Paul was having difficulties keeping this circuit in calibration, and he wanted to know if he could replace it with the following circuit. He wanted to inject a 100 mA somewhere in order to balance the bridge. He was planning on using an OPA27 op-amp in a Howland current pump configuration to generate the 100 mA of current. In the original design, a computer changes the voltage output of the DAC until there is 0.8 Volts across the 8-ohm resistor, indicating a current flow of 100mA through the bridge circuit.

Click here for more detail.

I think it presents a good example of how one dissects the operation of op-amps and a DAC/ADC circuit. It is a great real-world application of the items we have been discussing recently. If you have any questions that you would like to pose, please feel free to send them to me. The explanation below refers to the circuits diagrammed in the link above. I suggest printing out the diagrams in larger detail so you can look at it as you read the explanation.

Answer: Dear Paul, First, to answer your basic question, no, I cannot see any easy way to inject current into the bridge or op-amp inputs in order to get the signal you want. Everything I could think of will dramatically change the properties of the bridge circuit. I will endeavor to explain my reasoning. You are trying to replace a dynamic resistance with a current source where the voltage may change. Look at it simply with ohm's law. V=IR R=V/I I=V/R The DAC controls R to adjust the bridge biasing. It changes the amount of current that flows only through the sensor side (22.5) of the bridge by lowering the impedance of the 2.5 ohm resistor (the VCRs are parallel to this resistor). Now when R changes, BOTH V and I change in that part of the bridge (ohm's law). The op-amp circuit then increases or decreases the current flow through that part of the bridge to make its inputs equal. Now if you could hold V constant at that point, a changing R value could be represented with a changing I value. I don't see how that is possible without seriously affecting the operation of the bridge circuit. Second, it is very important to understand that the current flow through the bridge is dependent solely on the output of the LM759. The fact that this is a power op-amp is why it can source the 100ma needed for correct operation. In your new design, you need something to increase the power output of your op-amp. The specs of the OPA27 indicate that it will not handle the current load either (25mA max). If my memory is correct, the Howland current pump will only control the current in the feed back loop of the circuit (ungrounded impedance only). That would not work well with the grounded setup you suggest. I will go through my thought process below to try to give you some insight into my understanding of your application. Click here to read a description of the gas flow measurement using the method that you describe. Take a look at the drawing on that page. There is also a link to where this information can be found: SENSYFLOW - Fundamentals and Measuring principle. The process is basically a pressure and temperature compensated hot film (or wire) anemometer. By the way, it is actually quite an interesting circuit. To understand what is happening with these sensors, let's do a signal analysis starting with the bridge. The purpose of this bridge configuration is to measure gas flow and compensate for temperature and pressure variances of the gas (the volume of the gas varies significantly due to temperature). Resistor Rs in the document above are akin to the 22.5 ohm in your bridge. They can sense gas flow as evidenced by the fact that when you blow on it, it will cool down, changing its resistance while doing so (that's what the gas does as it flows). When Rs changes, so does Is. The problem is that the ambient temperature of the gas will affect the flow rate; so add second resistor Rt. This resistor is significantly larger than Rs and passes less current. Since it passes less current, the resistance value is basically due to the ambient gas temperature. These two elements are combined into a bridge circuit to generate the data that is needed. The amplifier that drives the bridge circuit attempts to keep the voltage at the two points of the bridge the same. The only way this happens is by heating Rs. The wire or film heats up, thus changing the resistance till Rs equals Rt. The heating of the Rs resistor compensates for the temperature loss due to gas flow. So if you can measure the amount this resistor is heated, you have a measure of gas flow. (In your application, this feedback voltage from Rs is lowered by the resistors and the pot to basically 1/4 of the value sensed at Rt. But the principle is still the same - it just heats Rs to 1/4 Rt.) Now lets look at signal analysis. Gas flow cools Rs, and then increases Is. The current flows through Rs not 'is'. This changes the voltage across the bridge, then causes bridge power to increase, which in turn heats Rs until inputs to the amp are equal. The resulting change in current through Rs equates to gas flow. The computer controlled DAC in your circuit mainly senses the current changes in Rs. It does this by adjusting the current flow through Rs by changing the resistance of the lower portion of the bridge. By compensating for the change in total current flow, it cancels out any differences due to the change in Rs. The resulting number in the DAC indicates the current change caused by Rs changing due to gas flow. This brings us to some topics which concern me. In this design, the bridge amplifiers will be stable at a 0 volt output, which results in 0 volts across the bridge. This particular problem causes the workbench to have problems as it tries to simulate it. This may not be the case in the real world since Rs and Rt are always changing due to external influences. This change may cause the bridge to start working correctly, but it is not a good idea to rely on it. The DAC connection is hooked up in a way that appears to affect the calibration of the unit (note that the FETs are in parallel to the calibration section). I can't imagine anyone getting a very linear signal change indicating gas flow. I suspect that the computer has to be calibrated to a significant curve. You should isolate the calibration signal from the DAC, which I presume is what you really want. The fact that the DAC changes resistance in the lower part of the bridge and may add an additional curve to the output will also complicate the issue. I would consider a circuit topology change, using an ADC to directly sense the voltage across the 2.5-ohm resistor in the Rs leg of the bridge. This would eliminate complexity, but will require new calibration curves for the computer. Here is a circuit that you could experiment with. It is a single sided power supply design. In this design, the computer would calculate the change in current through Rs and compare that to the total change in current, which will indicate the gas flow. I'm not sure if you need the total current measurement, but it might be useful to indicate temperature variations. The computer would not control anything on the bridge. It is for measurement only. Note that I have not hooked up all of the ADC reference pins. You will need to determine those requirements if you want to use this design. Click here for more detail. Good luck.

Product Engineering Archive

Copyright © Chipcenter 1999