I am working on a project that is wanting to measure strain of a structure and they have placed 128, 120ohm, quarter bridge strain gauges evenly throughout the structure. My understanding is this: With typical wiring of the quarter-bridge gauges into 4WFB120's and using AM16/32B multiplexers, the maximum amount of guages that could be monitored by a CR1000X datalogger is 64 because it can only connect up to 4 multiplexers and the multiplexers will need to be operated in 4x16 mode (the second channel is required for excitation of the bridge.) One option we may be considering is, if we provided an external excitation source (considering 5V), could we operate the multiplexers in 2x32 mode and, if so, will this complicate programming of the CR1000X? As I understand the operation of the multiplexers, we would simply be rotating through the 32 channels, each channel relaying a voltage reading back to the 4 input channels on the CR1000X which would then take this reading and perform its internal strain calculation. So, in my head it works, but I also understand that there could be an additional limitation here that I am not seeing.
The errors imposed by using a separate 5V source and running the multiplexer in 2X32 mode would be larger than the changes you expect to see in your readings. Signal to noise ration would be less than 1. Effectively your data would become useless.
Putting in 5V, you would be adding inaccuracy of that voltage source to your measurement. The datalogger must provide the excitation voltage, so it can do a full bridge ratiometric measurement. The rationmetric measurement completely factors out any error in the excitation voltage. The analog hardware of the CR1000X itself actually is 4 times as accurate on a full bridge measurement than on a differential voltage measurement.
Each Vx channel can only output about 25mA, which is not enough to excite 16 gauges at once, which is why we can't run all the excitation directly back to the CR1000X.
To read all the gauges, add a CDM-A116 with 4 AM16/32B connected to it.
JDavis,
To try to make sure I am understanding you correctly, using the AI Module, and 4 AM16/32B multiplexers, we should be able to monitor all 128 strain gauges? How does this get around having to measure in 4x16 mode on the 4 multiplexers in order to switch the excitation voltage? Or is 64 gauges still only possible?
Also, a few more (potentially) helpful pieces of information: I was mistaken, the project is not considering 120 ohm strain gauges, they are considering 350 ohm and discussions are underway to use full bridge gauges which should cut out the need of the TIMs.
As far as using an external excitation voltage source, our thought was to also read the external voltage into one of the open channels on the data logger to use for correction in the calculation, is this not achievable inside CRBasic?
You would have a total of 8 AM16/32B. The CR1000X has 4 connected to it, and the CDM-A116 has the other 4 connected to it.
The quarter bridge plus TIM and the full bridge gauges are both fundamentally the same analog measurement. You have the same basic connections on the AM16/32B. Using 350 Ohm gauges allows more sensors per excitation channel versus 120 Ohm gauges. Even with 350 Ohm gauges you can't run 16 gauges to one Vx channel. The excitation voltage would have to be reduced a lot to keep total current down, which reduces signal to noise ratio.
Reading an external voltage source adds another analog measurement, with its associated error. That would result in a number with 8 times the measurement error you would have doing a ratiometric measurement.
VoltDiff of excitation voltage plus VoltDiff of sensor signal:
(0.04% of voltage measurement + offset) * 2
BrFull measurement of sensor:
0.01% of voltage measurement + offset
If you are looking for big changes in strain, it might be acceptable to make the sacrifice of accuracy. I suggest running through the math to figure out what that measurement error would mean in terms of actual strain.