Comments/Questions/Suggestions for Project 1 -- Seismometer

• Some General Comments
  • You should all feel a sense of accomplishment after completing this project. You were able to generate a mechanical motion, sense it two ways to turn it into an electrical signal, condition one of the signals to be larger and possibly cleaner and display your results in a useful form on an oscilloscope and a computer. One more step and you could have stored the signal in a file on a computer and used the information for some purpose. In doing this work, you nearly all found some holes in your understanding of electronics, figured out how to correct your ideas and then made the system work. Good job by everyone.
  • The grades for this project are relatively high, since you all achieved the basic goals. In the comments that follow, you will notice that there were many things that most of you could have done to improve your work. These comments are not meant to indicate that you did a bad job. Rather, they are meant to be suggestions for future projects, offered in the context of this project in the hopes that they will be easier to understand than if just presented as a short lecture in class.

• Some General Issues That Relate To Any Project
  • Be sure that your report explains clearly what you are trying to do and why, how you did it and why the reader should be convinced that you have achieved your goals.
  • Do not include information that does not support your case.
  • Can you identify any design tradeoffs?
  • Of the various parameters that characterize a system under study, which have a significant effect on the desired performance?
  • In your report, relate your text to your figures. Do not try to explain just in words what can best be understood by a combination of words and pictures.
  • When you present your test results, explain how you obtained them. How did you excite your device? What happens when you excite it differently? Label your results.
  • After presenting your test results, discuss them. How do they compare with the predictions you made in your analysis? What features do you observe in the data? Can you explain any of the features? Why do you believe your results? How can you use the information in the results?
  • Be sure that your data is complete and convincing. Do not assume that the reader knows what you have done or why unless you tell them.

• Some Random Observations From The Project Reports
  • The pot setting was very sensitive. The pots we have in our kits are indeed difficult to set with great precision. There is another kind of pot that works better, the multi-turn pot, which can be set more easily. However, they cost more and the pot we have will work well enough if it is not the only resistor in the circuit. If an 1k pot, for example, is added to a 10k resistor, changing its setting will not have a great effect on the the circuit.
  • It was not possible to obtain a Lissajous pattern to print with HP-Benchlink, without using the AUTO STORE feature of the scope. Usually this is true, since the beam moves so slowly that only a part of the pattern is mapped out. The students who thought of this idea, did so without any staff help that I know of.
  • Understand the individual components first before trying the entire design. This excellent advice came from several groups who were not lucky enough to have their circuits work the first time. I have to admit that the groups who decided to first give the whole thing a try and see if everything worked before going to this step also were following a mostly reasonable path. However, there is always the possibility that whatever errors there may be might cause some of the components to self-destruct. This happened to a couple of groups that did not have the power connected properly to their op-amps. In general, it is best to test each sub-system before connecting them all together. This has the added benefit of producing data that can be understood.
  • Be careful while wiring your circuit. Things can get disconnected or burn up. Wires can easily break. Bad wiring practice can make it difficult to see such a break. It is also important to be sure you know where your power connections are being made. More than one group fried an op-amp by trying to make it work without power. Just about everyone will do this kind of thing at some time, particularly when we get to digital logic.
  • Since different beams were used each time and the individual component values can vary quite a bit, it is important to incorporate some flexibility in the design. Many groups added a pot here and there to be sure they could make adjustments when they had to use a different beam. A few groups complained about the variations of the beams.
  • Be sure you have power to your op-amps before you try to use them. It is easy to turn of the power to the op-amp when making circuit modifications. This is a good idea, since you never know what you might short out otherwise. However, it is all too easy to forget to turn a particular supply back on before applying a signal to the amplifier. Many times this will just produce some very odd results. Often, however, the op-amp burns up.

• My Comments
  • Many people do not seem to understand the concept of a practical schematic. Such a schematic should have each component easily identifiable so that building the circuit is made as easy as it can be. An example follows:
    
    Note that each component represents something to be loaded onto the protoboard and each pin of the op-amps is labeled. By the way, this is a very simple circuit for producing a square wave or a triangle wave and is, thus, a function generator.
  • Our seismometer-like projects generally consisted of four major sub-systems: the cantilever beam with its strain gauge and pickup coil, the bridge circuit for the strain gauge, the differential amplifier and a filter. We also used some power supplies, an oscilloscope and a computer (for HP-Benchlink). As some groups noted, it is very important to test each of these separately to be sure that they are working before trying to put everything together. Almost no one did this, however. While these systems are relatively simple, it is good to use a procedure that can be applied to more complex design tasks. Not only is it easier to spot problems this way, it is almost always easier to simulate the performance of each part separately and to understand what we see. This will also help to identify problems when they occur, whether or not they are important. For example, most groups had little problem getting a signal from their bridge circuit and, once they clarified some misunderstanding about the operation of a bridge and a differential amplifier, they were able to successfully connect them together. However, no one, that I can recall, commented on the fact that both of the inputs to the differential amplifier oscillated up and down together, even though one input was supposed to be from a DC voltage. This occurs because the negative feedback of the op-amp forces the two input voltages to be essentially the same. Thus, when one varies, the other must also. We see then that the bridge behaves differently with the differential amplifier connected. This different behaviour is something that we should notice but also recognize that it does not keep the amplifier from working properly. The information presented by several groups showing this effect was from PSpice. No one made any measurments with a scope except at the ultimate output of the circuit.
  • All the components mentioned in the last item were essential to the operation of the projects, except for the filter. More than half of the projects reported on invovled a filter, at least at the preliminary stage. Most groups were able to design a low-pass filter that had a corner frequency somewhat above the 20 Hz signal from the beam. The most popular approach was to use the formula for the corner frequency. Some groups also did a PSpice simulation of their filter. This is an example of the idea discussed in the previous item -- addressing each sub-system separately. By checking to be sure that the filter worked the way it was supposed to, some problems were avoided. Unfortunately, almost no one noticed that their choice of corner frequency was so close to 20 Hz that there was significant attenuation of the desired signal along with higher frequency noise. That is why we do the AC sweep analysis. If there is significant drop off in signal at 20 Hz, we need to raise the corner frequency. Having successfully designed a filter as the last stage of their project, most groups then did their final test and, being happy with the results, quit and wrote their report. No one did the obvious test of removing the filter to see what affect it had on the overall performance of the circuit. If they had done so, nearly all would have noticed two things. First, the filter made the signal at least somewhat smaller and, thus, negated part of what the amplifier had accomplished by increasing the signal. (One group did actually note that their amp had too much gain, but the filter brought it back to the desired level.) Second, the signal would be seen to be quite clean even without the filter. Thus, the filter proved to be an unnecessary addition to the project. I think there were at least two groups that had a filter in their preliminary design, but found good signal without one in their final design.
  • Much more often than not, reports included results from PSpice simulations and experiment which were not consistent. Sometimes there were comments on not trusting PSpice, but usually nothing was said. It is very important to include some discussion of how predictions and experiments compare. If they do not agree, there must be some good reason. The models used in PSpice are very realistic. Thus, the problem is usually in the way the circuit was setup. Most groups noted that it was at least somewhat difficult to balance the bridge well enough to get the amplifier to work properly. A significant imbalance caused the amplifier to saturate near plus and minus 15 volts. Most group's PSpice simulations showed the same kind of imbalance. The imbalance was due mostly to the difference between the two legs of their strain gauge bridges. The signal from the pot side, if only a pot was used, has about half the Thevenin resistance as the gauge side. This resistance gets added to the input resistance of the differential amp, which changes its gain. The two inputs do not have the same gain and thus a balanced bridge results in an unbalanced output. We saw this effect in Experiment 4, but no one discussed it or methods for mostly eliminating it. In a few cases, groups used the 1k pot as one resistor in the circuit and a fixed 1k resistor for the other. This should have largely removed the differnces in the Thevenin resistances of the two legs, except for variations in the individual components.
  • Part of the reason for asking that both the strain gauge bridge signal and the pickup coil signal be used in the project was to give you some experience with how different sensors see devices differently. There should have been a 90 degree phase shift between the two signals, since one is sensitive to position and one is to velocity. Only one or two groups discussed this in any detail and showed that their data was consistent with the expected difference. It is important to look for such features in the data that we can explain as evidence that things are working correctly.
  • Most groups changed the gain of their differential amplifier by changing the values of the feedback resistors while a few groups changed the values of the input resistors. Some changed both. One problem with changing the input resistors to raise the gain is that the resistor values need to become so small. Small resistors mean large currents, which sometimes cannot be handled by the op-amps. You should notice in most circuits (including the one above) that the resistors used are usually 10s of kilo-ohms. These larger resistors do not need so much current. Try to keep your values in this range and remember that you can change both the input and feedback resistors if you need a different gain.
  • In any design there will be some choices. Some system parameters can only be enhanced at the expense of others. Some parameters can affect the desired performance and some cannot. For example, it is possible to increase the signal from the bridge by increasing the DC voltage across it. Is this in any way preferable to increasing signal by raising the gain of the amp? Increasing the signal before amplification is usually preferred because the ratio of signal to noise will be better. The amplifier will usually also amplify the noise and the signal. However, in our case, the signal and noise levels were fine without raising the DC voltage. Raising this voltage will stress the strain gauge (it has a maximum power rating). Thus, it was neither necessary nor prudent to increase the DC voltage. One could consider putting the filter before the amplifier as another method for increasing the signal-to-noise ratio. Some groups did use a filter at this location, but others found it made the operation of their circuit difficult. The filter probably caused further imbalance in the differential inputs. As noted above, the filter was usually not necessary. Since most of you are new to circuit design, you probably find it difficult to make design choices like this. When you are not sure of the outcome, but see that there are some clear choices, you can usually go ahead and try them to see what happens. If you get some significant improvement, then you can take the time to find out what you did.
  • Some groups calculated the Thevenin voltage and resistance of the bridge, but then did little with it. The best approach is to determine these parameters for each side of the bridge and then use them as inputs to the differential amplifier in a PSpice simulation. This is what was done in Experiment 4. The groups that did this kind of thing were able to determine the overall frequency response of their design in this manner.
  • A minor point that we will return to when we consider op-amps again is terminology. The op-amp is the chip or the triangle in the circuit diagram. The differential or inverting amplifier is an op-amp configuration involving an op-amp and some other components. There are inputs for the amplifier and inputs for the op-amp. It is good to distinguish these things for clarity.
  • Another minor point regarding the cantilever beams. The wires connected to the strain gauge are taped near their solder points to provide some strain relief. It is very easy to break the solder joint if the wires are yanked on too hard. Please recognize the mechanical limitations of electrical devices and be careful not to damage them. You should always look things over to see where the weak points are before you try to use them. We have not had as many bridges break during this project as we had this term.
  • Several students suggested something to the effect that it is important to pay attention so you don't repeat your mistakes. If you don't pay attention, you won't be able to figure out what you did wrong.