XRF Capabilities - Correlating 2 different XRF machines - Ni/Au parts

Does anyone have experience correlating 2 different XRF machines? We do incoming inspection on Ni/Au parts. The supplier only measures Ni. The XRF manufacturer told us and the end customer 5-10% could be expected on both pieces of equipment (+/-5% for the cal standard, +/-5% for the machine).

Since we are not measuring the exact same part or the same plating (Ni vs Ni/Au) how do we complete GR&R, or can we?

FPT2001 said:

Does anyone have experience correlating 2 different XRF machines? We do incoming inspection on Ni/Au parts. The supplier only measures Ni. The XRF manufacturer told us and the end customer 5-10% could be expected on both pieces of equipment (+/-5% for the cal standard, +/-5% for the machine).

Since we are not measuring the exact same part or the same plating (Ni vs Ni/Au) how do we complete GR&R, or can we?

Click to expand...

For clarification purposes, Au=gold, and Ni=nickel, and I'm assuming that what you're dealing with is Au/Ni and not the other way around. Why is the supplier measuring only the nickel? Have you attempted correlation? Are you dealing with electronic connectors? If they're small parts you should have the supplier measure some, mark them and send them to you for measurement.

Added in Edit: If you're making comparisons, it's also important that you and the supplier are using the same size collimator and aiming it at the same spot on the parts. Also, I'm not sure about your reference to GR&R--you and the supplier should be doing that separately, but you could use the same parts for correlation purposes.

Thanks Jim,

These are stainless parts that are Ni plated on a reel plater and then go through electroless Au. Your first thought is that the plater is trying to make sure the Ni is within spec before they plate Au... not so, we just found out they are stripping the Au because their XRF Ni measurement variation is too high when they measure through Au.

We are simply doing our own GR&R and microsectioning to confirm the plating thicknesses.

FPT2001 said:

Thanks Jim,

These are stainless parts that are Ni plated on a reel plater and then go through electroless Au. Your first thought is that the plater is trying to make sure the Ni is within spec before they plate Au... not so, we just found out they are stripping the Au because their XRF Ni measurement variation is too high when they measure through Au.

We are simply doing our own GR&R and microsectioning to confirm the plating thicknesses.

Click to expand...

Although I have no experience with SS as a substrate for any sort of plating, Are you saying that the NI is plated in a continuous (reel-to-reel) operation, and then subsequently goes through an electroless Au process? If that's the case, why can't they measure the Ni before the Au is applied?
At any rate, once again I suggest that you get together with your supplier and measure the same parts. Because microsections will always be the "referee" if a pi$$ing match develops, both of you can compare your XRF results to microsections of the same parts if necessary.

Without getting into the specifics of the process, An MSA is specific to a gage or family of gages. While you may make some limited comparisons of separate MSAs done on different gages such as which has less measurement error, your problem sound more like inter-laboratory correlation.

The most appropriate tools would be a Youden Plot (a.k.a. Iso-plot) followed by an Intra-class correlation.

Miner said:

The most appropriate tools would be a Youden Plot (a.k.a. Iso-plot) followed by an Intra-class correlation.

Click to expand...

With due respect, that's probably overkill. It's usually enough in these situations (where there are often "discussions" over the seventh decimal place) just to compare the numbers.

FPT2001 said:

Does anyone have experience correlating 2 different XRF machines? ..............................

Click to expand...

I am curious whether you validate the machines and if you do, have you tried doing it by using the same certified reference material?

Yes, we validate, but only IQ & OQ's.

We recently had a rep bring several standards and we saw some variation due to the stainless steel types. Seems iron content, even slight had an impact on the readings. This is a surprise even though Ni & Fe are close on the periodic table.

We have found that using actual parts as standards works best for our application. It takes a few microsections but it is worth it