measuring instrument that I wish to make an analysis of this measurement system MSA (Gage R&R study + Gage Linearity and Bias Study) to verify its performance.
This instrument is inserted into the production line and does the measurement of 4 types (A,B,C,D) of parts with the following specification limits:

Type A > LSL: 1,5; USL: 2.5
Type B > LSL: 2.5; USL: 3.5
Type C > LSL: 3.5; USL: 4.5
Type D > LSL: 4.5; USL: 5.5

I will collect samples of typical production for each of 4 types (A,B,C,D).
My question is if I should do a single study with data from all 4 types together considering a single specification limit (LSL:1.5 and USL:5.5) or if I should do a study separately for each type ?

[Miner]

In Reply to Parent Post by Berg.Jlle

1) What is the difference in use or not of "Historical standard deviation" in a Gage R&R study, since the %SV and %Tolerance are not affected by this value ?

Entering the historical standard deviation will provide another metric called the %PV or %Process Variation. The %PV should be used instead of the %SV as it is based (presumably) on many more parts than was used in the R&R study. The historical standard deviation is typically obtained from a capability study or based on SPC data.

In Reply to Parent Post by Berg.Jlle

2) I have a measuring instrument that I wish to make an analysis of this measurement system MSA (Gage R&R study + Gage Linearity and Bias Study) to verify its performance.
This instrument is inserted into the production line and does the measurement of 4 types (A,B,C,D) of parts with the following specification limits:

Type A > LSL: 1,5; USL: 2.5
Type B > LSL: 2.5; USL: 3.5
Type C > LSL: 3.5; USL: 4.5
Type D > LSL: 4.5; USL: 5.5

I will collect samples of typical production for each of 4 types (A,B,C,D).
My question is if I should do a single study with data from all 4 types together considering a single specification limit (LSL:1.5 and USL:5.5) or if I should do a study separately for each type ?

Are these similar types of measurements? For example, they are all lengths of the same type, with different nominal dimensions? If they are, you may be able to treat them as part families, and perform R&R studies as I describe in my linked blog.

Linearity and bias studies are typically performed using Master parts that have been measured using a more precise measuring device. If you can do that, you can use the results of all four parts.

[Berg.Jlle]

Hi Miner.

First thank you for your patience.


1) There is only one measurement device used for all four product types (A,B,C,D).
2) All measurements are voltage measurements [V] and they are measured in the same point of the circuit. The value is read on a digital display.
3) Spec for product types:

A: Target = 2.00 V
LSL = 1.50 V
USL = 2.50 V

B: Target = 3.00 V
LSL = 2.50 V
USL = 3.50 V

C: Target = 4.00 V
LSL = 3.50 V
USL = 4.50 V

D: Target = 5.00 V
LSL = 4.50 V
USL = 5.50 V

4) I intend to take a sample of 10 parts of each product type (A,B,C,D) with a total of 240 measurements (10 parts X 3 operators X 2 measurements each operator X 4 product types).
5) I wish to use the gauge for SPC. I don't know the process variation (std dev) of each product type yet.

Suppose the process variation (std dev) are: A (0.12), B (0.10), C (0.11) and D (0.13). Could I make a single Gage R&R study considering as "historical standard deviation" 0.10 (smallest std dev) using all 240 measurements ?


Miner, sorry if I wrong understanding.

Best regards.

[Miner]

No. Since you do not actually have a historical standard deviation, stay with the %SV using the part variation.

[Berg.Jlle]

In Reply to Parent Post by Miner

No. Since you do not actually have a historical standard deviation, stay with the %SV using the part variation.

Miner,

What is still not clear to me is whether I should do a Gage R&R study for each of the product types (4 studies X 60 measurements) or can I consider the four product types as a single family and do a single Gage R&R study (240 measurements) ?

Is it correct to conclude as follows ?

1) If I have the following conditions satisfied for the four product types (A,B,C,D), then I can consider the four product types as a single family and do a single Gage R&R study.

- Same gauges.
- Same product characteristics.
- Same process tolerances or variation depending on use of gauge.

2) If some of the conditions listed above are not met or if I'm not sure of all them, I must consider the four product types (A,B,C,D) as four families and do four Gage R&R studies separately.

[Berg.Jlle]

In Reply to Parent Post by Berg.Jlle

Miner,

What is still not clear to me is whether I should do a Gage R&R study for each of the product types (4 studies X 60 measurements) or can I consider the four product types as a single family and do a single Gage R&R study (240 measurements) ?

Is it correct to conclude as follows ?

1) If I have the following conditions satisfied for the four product types (A,B,C,D), then I can consider the four product types as a single family and do a single Gage R&R study.

- Same gauges.
- Same product characteristics.
- Same process tolerances or variation depending on use of gauge.

2) If some of the conditions listed above are not met or if I'm not sure of all them, I must consider the four product types (A,B,C,D) as four families and do four Gage R&R studies separately.

Please, other participants are also welcome!

[Atul Khandekar]

In Reply to Parent Post by Berg.Jlle

What is still not clear to me is whether I should do a Gage R&R study for each of the product types (4 studies X 60 measurements) or can I consider the four product types as a single family and do a single Gage R&R study (240 measurements) ?

Separate study for each product type. Clubbing all types (nominals) into one single study would artificially inflate the Part Variation.

In Reply to Parent Post by Berg.Jlle

1) If I have the following conditions satisfied for the four product types (A,B,C,D), then I can consider the four product types as a single family and do a single Gage R&R study.

- Same gauges.
- Same product characteristics.
- Same process tolerances or variation depending on use of gauge.

Yes.

In Reply to Parent Post by Berg.Jlle

2) If some of the conditions listed above are not met or if I'm not sure of all them, I must consider the four product types (A,B,C,D) as four families and do four Gage R&R studies separately.

Yes.