Testing Tips: February 11th, 2016
How do you set up the Amplitude Controller?
Last week we talked about how the Amplitude Controller could automatically alter the command drive signal so that the desired peaks are being met. This week we will talk about techniques for setting up the Amplitude Controller so that it does its job correctly. Here are several tricks for setting up the Amplitude Controller properly:
- When you calibrate the test system, give yourself at least 20% headroom on your test levels. In other words, if you want to run a test at +/- 100N, don't use a transducer that is calibrated to 100N. Instead use a transducer that is calibrated to 120N fullscale. The reason for this is that a transducer that is calibrated to 100N will have a signal that will saturate and clip at 100N. This means that if the applied load is something like 102N, the transducer will only display 100N. If the amplitude controller overdrives the command signal, it has no way of knowing it has done so. Likewise, if you keep the test running well within the calibrated range, you reduce the chance for waveform distortion that can sometimes occur near the end of the transducer full scale range.
- If your system has software filters for reducing the noise on the transducer signal, set your filter cutoff frequency at least 10 times the desired test frequency. Setting the filter at too low a cutoff frequency will attenuate (reduce) the signal undesirably. For example, if you are running your test at 50Hz and use a first order filter with a 250Hz cutoff frequency, the signal will be reduced by 2% whereas a 500Hz filter will reduce it less than 0.5%. On the flip side if you don't use any filter or a filter setting that is too high, noise artifacts that can create erroneous peak readings may be let through.
- Before starting the test, turn your PID settings down. The last thing you want to do when starting the test is to overstress the specimen. Ideally you will want to set the test up initially with a specimen that has similar response to the one you will ultimately be testing. Once the test is running, you can adjust the P setting upward until you achieve peaks that are about 90% of the desired peak levels. By boosting the command waveform the Amplitude Controller increases the error signal that is output to the drive amplifier and this is similar to increasing the P term. The Amplitude Controller also compensates for any offsets in the control loop so the I-term is not usually needed. When using the Amplitude Controller, the D-term is also not usually needed.
- Make sure your peak/valley readings are reliable. For example, let's say you are trying to run a test at +/-100N. After you start the test you see from your scope screen that the transducer trace is cycling between +/-90N. At the same time you notice that your peak/valley readings appear like random readings. Random peak/valley readings are usually an indication that the tolerance band for your peak/valley detection routines is too small. In this case the peak/valley routines are actually displaying the transducer noise and since the signal is also changing substantially the peaks and valleys appear random. To make the peak/valley readings display properly you will need to increase the tolerance band to about 2 to 5% of your test waveform amplitude. So if your waveform is +/- 100N, set the tolerance band to 2 to 5N. We will talk more about setting the peak/valley tolerance band in an upcoming test tip.
- Don't set the Amplitude Correction level too high. We mentioned before that amplitude correction is an iterative process whereby the difference between the desired and actual peak levels is multiplied by a small percentage and the resulting number is added to the command level signal during each cycle. If you set the amplitude correction too low, it will take more cycles to reach the desired levels. This is actually OK and it means you will have to wait a little longer to reach the desired levels. If you set the percentage too high, you will reach the desired levels quickly but your peak readings will look noisy as the system overcorrects above and below the desired peak levels. It's OK to set the percentage on the lighter side.
As long as the specimen and system response are somewhat linear, the Amplitude Controller is easy to set up so that your desired test levels are applied precisely throughout the test. When we say linear response we mean that an increase in command creates a proportionate increase in the transducer reading. Where it starts to get tricky is when an increase (or decrease) in command creates a non-linear response in the transducer reading. An example of a system/specimen that generates a non-linear response would be a set-up where the specimen is squeezed between two platens in load control. Any time a compressive load is commanded the resulting load would be somewhat linear. Any time a tension load is commanded the load always reads zero since the specimen is not coupled to the load cell. In next week’s test tip we will talk about tricks for using the Amplitude Controller to control non-linear types of systems.