A quick survey by Robert Tinker of some of the important new handheld computer and sensor ideas and developments that might help improve teaching.
Becoming Available: Probing Outside | USB
Coming Soon: Smart Probes | Cheaper Motion | Do it yourself | Smart Software | Wireless
Over the last seven years, we have experimented with a number of configurations of portable computers and sensors as part of our Student Learning in Context (SLiC) research. To support classroom-based research, we have developed sensors, interfaces, and software for the Newton, e-Mate, and Palm computers. Portable Technologies: Science Learning in Context is a recently published book about this particular research.
We found that it is important to take outside the fewest possible numbers of parts and wires. Any extra interfaces or wires are all that more to set up on site and tend to get lost. A real-time graph is often essential, too, to allow students to see their dataset as it is collected. Ideally, the probe is part of the graphing display device. As a practical matter, however, it is probably cheaper to have one that simply plugs in and can be reused with other computers back in the lab.
A number of configurations are getting close to this ideal. The graphing calculators have temperature, light, pH, motion and other sensors that plug in directly. There are small interfaces that attach to the Palm, Visor, and Pocket PC computers and accept a wide range of sensors, one or two at a time. There are portable loggers that lack the real-time graphing but accept various sensors. We also sponsored a meeting to design the ideal hardware. See the result at http://www.cilt.org/images/DataGotchi.pdf.
[ back to top ]
Many people first became aware of the "USB" interface when they tried to connect their old keyboard, disk drive, or printer to their new computer. USB, which stands for "Universal Serial Bus" is rapidly replacing the old RS-232 serial interface for a number of good reasons. USB is much faster, it can be "daisy chained" from one device to another, and it carries power.
Most sensors connect to an interface that communicates to your computer over the RS-232 serial interface. We started that trend back in 1982 with a serial "blue box". That was an improvement over our "Red Box" that connected only to the Apple II game port. The Red Box was faster and more versatile, but vanished with the Apple II computer. No educator wants to invest in probeware that cannot be moved to new computers as they become available. When we decided to use the standard serial port, we knew that it was a terrible interface for sensors. It was, however, good enough for many applications, and certain to be available on all future computers. This has been true, even for odd computers like the Newton, most "palm-top" computers, and even the Pilot. Until USB, that is.
USB makes a far better computer interface for sensors. Because it is faster, we can imagine capturing and displaying sound and the forces in collisions, two important areas that are too fast for a serial port. Because it carries power, smart probes can be designed that draw off power to run their logic. These "smart probes" don't need batteries or external power supplies, making them ideal for applications requiring portability. Because it daisy-chains, many smart probes could be connected together and still not use up a single USB port on your computer.
So, the good news is that USB will support better, smarter, faster, more flexible, more numerous, easier-to-use sensors. The bad news is that you will have to purchase new USB sensors or interfaces. The first USB sensors are just coming on the market. If they get an enthusiastic reception, competitors should join in soon. Don't worry, some smart vendor will make a USB interface that uses all the old sensors you have purchased for your serial interface. The shift from RS-232 to USB will be painful, but USB is here to stay and it will make educational probeware much better.[ back to top ]
[ back to top ]
Complete microcomputers on a chip now cost under a dollar. This means that they can easily be incorporated into a sensor, making it a smart sensor. At very least, a smart sensor can have information about itself that it sends to any computer it attaches to. It can tell what kind of sensor it is, how accurate it is, and when it was last calibrated. It can also contain its calibration information and report its values using physical units like degrees, Newtons, or Joules. This would save the host computer from converting from the usual "raw data" inputs and would also greatly simplify classroom management. Usually, calibration data are stored on disk and are different for each probe. To avoid re-calibrating each lab, the sensors need to be matched to the computer containing the latest calibration data. Not so with a smart probe that carries its own calibration.
Smart probes with calibration, other information, and physical outputs may not be revolutionary, but they certainly would be useful for a very small additional cost. In fact, the cost may be negligible, since the microprocessor that provides the "smarts" can also replace other electronics needed in "dumb" sensors.
In 1997, as our contribution to the Center for Innovative Learning Technologies, we built a smart motion and motion sensors. These were built for the standard RS-232 serial port. As a result, these smart probes connect directly to most computers, including handhelds.
Our smart probes were not as simple as we liked, however, because they required some circuitry to be compatible with the serial port. In addition, they either required batteries or had additional circuitry to extract their power from the RS-232 port. Now we would recommend avoiding these extra circuits by building smart probes for the USB port. In effect, USB makes smarts in sensors not only feasible, but almost free. Let's hope some vendors see the advantage of smart probes soon.[ back to top ]
Our group, when it was at TERC in 1985, invented the ultrasonic motion probe. Dr. Jim Pengra from Whitman College, then on sabbatical with us, first connected and Apple II to the TI ultrasonic sensor and electronics used to focus the Polaroid Sun camera. This proved to be one of the most important educational sensors ever developed, since it makes graphs of position so immediate and clear. It also shows better than any other approach, the relation between position, velocity, and acceleration.
The ultrasonic detector is, however, far from ideal. It is pricey, a bit finicky, and a bit too magical. There doesn't seem to be any way to manufacture and sell it for less than $90, in spite of large volumes of sales. It has trouble bouncing sound off soft surfaces like sweaters, for instance. You cannot use a bunch in a single room at the same time because they interfere with each other. Finally, inexperienced users record jumps and spikes because they don't understand and cannot see where the beam is directed.
To address these problems, we hope to develop a simpler, more immediate distance detector. Our idea is based on the "fifth wheel" often used on cars. A light wheel that uses light to detect the motion of its spokes would be easier to understand and cheaper to build. Its own microcomputer could convert the light data into position, velocity, and acceleration data and pass it out quickly to a USB port. Around the time USB becomes widespread, I hope that this improved motion detector will be, too.[ back to top ]
We hope schools will want to make a major commitment to probeware and use it in
most laboratory investigations. Schools that do will find that they need a
range of sensors and that the cost of sensors ends up being a major problem.
Our solution is "do-it-yourself" sensors. We have made a prototype
interface that is unbelievably sensitive.
One side connects to a desktop or
handheld computer and the other will accept sensors you make. Because the
interface is so sensitive, your sensors can be relatively simple and easy to
make. Many sensors that usually need an amplifier can be connected directly. As
a result, many sensors involve simply connecting a few wires.
For instance, a thermocouple is a standard temperature sensor consisting of two wires of different metals. It generates a few tens of microvolts for each degree change at the junction between the two metals. It can be placed in a flame or used to measure very small temperature differences. Normally, though, you would need to amplify the thermocouple's voltage before feeding it into a computer. Not so with our interface. Just attach the two wires to the input and, vol‡, you have a probe. The wire costs only pennies, so you can afford to make many sensors.
Sensors for temperature, light, force, pressure, voltage, and much more can be made this way. The most complex might involve a resistor and the sensor. Our question is whether teachers and students will want to undertake probe construction as part of math and science lessons that involve sensors. We hope to make it as easy as possible by providing video construction tips and building smart software (see next section.)
Find out more about our Do-it-yourself CC Probekit.
[ back to top ]
Here is a typical problem: we build a beautiful probe software application that contains every possible option a student might need. It will even, for instance, make a least-squares straight line fit to a subset of the data. The beginning student, however, starts collecting data that is out of the range of the graph and sees only a straight line at the top of the screen. Totally confused, he or she thinks the computer has already made a straight line fit.
The problem is that we have built a tool that is too complex for the beginner but is unresponsive to the dilemmas beginners regularly encounter. Paul Horwitz had this same problem when he built a powerful genetics modeling package. His solution was to build Pedagogica, a scripting language that communicated with the genetics package. The result was BioLogica. Pedagogica can remove unneeded options, detect student errors, and strike up a conversation with the user.
We plan to use Pedagogica with a probeware package to address the shortcomings of previous packages. Now if the data are off-scale, the probeware will inform Pedagogica of the problem. Pedagogica might ignore the problem, or it might respond with a notice and a suggestion for correction. Or it might simply fix the problem by providing a different scale. The point is that this choice doesnŐt have to be built into the probe software, it can be in Pedagogica and easily changed depending on the instructional goals.[ back to top ]
Wireless portable networking is coming. Already, there are handheld computers with wireless, and cell phones that are portable computers. These two technologies will soon merge and offer high-speed wireless connectivity to the Internet.
This will be an important development for portable probeware for many reasons. First, imagine kids doing environmental measurements in study plot. Each group of students could instantly see all the data collected by all groups. Trends could be spotted in the field, hunches formulated, and further investigations undertaken.
Wireless networking will also bring better educational software. There is now great reluctance on the part of vendors to develop educational software for handhelds. The installed base is small and the hardware might change. But Internet access uses standards. Software developed for these standards will run on all handhelds as well as regular desktop and portable computers. This is a larger market and one that will not vanish when the hardware changes.
We are watching a more unusual use of wireless, too. There will soon be inexpensive very low power, wireless chips that can communicate with each other over short distances. Even if moved, they establish a network that can send data from any node to any other. The nodes might be wireless sensors and the data could be shared directly between sensors. The data might be used in the field or sent back to the classroom.[ back to top ]