Tech Tips December 2006
December 26, 2006
TECH TIP OF THE WEEK:
More servomotor inertial matching.
The following is a reader response to a Technology Update called 'Matching moments of inertia' that appears in the December 1, 2006, issue of Control Engineering magazine. Click here to read that.
As 'Matching moments of inertia' points out, for best performance the moments of inertia of the load and servomotor armature should have a 1:1 match. The article went on to point out that performance is still adequate with a 3:1 mismatch.
There is more to 'match' thinking, however. Most servo motors for general industrial applications use ceramic magnets, which still get satisfactory performance with a 3:1 mismatch. There is good 'load isolation' with this ratio, since the load inertia reflected to the motor is reduced by the square of the gear ratio. This fact is very important when machine axes reach 20,000 lb or more.
There are also high performance servo motors called 'low inertia motors' that use high coercive force magnets like neodymium-iron-boron. The best match for these motors is still 1:1, but now the reflected inertia must be much lower to get increased performance.
Unfortunately, many users think they can still get the desired performance with these motors on machine axes that worked well at higher reflected inertia with a ceramic-magnet motor.
Manufacturers of low inertia motors claim a 10:1 inertia mismatch is viable, but my experience shows that it can lead to instability if you don't know what you are doing. As the Control Engineering article points out, to avoid instability the acceleration time must be increased.
Another problem that is often overlooked with low inertia motors is that they have a much shorter thermal time constant, which can cause overheating problems in a drive providing the power needed to force-repeated accelerations.
SOURCE: George Younkin, P.E., MSEE is a Life FELLOW of the IEEE. He is author of the book Industrial Servo Control Systems: Fundamentals and Applications, Second Edition, published by CRC Press. For more information or to order a copy of the book, visit CRC Press online at www.crcpress.com. Younkin is currently working with Industrial Controls Consulting, A Division of Bull's Eye Research, in Fond du Lac, WI. For more information, contact him by telephone at 920-929-6544 or by email at firstname.lastname@example.org
December 19, 2006
TECH TIP OF THE WEEK:
Use four-wire circuit for accurate measurements.
The large motor drive current (black path) produces a voltage drop across a series resistor, while a miniscule current flows in the measurement path (green).
Measuring the voltage drop across a load is surprisingly hard to do accurately. The problem is that contact and lead resistances contribute error voltages that depend on the load current. To get around this problem metrologists—who spend their careers pushing the limits of measurement accuracy—have developed the Kelvin, or 'four-wire' measurement technique.
Suppose you want to monitor the current through a dc motor by placing a small resistor in series with it. Measuring the voltage drop across the shunt resistor and a little math using Ohm's law yields the current through the resistor and, by implication, the current going through the motor.
Let's say it's a 24 V motor drawing 10 A. A pretty typical resistor made for such an application would have a resistance of 0.1 W and give a 1 V drop with 10 A flowing through it.
The problem is that we can expect lead and contact resistances to be high enough to confuse the measurement. Five feet of #18 copper wire in the motor power circuit—reasonable for this application—has a resistance of 0.03 W. That's enough to cause a 30% error in the current measurement. Contact resistances at terminal blocks can be even higher, causing more trouble.
Stack terminal lugs so that the drive current does not flow through the voltage-measurement connection.
The four-wire measurement technique, shown schematically here, avoids these problems. The idea is to connect sensing wires as closely as possible to the series resistor's ends. The motor-current path flows directly through the resistor, creating the voltage drop we wish to measure, but not through the sensing wires.
Since voltage-measuring equipment tends to have very high impedance (tens of megohms or more), only a miniscule current flows through the sensing leads. They can be quite light gauge without introducing measureable errors.
The graphic, 'Right and wrong ways to make four-wire sense connection,' QA19DEC06a2.bmp shows how to stack the terminal lugs on a post. The sense leads should always be stacked so that the supply current does not have to flow through the sense lead connection. This minimizes the effect of lug-to-lug contact resistance.
While I've used a dc motor drive as an example, the same technique can be used to measure voltage across any circuit element. It works for ac as well as dc. It works when the impedances are reactive (inductive or capacitive) as well as resistive.
— C.G. Masi , Control Engineering Senior Editor
December 12, 2006
TECH TIP OF THE WEEK:
Consider liquid cooling for high-power servosystem components.
Control system engineers, who nowadays tend to be electronics engineers, tend to ignore component-cooling issues. As long as they specify components whose power ratings exceed the expected in-service conditions, they generally don't consider the amount of heat their equipment generates. At best, they might estimate the total system power dissipation and report it to the facility HVAC engineers to help them size their air-conditioning units.
At one level, this is a perfectly reasonable thing to do. Nearly all electrical and electromechanical components are air cooled. Part of the control-component development process involves modeling and testing to determine the operating conditions under which the unit won't turn itself into a component breakfast bagel. So, as long as you follow the manufacturer's recommendations for mounting and sizing (Don't take a 1/4 hp electric motor, mount it in an air-tight insulated box, and ask it to deal with a 3 hp continuous load!), you can expect it to operate at a reasonable temperature conducive to long life.
At another level, though, it may be possible to save money and space by looking into liquid-cooled components.
Liquid cooling is better than air-cooling.
One reason the P-51 Mustang, generally considered to be the best propeller-driven military fighter aircraft ever built, was so successful was that its 12-cylinder Merlin engine was liquid cooled. Most earlier designs opted for air-cooled radial engines because they were much simpler and lighter than comparable liquid-cooled units. Experimental aircraft built for the Schneider Cup racing series had, however, shown the valuable ancilliary benefits of liquid-cooling.
Better thermal contact —Heat transfer between a solid surface and the cooling fluid is better with liquid cooling. Liquid coolant, which is generally a mixture of ethylene glycol and water, has a higher density and greater viscosity than air. Together these properties reduce the surface area needed to maintain a given heat flow at a given temperature difference.
Better heat capacity —Liquid coolant has a much (much!) higher heat capacity than air. Heat capacity is defined as the ratio of the heat energy stored divided by coolant volume, divided by coolant temperature rise. Greater heat capacity means you can keep a component cooler with less coolant.
Separation between the heat source and heat sink —Ultimately, you have to somehow dump all the heat you generate into the environment. The point where you dump the heat is called the 'heat sink.' (Think of it as a grounding rod for heat.) Those big finned aluminum things simply couple heat from your heat source into the atmosphere. Liquid cooling separates the points where you take the heat out of your control component and where you sink it to the environment. That creates opportunities for more effective heat-sink systems, such as chilled-water, that actually can bring your component's operating temperature below ambient !
The P-51 Mustang's designers used these attributes to create a lighter, slimmer aircraft with much lower drag. They even put the radiator in an enclosed duct, which they contoured to minimize its drag.
It is possible to achieve similar benefits through liquid cooling of motion-control components. 'Heat is generated in the bearings and stator windings of motors,' according to Dr. Gunther Vogt, managing director of AMK Arnold Muller GmbH. 'By concentrating cooling flow at those points, we can get most of the heat out of the motor before it gets to the housing.'
AMK is one servo-drive manufacturer that made a conscious effort to develop liquid-cooled motion-control components for its product line. 'In Europe, we have seen a steady growth in sales of liquid-cooled motors and drives, but not so much in America.'
One concern Dr. Vogt reports common among control engineers is fear of putting water into control cabinets with relatively high-voltage electricity. AMK and many others who have built liquid-cooled electric power systems solve this problem by mounting their power transistors and other heat-generating components on an aluminum plate at the back of the cabinet. Cooling channels drilled into that plate carry liquid coolant in a serpentine pattern, cooling all parts of the plate. Thermal conduction then carries heat from the components to the plate just as if it were a conventional heat sink. Mounting this structure at the back makes it easy to connect chilled coolant supply and drain lines outside the cabinet.
Since the motor housings and drive enclosures no longer have to sink heat to the environment by air convection, they can be made smaller and be put into tighter spaces. Since cooling is more efficient, physically smaller motors can do the same jobs as their larger air-cooled brothers. Similar advantages accrue to liquid-cooled drives (inverters). Finally, since it is possible to use chillers or heat pumps to sink heat to the atmosphere, it is possible to run motion control systems under otherwise impossible environmental conditions.
SOURCE: Control Engineering private communication with Dr.-Ing. Gunther Vogt at SPS/IPC/Drives Electric Automation Systems and Components, Nuremburg, Germany.
Have a really good idea to share with other engineers? Email your control-system tip to email@example.com with 'Tips' in the subject line and question, and full contact information in the body of the email. If we use it, we'll send you an 'Engineer and proud of it!' pocket protector.
December 5, 2006
TECH TIP OF THE WEEK:
Use current loops for reliable analog signaling
The workhorse method for sending sensor output signals is analog voltage, where the voltage between the 'hot' and 'neutral' signal lines is a function (usually linear) of the value of the sensed real-world parameter. For example, the output from a thermocouple equals the difference between the temperatures of the 'hot' and 'cold' junctions times a thermocouple constant (expressed in volts per degree). Analog voltage is useful because it is simple to generate and robust under most conditions.
4-20 mA current loops provide reliable analog signaling.
That is, analog voltage is robust as long as the series resistances within the signal lines are low compared to the load impedance on the receiving end. There are times, however, when you can't guarantee that the analog voltage will reach the receiver unmodified. When signal lines get long, or there are multiple connectors in the signal path, or when more than a few microamperes flow in the signal wires, unacceptable error voltages can appear without your even knowing it.
'Who ya gonna call?' Call 4-20 mA current loops!
The advantage that 4-20 mA current loops have is that (assuming no short circuits) every electron pushed out of the hot lead at the transmitter end must go into the corresponding terminal at the receiver. In addition, you can have multiple receivers because every electron pushed out of the transmitter hot terminal must make the complete circuit to flow into the cold terminal.
This figure shows how a 4-20 mA current loop works. Instead of having a low transmitter impedance and a high load impedance, it has a high transmitter output impedance and low load impedance.
For the thermocouple example, the analog output voltage goes into a voltage-to-current converter-which is simply a current regulated power supply whose current set point depends on the input voltage. Zero voltage input translates into 4 mA output. The full-scale current of 20 mA corresponds to the maximum sensor-output voltage.
Suppose you have a sensor set up 10 m outside your building. Signal lines have to go from the sensor to a connector plugged into a wall feed through. A second connector plugs into the feedthrough on the inside. There'll be, maybe, another 2-3 m of cable carrying the signal to another connector at the receiver end. Each connector, of course, introduces two ohmic contacts (one for the hot side and another for the neutral) into the current loop.
Now, let the wind blow, the ambient temperature change, and a klutz trip over the signal line inside. Each time something happens, it changes the resistance of at least one of those ohmic contacts, changing the signal circuit's total resistance.
BUT, WHO CARES!?
It's not a problem because the current through the loop is regulated. Each time the loop resistance changes, the transmitter's current regulator adjusts its output voltage to compensate, keeping the current locked to its appropriate value. Your data is safe. Your measurements are secure. You can sleep soundly at night, confident that (barring a catastrophe that would completely knock out the data gathering system) the results you get will be correct within your predictable measurement error.
For more information about 4-20 mA current loops, visit the following online sources:
— C.G. Masi , Control Engineering Senior Editor