Strain gages and DAQ systems... 3

[chazRoman]

I'm looking to observe the deformation during a cycle test of a piece of exercise equipment. I figure I can use FEA to determine points of greatest concern and simply place strain gages at those points (assuming, of course, they are external).
My issue is that the company I work for doesn't have any program for such testing, so I need to propose the idea for funding.
Does anyone have any experience with selecting the right set up? For instance, I've seen two sets of strain gages that look almost Identical and and both constantan-based that vary in price by about 60%. The same goes for DAQ software, which I have found rudimentary versions of for free, or professional versions for over $4000 U.S.

How am I to know what I need?

 

[IceStationZebra]

I have installed hundreds of strain gages on off-road machinery and have turned many machine components into torque transducers and load cells. I have selected and setup the data acquisition for two different companies and have dealt with at least 8 different data acquisition systems.

First, strain gages are sensitive to orientation. You have three choices here.
A. Use a rosette, which you may not not be able to install depending on the location. If you are near a stress riser the best bet is to use a stacked rosette. Non-stacked rosettes are good in open areas.
B. Guess
C. Try to use FEA deformation to figure out the principle strain axis.

Second, strain gage selection
A. The gage has to be selected for the material in question.
B. Size - Strain gages "average" the strain over the area they cover. Large gages average a bigger area, small gages are better at determining stress risers.
C. Resistance - The big two are 120 & 350 ohm. 350 has less current draw and self heating but 120 is a little easier to get. Assuming you will be using 1/4 bridge circuit, so your gage must match your completion resistors. (A way to get around this is to put 3 dummy gages on a coupon and leave it near the 4th gage so they are all at the same temperature. The hook them up as a full bridge.)
D. Prewired costs more but will probably save some headaches for your first time - especially if you are not good at soldering.
E. Recommendations (I am assuming steel given the application.)
Omega.com - KFG-3-350-C1-11L3M3R ($165.00 for 10, pre-wired, 3 mm grid, 350 ohms)
Vishay.com - 125UN-CEA

Instrumentation - This depends on your needs. Give me more info and I will elaborate.
- Do you want to do static/quasi-static or dynamic testing? This determines the sampling rate.
- I would recommend a quarter bridge with 3-wire hook-up.
- How many at one time?
- Is it on a rotating component?

Installation can be a tricky so plan on wasting some gages. Most people who sell gages sell a kit to get started, but you can buy the stuff much cheaper on your own.

Omega and Vishay both have good info online.

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

ISZ

 

[rb1957]

well after that long and very complete post, i'd add only that you have to be very carefull if you are trying to detect the stress peak from your FEM, 'cause it is usually near a discontinuity and in a high stress gradient. i'd recommend looking for a couple of points where maybe the FEM shows some reasonably detectable difference for the load being applied, some points to calibrate the model.

 

[IceStationZebra]

Good point RB1957. I have one FEA guy who always requests several rosettes in the middle of large pieces so he can calibrate his models. I always give him a good natured ribbing for taking up precious channels with "frivolous & expensive" rosettes. (His models are usually correct)

ISZ

 

[chazRoman]

IceStationZebra,

to answer your questions, we are performing quasi-static testing.
I believe we would be using three separate locations, possibly with two or three axes each, therefore, I think up to nine channels may be necessary. As for rotation, no, these devices move back an forth linearly.

 

[IceStationZebra]

This is pretty simple from an instrumentation standpoint. The sample rate requirement is low so almost any data acquisition will work so long as it can handle all 9 channels and has onboard excitation. The big question now is do you want a stand alone data logger or can you use a laptop? The laptop will probably be the cheaper alternative. You should be able to get away with using Excel to read the data as long as you don't go over 63k samples.

As far as resolution goes you should get at least 12 bit, which isn't that hard to find. Esentially you are taking an analog signal, which can be any consevable value, and recording it as one of X descrete numbers. A 12 bit conversion takes the total measurement range and breaks it up into 4096 descrete values, and the actual analog value is reported as whichever digital value it most closely matches.
10 bit = 2^10 = 1024 values
12 bit = 2^12 = 4096 values

I will plug a book that is a good introduction to data acquisition. It used to be free, but is in such demand that they now charge $30.

http://www.iotech.com/handbook/handbook.html

ISZ