Strain Gauge Surface Prep - Why Higher Grit?

[brianpaul]

The engineers in our group are having a discussion as to how to convince our tech's they need to be increasing their sand paper fineness when they prepare surfaces for installing gages. (Instead of arguing, it is some times easier to convince.)

Everything I find in the literature make your surface flat (grind if you have to), start with 220, move to 320, and on up to 400, with different sources saying when to apply surface prep.

The question naturally arose, how do we know what is smooth enough. I found literature that says a surface rms of 63-125 microinches is good for general stress analysis. A surface roughness meter didn't seem practical (or funded), so I bought a Surface Roughness Standards book to do finger scratch comparisons.

The problem starts because after applying the pneumatic grinder, the grinder pad has been flattened down and the surface appears polished. Its an uncontrolled polish, but finger scratch compares. This doesn't seem good enough, but how do you convince a tech to then begin incrementing with sand paper.

 

[brianpaul]

In otherwords is there any literature, or convincing description that describes the relationship between adhesive film height/distribution compared with gage length. If the gage is attached to the adhesive, and the adhesive is attached to the steel, and the steel strain causes adhesive strain which in turn causes gage strain, how bad is bad?

How does a rough surface filled in with adhesive lead to poor enough strain distribution to inadequately describe the strain underneath the rough surface?

Conversationally we are describing the rough surfaces as peaks and valleys being shifted by the strain underneath the peaks and valleys. The glue is filling in between and attaching to the peaks and valleys, transfering strain to the gage.

How deep can those valleys be before causing a problem? What does the stress distribution look like through the glue to really be causing this problem?

 

[brianpaul]

Another thing! An engineer here argues that the error caused by bad surface prep is small compared with the total error in the data acquisition system.

I say that isn't true, but how?

Does anybody have a referenced study proving this?

 

[Bribyk]

Have you talked to a strain gauge manufacturer?

 

[BobM3]

Nothing like reading it in writing from the manufacturer. Check out application note B-129 from Vishay:

http://www.vishay.com/strain-gages/knowledge-base-list/appnotes-list/

 

[TheTick]

Have you tried comparing strain gage results with calculations? Maybe the current methods are working and need no refinement.

 

[brianpaul]

Bribyk,

I just got off the phone with one of the Vishay guys. He said some interesting things.

1) I described our tech using the pneumatic with scotch-brite pad to remove the paint and flatten the surface. As the pad is used, the pad is smoothed, causing the surface to be polished. The Vishay guy this surface was too smooth. So the subsequent 220-320-400 roughens the surface to allow better bonding.

2) I asked what is the difference in accuracy between a roughness of 100 micro inches (general stress) and roughness of 250 micro inches (high elongation). He replied less than 1%.

I finally asked if there have been any studies written up detailing this, and he replied that there might have been, but he's never seen any.

----------------

This is all very interesting, because 5 minutes earlier I had an interesting conversation with an old tech (who doesn't do strain gages) with alot of machine shop experience. He said using a pneumatic grinder with scotch-brite pads followed by finger-sanding has to be causing significant waves on the metal surface. Maybe you couldn't see them with your eye, but it was essentially gouging on the metal.

When I asked his opion if such waves would cause significant strain variances, he said he was certain they did.

 

[brianpaul]

BobM3,
Thanks for the help, but that was the paper I already referenced.

TheTick,
I appreciate your advice. Are gages are usually in locations that are difficult to calculate for with loading that isn't measured.

 

[IceStationZebra]

"He said using a pneumatic grinder with scotch-brite pads followed by finger-sanding has to be causing significant waves on the metal surface."
I do this all the time with no problems. You can do a lot more damage with a power tool than with your finger. You only need to rub the emery paper ~10 times to roughen the surface. I normally only perform one wet abrasion step with 320 or 400 grit and M-Prep Conditioner A. (This helps keep the emery paper from loading up with material and makes the next cleaning step easier.) If you are getting marginal adhesion you can tell because you will accidentally pull some of your gages off when you pull the tape off. (And most novice installers use too much glue or don't clean well enough.)

One thing the paper doesn't specify, but I think is alluded to with the different installation types, is the type of adhesive used. In general analysis you normally use cyanoacrylate glue, whereas the other types use epoxies.

In general I agree with your engineer that you are chasing a small error. There are other factors that can create much bigger issues:
-Stretching gage during installation (caused by "wiping" the gage onto the surface, or using too much uneven clamping)
-Wiring (poor soldering, 2 vs. 3 vs. 4 wires)
-Instrumentation error (hardware and operator error)
-Orientation error to axis of strain

ISZ