|
Why not just use an inexpensive development system with a plug-on A/D board?
All Lawson Labs data acquisition products have three things that you might not know you need. Most competing systems are missing at least one. 1.) True differential high impedance analog inputs. The A/D converter itself will not have particularly high impedance inputs, and may have poor common mode rejection. As soon as you add a differential buffer in front of the A/D converter to meet these requirements, you compromise any internal calibration. Calibration signals really need to come in through the added differential buffer. We do that. It costs a little more, but it is essential for accurate, high impedance differential measurements. 2.) Protected inputs. Our products can survive hard treatment. 3.) Isolation. Isolation breaks ground loops and improves protection. What are Differential Inputs, and why do they matter? The notion that there is voltage potential called "ground" or "earth" with a potential of zero volts is necessary and useful. The term earth has its origins in telegraphy. In the early 1800s it was found that only one telegraph wire was needed; the return electrical path for the circuit could be through the earth. (Later in the century, it was found that the earth is not really of uniform potential, so telegraph wires were again employed in pairs.) For precision measurements, it is necessary to consider ground in relative terms, not absolute terms. Think of all voltages as relative. "Ground" points at different places in a system are apt to be at different potentials, even if they are connected together. If the connections have non-zero resistance and there are currents flowing in the grounding wires, voltage drops will result. To make matters worse, wiring has intrinsic inductance which goes up with its length, so even if the wiring resistance could be made to be zero, AC voltages would still develop across the wiring inductance whenever disturbances occurred. The AC effects get more pronounced as the frequency increases. So, to distill the above, the actual potential of a "ground" depends on exactly where and when it is measured. For precision voltage measurement, all voltages are voltage differences. For example, Point A is 100 uV more positive than Point B. Even if Points A and B are both nominally "Ground", the measurement must be made differentially between the two points in order to be meaningful. Think of the red and black leads of a battery-powered voltmeter. The reading given is the potential at the red lead minus the potential at the black lead. What is Common Mode Rejection, and why does it matter? In the case of a battery-powered voltmeter, the meter itself is not referenced to ground. Its internal power supply floats near the voltages being measured. To distinguish a "floating ground" from "earth", it is sometimes called "common", meaning "shared". An isolated data system such as the Model 201 can be used in the same fashion as a battery-powered voltmeter, but there are some secondary considerations that become important for high precision measurements. The computer that the data system connects to is normally referenced to earth. The voltage difference between "earth" and "common" appears across the isolation barrier. That same difference also appears between earth and the minus output of the power supply running the Model 201. Think back to the case of the battery-powered meter. The only current that flows in the red and black input wires is the input current for the voltmeter itself. There is no other path for current to take. In the case of the Model 201, there are power and computer serial port connections. Because the power supply and serial port are electrically isolated, any current will be very small, but even a tiny current can cause measurable errors given the high resolution of the system. Some currents may be AC currents only, but there are multiple ways for small AC effects to be rectified and to appear as small DC errors. The way to avoid such errors is to take full advantage of the true differential inputs of the data system by running a third wire for "common". That third wire connects to ground, or common, or "minus out" at the voltage source, and to a ground at the Model 201, or at another true-differential data system. Any currents flowing in that third wire will not cause voltage drops in the A/D signal input wiring. One way to tell that you have sub-optimal input wiring is if there is more than one connection at the A/D end of a minus input pin. What happens if the data system is allowed to float, like a battery-powered meter? Everything may appear proper, but the chances are good that if you watch the system carefully for an extended period of time, anomalous readings will occur, a few at a time, followed by long periods of stability. What is going on? There is almost always some leakage across an isolation barrier. The result is that the common voltage at the Model 201 compared to "earth" will slowly rise or fall, depending on the polarity of the leakage. If there is some other earth reference on the signal input side of the system, the Model 201 common will drift away from the input common in the absence of a DC connection between them. When the difference becomes more than 6 volts, the input range of the Model 201 is exceeded, and a small fault current will flow in a protection diode. That will bring the input voltage within range, so the normal behavior will resume, until the next time the process repeats. The common mode voltage is a voltage applied to both inputs of a differential amplifier. In theory, the amplifier should not respond at all to common mode voltages. In practice, there are limits. If you apply a sine wave to both amplifier inputs and gradually increase the amplitude, at some point, the amplifier will stop rejecting the common mode voltage due to range limitations. The range where common mode voltages are effectively rejected is called the Common Mode Range, or DC Common Mode Range. The ability of a differential amplifier to ignore common mode voltages is also limited by frequency. The higher the frequency of the common mode voltage, the harder it is to reject it due to timing mismatches between the positive and negative amplifier inputs. Common mode noise at low frequencies is effectively eliminated by Lawson Labs A/Ds. High frequency noise may need to be filtered before the signals enter the data system. What is input impedance, and why does it matter? Some interaction between the measuring system and the measured voltage is unavoidable. Raising the input impedance reduces any such interactions. Suppose you are measuring a battery's voltage through a 1 meg ohm resistor. If the voltmeter has a 1 meg ohm input impedance, the two resistors form a divider, and the measured voltage will equal half of the actual voltage. That may sound like an extreme case, but it can get worse. Some electrochemistry electrodes have output impedances in the 100s of meg ohms. The input impedance of the measuring system needs to be many orders of magnitude higher in order to get accurate results. What are the basic rules of thumb for Grounding? First, be aware of a widely believed fallacy: If point A is wired to point B and point A is wired to point C, that is equivalent to point A being wired to point B and point B being wired to point C. The problem is, that every connection includes a series resistance and a series inductance and some capacitance to nearby conductors. In the world of 24-bit A/D converters, these stray effects matter. When ground connections are daisy-chained, ground currents can add and subtract and interact with series resistance in various ways, causing complex, non-ideal behavior. Ideal grounding would have an individual ground wire from each grounded node to a central, single-point ground. The single point is sometimes referred to as a star ground, or mecca ground. Due to practical limitations, some compromises are almost always needed. One practice is to designate different classes of grounds. Each group forms its own star ground and the star grounds are, in turn, individually connected to mecca. For example, signal ground could be designated Ground A, with only tiny currents involved. Power ground could be designated Ground B, with significant currents possible. Chassis ground, with all associated shields connected, could be designated ground C. All three grounds would join at the earth connection point for minimum interaction. What is a ground loop? A ground loop is formed by any redundant ground connection. Two grounds are not better than one. If more than one ground point in a system is connected to earth, a ground loop will be formed. A typical example might be two pieces of equipment running from AC power each having a chassis grounded via a three-wire power cord, these pieces of equipment being connected together by a shielded data cable. If the cable shield is connected at both ends to chassis ground, a ground loop is formed through the building wiring. If the two earth grounds on the two pieces of equipment go to separate AC circuits in the building wiring, the two grounds could be at very different potentials, causing ground current to flow. The polarity and magnitude of the ground current would depend on the relative potentials of the two grounds, which would be influenced by other equipment turning on and off in the area. Such ground currents are non-deterministic and can cause erratic behavior in any of the equipment participating in the ground loop. Disconnecting the earth connection on one of the pieces of equipment might remove the ground loop, but could also create a shock hazard, especially in the event of a failure. Therefore, you should leave the earth connections intact and break any chassis-to-chassis connections, instead. It is also generally beneficial whenever possible, to run the cross connected equipment from a single outlet or power strip. True differential inputs are a powerful tool for breaking ground loops because a sense wire in a properly wired system will not complete a ground loop. In similar fashion, the optical isolation on Lawson Labs data systems helps insure that ground loops can be avoided. Isolation allows the freedom to do proper grounding, but it also makes it easier to leave circuitry floating that should be grounded. Generally, there should be one, and only one, connection to ground for each circuit block in a system. How do I diagnose and eliminate a ground loop? Symptoms of a ground loop can be varied. Problems with data stability or Digital I/O irregularities that come and go unpredictably can be caused by ground loops. If the loop is completed through building wiring, look for correlations with HVAC activity or refrigeration cycling or other activity involving large heaters or motors. Damp weather can contribute extra ground paths, or can improve the quality of an earth ground. If you suspect a ground loop, with the system turned off, try temporarily opening the intended ground connection and measuring the resistance to ground with an ohm meter. If you see a high impedance, say over 1000 ohms, you do not have a ground loop. If you see 10 ohms or less, you do have a second ground connection somewhere. Places to look include shielded connections, metal enclosures on conductive surfaces, grounded sensors, and conductive fluids. How do I diagnose and eliminate a missing ground? See the section above called "What happens if the data system is allowed to float?" for the symptoms of a missing ground. If you suspect a missing ground, with the power off, measure resistance with an ohm meter to earth from various points in your system that are nominally at ground potential. If you find a missing or bad connection, fix it. Ground connections that have seemed satisfactory for years can fail suddenly if excessive ground currents have caused electrolytic action in dissimilar metal junctions. What is meant by loop area? Early AC house wiring was called knob and tube. Porcelain tubes held a wire when it passed through a wooden structural member and knobs held wires when they passed over framing members. The two sides of the AC circuit could be far removed from each other. The further apart the wires, the larger the magnetic field they create. Modern wiring uses paired wires to minimize the effect. The area inside a conductive loop is called loop area. A loop will generate a magnetic field if a current flows in it, and the magnetic field grows larger with the loop area. A loop in the presence of a magnetic field will generate current, and the larger the loop, the more the current. Minimizing loop area will reduce both emissions and susceptibility to magnetic interference. Twisted pair wiring is one way to minimize loop area. Coaxial cable is another. Even when using point to point wiring, make an effort to keep the loop area of analog signal wiring small. Also, review your power wiring to minimize the loop area of higher current circuits, especially for devices that turn off and on. In the presence of strong fields, take extra care to keep input wiring as short as possible, and strive for symmetry between the plus and minus input terminal wiring. |