From the Lab February 1997 [Archive]

skranish

Feb 01, 1997, 01:00 AM

Beginner's Corner: Accuracy of Measurements and Calculations


If you have been following much of the traffic on the eflight listserv, it is
impossible to miss all the messages that quote performance calculations from various
computer programs. 'Electricalc' seems to be the tool of choice at the moment, but there
are others, such as Aero*Comp and the Electric Airplane Design Analysis Program. These
programs seem like the ideal way to tweak a design or power train - adjust a few numbers,
recalculate, and check the results. It really is that simple, isn't it? Well, not really. 

First of all, let me make one thing perfectly clear: I am not out to trash Electricalc
or any of the other programs. My intent here is to explain why these programs cannot be
perfect, and why the answers they provide may not always match actual flight testing or
even be what they seem. 

Approximations in Engineering and Science 

This may come as a surprise or shock to some of you, but engineering is frequently not
an exact science. Actually, science is not always an exact science, either. In
engineering, we frequently make use of "engineering approximations" - numbers or
equations that are arrived at by a variety of methods, and serve the purpose of being
'close enough' for the task at hand. In some sciences - Astronomy is a good
example - the approximations are even coarser. I took a technically-oriented Astronomy
course in college (not the one taught by the late Carl Sagan, unfortunately) where test
answers were considered correct if they were within a factor of 10 of the 'correct'
answer. In reality, a lot of Astronomy calculations are even less accurate that that!


You may not realize it, but you probably use approximations, too. When you read an
analog clock (you know, the old fashioned kind with the round dial and the hands) you
probably interpret the time as "about a quarter to one" rather than the more
exact 12:47. Most of the time, the difference of about 2 minutes is not important, so the
approximation works just fine. 

Why do we use approximations? Well, simply...to keep things simple! A cumbersome,
complex calculation may be simplified by removing some elements that do not
significantly affect the result, or replacing others with a constant value that
is 'close enough'. An element of an equation that changes the result by a fraction of
a percent is really not important if the result is only accurate to 10%. If done properly,
these simplifications introduce very little error - provided you understand the
limitations of the simplification. 

Some engineering approximations are really quite simple. A classic one is "The
square root of 5 is equal to 2, for small values of 5, and large values of 2". This
may sound odd, but if you get out your calculator, you will find that the square root of
5.0 is actually 2.24, which means that this approximation is off by about 11%. Conversely,
2.0 is the square root of 4.0, which is off from our approximate value of 5.0 by 20%. In
many calculations, these approximate values may be 'close enough'. 

Another, more subtle and perhaps more sophisticated approximation may be found in the
use of trigonometry: "The sin of an angle (in radians) is equal to the angle for
small angles". This means that the calculation sin(X) may be replaced by X,
 provided X is a small value. This sort of approximation may find its way into
calculations for full-size (well, pilot-occupied) airplanes, because (for most airplanes)
both the angle of flight (climb or descent) and the angle of attack never get very large.
But if X is large, the approximation falls apart, and a large error is introduced. Well,
our models typically have a flight envelope - and the resultant flight angles
- that would make even the most macho test pilot lose their lunch, so this
approximation is probably not valid for a calculation used for a model. There are almost
always limits to the usefulness of a particular approximation. 

Accuracy of Numbers 

With the possible exception of phone numbers, numbers are never exact. Even your bank
balance isn't all that exact - and banks can make a lot of money from daily inaccuracies
of about half a penny for each account. Let's try a simple example so see what happens
when the numbers used in a calculation are not known exactly. Consider the equation A *
B = C, where A and B are both accurate to +-10%, which is actually an
uncertainty range of  20%. How accurate will the result C be? Care to guess?
(Hint, hint!) 

An uncertainty range is the range of values that a number can have. We know the number
is somewhere in the range, but we don't really know where. A number that is known to +-10%
may be anywhere from 90% to 110% of its 'nominal' value. 

OK, to make the example simple, A = B = 10. If we know the numbers exactly, A *
B = 10 * 10 = 100. 

The worst inaccuracy for this equation will be for the situations where both A and
B are low by the maximum uncertainty, and where A and B are high by the maximum
uncertainty. 

Case number one, 10% low: A = B = (90% of 10) = 9, so (A * B) = (9 * 9) = 80. See
something funny here? Isn't (9 * 9) = 81? Well, yes and no. If A and B are known
exactly, then 9 * 9 = 81. But we started with the understanding that we only know A and B
to +-10% accuracy, or one part in 10. The result cannot be more accurate than the
numbers we start with  - we cannot increase the certainty of numbers, only the
uncertainty. The '1' in '81' is 1 part in 81, which is far more accurate than the numbers
we started with. So we say that (9 * 9) = approximately 80. This case is 20%
below the exact example. 

Case number two, 10% high: A = B = (110% of 10) = 11, so (A * B) = (11 * 11) = approximately
120. This is 20% above the exact example. 

So, having started with two numbers that are known to +- 10%, this very simple equation
produces a result that is +-20%, for a total uncertainty range of 40%! 

Before someone jumps up and screams that the problem here is that the numbers are being
multiplied, let's try a similar example using division: 

Consider the equation A / B = C, where A and B are both accurate
to +-10%. 

To make the example simple, A = B = 10. If we know the numbers exactly, A / B =
10 / 10 = 1. 

The worst inaccuracy will be for the situations where one number is high and the
other is low, such as A = (10 - 10%) = 9 and B = (10 + 10%) = 11 

Case number one, A = 9 and B = 11, so (A / B) = (9 / 11) = approximately
0.8. This is 20% below the exact example. 

Case number two, A = 11 and B = 9, so (A / B) = (11 / 9) = approximately
1.2. This is 20% above the exact example. 

So again, having started with two numbers that are known to +-10%, a very simple
equation produces a result that is +-20%, for a total uncertainty range of 40% of the
'nominal' value! 

If the equations are more complicated, which is usually the case, the resultant
uncertainty gets even larger. A large or complex equation may include many elements of
varying uncertainty, but a single important element of great uncertainty may 'taint' the
resultant calculation. 

The only way to avoid this expansion of uncertainty is to limit the equation to
addition and subtraction, which propagate, but do not increase uncertainty. There aren't a
lot of useful equations that are limited to addition and subtraction, so we are forced to
live with uncertainty. 

Significant Digits 

There are a lot of ways to write a number. Consider:

<ul>
<li>23 </li>
<li>23.0 </li>
<li>23.04 </li>
<li>23.040 </li>
<li>23.0403 </li>
<li>2.30 * 101  or 2.30E+1  (this is scientific notation, if you are
not familiar with it) </li>
</ul>

Each of the above lines represents approximately the same number. The difference
between them is the number of digits presented, and the number of digits implies or
gives the impression of accuracy. 23.501 looks like a more accurate number than 23 because
it includes digits that represent very small units - all the way out to thousandths. But
is it really more accurate? Or does it simply include more digits, just for show? 

The digits shown in a number are called significant digits  - that is they
signify a quantity, rather than simply filling a place. A 0 on the right end of a
number may be a placekeeper, or it may be a significant digit. All non-zero digits
are significant - or should be. 

Once upon a time, the best tool most of us had for doing calculations was a slide
rule. A good slide rule could provide 3 significant digits or usable numbers for
some calculations. Most measurements were made with instruments that used moving coil
meters, the best of which can be read to about 3 significant digits. Most of us did not
have access to tools that could provide more accuracy, and frankly, for most things it did
not matter. 

Then along came computers that could do calculations with 8, 10, or even 12 digits. Not
digits of accuracy, mind you, just digits. There are good reasons for using a lot of
digits in a calculation, but early computer software was very bad at formatting output, so
even if the input numbers were only accurate to, say, 10%, the output would be printed
to 8 or 10 digits, most of which were rubbish. The arrival of affordable calculators
actually made the situation worse - all of a sudden everyone had access to a tool
that could do calculations to 8 digits or more. More sophisticated calculators allow the
user to set the number of digits displayed, but it is still up to the user to use
good sense in how many digits are actually used when entering a number. 

The number of significant digits used in a number implies the accuracy of the
number. If a number is accurate to only 10%, then only one significant digit should be
used, regardless of how large or small the number is. A single count (changing the value
by one) of that single digit represents at least 10% of the value, so no more digits
can be used without implying greater precision or certainty than is really available. On
the other hand, a number that is accurate to 1% may be represented by 2 significant
digits, because a single count (changing the value by one) of the second digit represents
at least 1% of the value. For our purposes, two significant digits is probably the maximum
that is reasonable. You might measure dimensional data for your model to 3 significant
digits, and perhaps measure its weight to 3 significant digits, but everything else should
be limited to two - or maybe just one. 

The use of significant digits does not really follow any exact rules. It is more
a matter of what is appropriate, and is often an indicator of how realistic
the numbers are. If you look closely at an outdoor mercury thermometer, you will
probably find that it is really readable to 2 or 2.5 degrees F. Using this
as your measurement instrument, it is OK to say the temperature is 68F, but rather silly
to say that it is 68.5F, and downright stupid to say it is 68.52F. The last digit
displayed may have some uncertainty - in this case the last digit may represent about 1%
of the value, even though we can only measure to 2 %or 3% of the value. As a result, we
may say the temperature is 68F, but it may really be 67F or 69F - we really are not that
sure, and cannot measure it more accurately with this instrument. 

Another example that is perhaps a little closer to our applications is measuring the
rotational speed of a motor. This is usually given as RPM, or revolutions per minute. Even
though the given unit is a minute, most inexpensive tachometers only measure for one
second, because they are really made for the slimer market, and people want to use them to
adjust the engine mixture - so they want a reasonably fast update of the display. If a
motor is rotating at 12000 RPM, it can also be considered to be rotating at 200
revolutions per second, or 200 RPS. The measurement time is one second, so the measurement
is accurate - at best - to one part in 200, or 0.5%. A number of this accuracy can be
displayed with 3 significant digits, but no more. And when the number is translated
from RPS to RPM,  by multiplying it by 60, the last digit representing 1 RPS now
represents 60 RPM, so you should be very skeptical of a number that shows more accuracy
than 60 RPM. A measurement of 12000, or 12600 RPM makes sense, but 12650 or 12653 are just
made up numbers, unless the measurement was made with much more sophisticated equipment -
and the additional significant digits implies just that. 

Accuracy of Mathematical Models 

All of those amazing performance calculations made by the above mentioned programs come
from some sort of programmed mathematical equation. The equation is a mathematical model,
or representation of a physical system, such as the drive train or airframe. Mathematical
models are a combination of:

<ul>
<li>physical constants, which are really hard to change (care to change the
gravitational constant?) 
</li>
<li>measurements of the physical system being modeled  (how accurately
do you know your propeller pitch?) 
</li>
<li>approximations of values or characteristics than cannot easily be
measured (such as coefficients of lift and drag) 
</li>
<li>(usually) some educated guesses (you're sure you can built an 82" Bearcat
that weighs only 10 pounds?) </li>
<li>(occasionally) some wild guesses for good measure (of course, you would really like the
Bearcat to weigh 5 pounds) </li>
<li>and lastly, an equation to tie all the pieces together and model the system - sort of </li>
</ul>

Even if all of the numbers are known exactly, the equation that models the system will
probably include some approximations and simplifications. The result, is, well, an approximate
representation of the physical system. 

Do you remember your high school physics class? Remember the simple equation that
describes the effect of a constant force on an object? F=ma, where F is
force, m is mass (well, weight for our purposes, but there is a difference),
and a is the resultant acceleration of the object. It is completely accurate, isn't
it? Well, not really. This oversimplification of a physical system ignores (or at
least does not explicitly account for) surface contact friction and air friction (we call
that drag!), both of which happily change with the speed of the object and the
environment the it is moving in. And you did remember the force applied by gravity,
didn't you? Even though it does not explicitly appear in the equation, gravity must also
be considered as part of the force, because it really applies a force F in whatever
direction it feels like (ever notice that on Star Trek, that is ALWAYS towards the bottom
of the screen, even though there is no reason for it?). To make things even more
complicated, the gravitational field changes predictably with altitude, and less
predictably with location. (In some places on our little planet, gravity really is
stronger than in other places.) Even though it is a simplification, that little equation F=ma
does serve the purpose of illustrating an important physical relationship, without
cluttering itself up with a lot of details. 

Why aren't the equations that model the system more accurate? There are a lot of
reasons, some historical, some physical, and some practical. Let's consider history: once
upon a time, most engineers had only three calculating tools available: a slide rule, a
book of pre-calculated  standard tables, and a pencil. If you remember using these
stone-age tools, complex calculations were very, very tedious and time consuming. Any reasonable
simplification of the calculation saved time and effort. How does this affect us with our
whiz-bang high speed computers? Well, we need to remember where the equations used by the
computer really come from. The people who write programs for model airplane performance
probably aren't doing all the required analysis and investigation on their own.
Engineering is, if nothing else, the art of building on prior art, so like most engineers,
they are using, at least as a starting point, available aerodynamics texts and reference
books. This is both good and bad. It is good, because a lot of very good aircraft
performance and control work was done a long time ago, and brought us the remarkable
aircraft that flew in World War II. and many of the light aircraft that fly today. It is
bad simply because this work was done a long time ago, when it was difficult to do
complex calculations, so the equations were simplified if at all possible. Those
simplifications from long ago have carried forth to the present because not that much new
work is being done on low speed flight, other than airfoil design, and in the grand scheme
of things, our models fly pretty slow. So in the end we are dependent on some very old
(but probably very good) work from very old books. Actually, even newer books frequently
use older books as reference material. 

Many older texts - and some newer ones, too - use nomographs to show the results of
some calculations. A nomograph is a type of graph, and the lines on it represent
approximate values from some equation, which may actually be very complex. Multiple
equations may be used to generate a single line, because of some characteristic that
changes relative to another characteristic. In general, nomographs were used to replace
calculations that were very complex. Unfortunately, they don't translate very well onto
computers, which could have happily done the complex calculations, except that they were
not available at the time. If older texts present information as a nomograph, it may be
translated for a computer program as an approximation. 

I got out my college aerodynamics texts from 20 (gasp!) years ago. Airplane
Aerodynamics was published in 1967, but it is actually the fourth edition of a book
first published in 1951. Stability and Control of Airplanes and Helicopters was
published in 1964. I also took a look at some my more recent purchases: Theory of
Flight (available at Barnes & Noble for $15.95) was first published in 1945.
Martin Simons' book Model Aircraft Aerodynamics is actually the third edition of a
book first published in 1978, but some of his references are much older. Alasdair
Sutherland's book Basic Aeronautics for Modelers lists a number of references,
including Simons' book, but does not give any dates. Chevalier's book on book Model
Airplane Design and Performance for the Modeler and Andy Lennon's new book R/C
Model Aircraft Design are both almost devoid of references, so it is hard to tell
where any of their material comes from, but I doubt it is of recent vintage. Astro Bob
Boucher's book on electric motors has a page of references, but almost a third of those
reference.. Astro Bob! The others are of varying vintage, and there are no notes in the
text to indicate where material from a reference source was used. 

There is nothing wrong with using material from older books - often older books have
more details on basic material that is now considered too simple to cover in much detail,
because more complicated things are available to fill the pages. You just need to be
aware of the simplifications and approximations that may have been made at that time due
to limitations of the available computational tools. 

Let's consider physical reasons for simplification of equations. Many physical
quantities are difficult or expensive to measure, so it may simply be more practical to
use an approximation. If an equation includes a number of physical measurement values that
are not readily available, it may be preferable to simplify the equation by using
approximate values, instead. The coefficients of lift and drag are good examples of this,
because it is not practical to measure these value unless you happen have a wind tunnel in
your basement. The best you may be able to do is look up the coefficients for your
particular airfoil. You do know what airfoil you are using, don't you? 

The practical reasons for simplification of equations have been pretty well covered
above in the section on Approximations in Engineering and Science. 

Want a more complete, accurate, and up-to-date model? Get a degree in Aeronautical
Engineering, and plan on spending a LOT of money on a wind tunnel and a computer, and
waiting a long time for the results. The more accurate the model, the more complex, and
the more time consuming to calculate. There is really no way around it. Most of the
really, really expensive high speed computers now in use are used for things like modeling
airflow around airframes, and the behaviour of weather systems. The airflow modeling has
become very accurate, and is now replacing the use of wind tunnels, but we know how
accurately weather can be modeled - and predicted - don't we? 

What This Means To You 

What does all of this mean to you? Well, for starters, you should very skeptical of the
numbers generated by any computer program. If you have access to the source code, and know
what sort of mathematical model the calculations are based on, you might have a bit more
confidence in the calculations. If you really want to look knowledgeable, appear totally
aghast at any calculated result that has more than 2 significant digits. Can you tell
the difference between a climb rate of 1682 ft/min and 1725 ft/min? Do you really
think the numbers you enter into the program are accurate enough to give results of this
resolution? 

While we are on the subject of rate-of-climb, let's consider a simple physical
constraint on climb rate. In theory, the rate of climb is unaffected by wind speed,
because if you are trying for maximum climb, you just keep the plane pointed into the
wind, right? Well, if there is NO wind, the plane proceeds upwind at its flying speed, and
very quickly gets very hard to see - so you have no choice but to turn downwind, and the
turn affects the efficiency of the climb. On the other hand, if there is a stiff
breeze, the plane may climb to a high altitude while covering almost no ground, and
without requiring any turns. Does your program account for this? 

You should always be skeptical of any program that uses unknown, undocumented, or
unspecified equations. If you have access to the equations, you should be suspicious of
any constants in the equations, because constants (other than the common physical
constants, like gravitational acceleration) are almost always a sign of simplifications.
And you should be skeptical of any program that uses a constant for something you know is
not a constant at all, or uses more significant digits that can reasonably be used for a
number. How accurately do you know the motor constants? Or the pitch of your propeller? 

For instance, battery pack voltage and impedance are never constant. They change - and
change a lot - with state of charge, current draw, temperature, age of the cell, and
probably phase of the moon. If you use only a constant value, you are using an
approximation, and perhaps an unnecessary one. Such an approximation may be propagate
through a variety of calculations, including power in, power out, run time, and motor RPM.
To avoid using this approximation, we have to better understand and model the
characteristics of a battery - and that is a topic for another time. 

This does not mean that you should not bother to use a computer program to help you
design an airplane. On the contrary, they can be very useful, if used wisely:

<ul>
<li>Try to understand what the calculation is, and what data is important to the result. If
at all possible, get the actual equations used, and try to understand them. </li>
<li>If you cannot get the actual equations, be skeptical of the results. This is not magic! </li>
<li>When entering numbers, use an appropriate number of significant digits </li>
<li>If the program produces more significant digits than is really appropriate, round off
the numbers to get an appropriate number of digits. </li>
<li>When measuring physical data, such as dimensions and weight, be as accurate as possible.
</li>
<li>If at all possible, don't guess about numbers. There is a lot of information
available on the lift and drag of specific airfoils. If motor constants are not available,
they can be measured. Thrust produced can be measured, if you are willing to go to the
trouble. </li>
<li>If the program uses a constant value for a number that you know is not constant, try to
use an average value. For example, use 1.05V per cell as the average voltage from a
battery, because it will actually range from 1.20V down to about 0.90V - or lower. </li>
<li>Most important of all: Use the computer programs as comparison tools, rather than trying
to generate exact numbers. Change only one item, such as the pitch of a prop, or the
capacity (and hence weight) of a battery, and compare the results. Do not expect to do
accurate comparisons if you change an entire airframe or drive train. </li>
<li>When comparing the results from the program, keep in mind the accuracy of the input
numbers and the calculation. If the difference between two results is small,
consider them to be the same. </li>
</ul>

Additional Reading 

Most college Physics texts, especially those that include laboratory material, will
have a section of the accuracy of measurements and calculations. I used the classic
Halliday and Resnick. 

Next Time 

Is there something that you don't understand that you would like to see explained?
Please email me at <a href="#skranish(at)ezonemag.com">skranish(at)ezonemag.com</a>


<hr>

Fuses: The Inside Story 

A fuse is essentially a piece of wire of known poor quality. Unlike a piece of wire of
good quality that can carry more current (the quantity of ping pong balls per second,
remember?) than it was designed for, and only gets warm, a fuse gets even: when it tries
to carry too much current it self destructs. This can actually be a good thing, because a
fuse can be a very important safety device. But it is very important to understand that a
fuse is a thermal device, that is, it really self destructs because of heat, rather
than because of current. 

A fuse, like any piece of wire, has a resistance. The resistance will typically be very
small, and with common instruments like a multimeter, it will appear as a short circuit -
essentially zero ohms. But there is a resistance there nonetheless, and the
resistance is important, because that is what makes the fuse dissipate power and - you
guessed it - get warm. If you remember from the discussion of electrons and
resistance in the December 96 column: 

".. they do what electrons always do when they get confused about what they are
supposed to do: they turn the energy they were supposed to do something useful with
into heat, which is largely useless, unless that is what you were after in the first
place." 

Well, heat, which is a form of power, is actually what we are after here, so there can
actually be a benefit to resistance. We discussed power in the last column, but here we
need to consider a different equation for power: Power = Amps2 * Ohms (that is
Amps squared times Ohms, in case it doesn't come out right in your browser). Even with a
very small resistance (very few ohms), if there is a lot of current flowing, there will be
a lot of power dissipated in the fuse, and if there is too much power, it will get very
hot and go POOF! 

An important thing to remember is that temperatures do not rise instantaneously, so
even if the fuse is dissipating a lot of power, it takes a while before it gets so warm
that it self destructs. This is both good and bad. A fuse will not blow fast enough to
protect a MOSFET or other semiconductor device - the semiconductor will almost always self
destruct first. On the other hand, it is possible to overstress a fuse by letting it pass
far more than its rated current, provided we do not do it for very long. 

Let's look at the specifications for a typical fuse.  The 'rating' of a fuse
is the typical current at which it will blow. These specifications vary a LOT with
different types of fuses, and these are for the common '3AG Fast Acting' fuse found in a
lot of electronic equipment:

<ul>
<li>110% of rating for 4 hours minimum </li>
<li>135% of rating for 1 hour, maximum </li>
<li>200% of rating for 5 seconds, maximum </li>
</ul>

These specifications tell us that the fuse can handle a little overcurrent for
a long time, but at twice the current it will blow in a few seconds. What the specs
don't tell us is that the fuse can handle even higher currents, provided it is only for a
VERY short time. 

The ability to handle very high currents for a very short time is very important in
electric motor applications. The Graupner Speed 600 8.4V BB motor (number 3316) is
typically used at about 18 to 20 Amperes. But the startup or stall current is actually 70
A! This has to do with the characteristics of DC motors, and is will be covered at some
other time - but it does happen. During normal operation, this startup current is present
for only a small fraction of a second, until the motor starts turning. The current then
falls to a more reasonable level. The ability of the fuse to handle a much higher current
for a very short time allows us to use a fuse rated for the running current, rather than
the starting current. If the motor is stalled, so it cannot rotate, it will draw the
starting current, and the fuse will blow quickly because it will be at a high current for
too long. 

<hr>

Safety Interlock Switches: What The Specifications Mean, and Why You
Don't Need to Care 

There are basically two configurations of safety interlock or "arming"
switches: an electronic interlock, which simply enables the controller, but does not
physically disconnect the battery from the motor controller, and an electrical interlock,
which physically disconnects the battery from the motor controller. If your motor
controller has really tiny wires running to a really tiny switch, it is an electronic
interlock, and there isn't much more for us to say about it other than to suggest that you
use it properly. 

On the other hand, if you have heavy wires running to a fairly substantial interlock
switch, you have an electrical interlock, and we can discuss what type of switch should be
used here and why. If you look at the side of a typical, good quality single pole, double
throw (also known as SPDT) toggle switch, you will probably see some numbers like: 

        2A 250V AC 

        5A 125V AC 

There may be some other numbers, but these are the ones we care about. The above
numbers are the current ratings for two different AC voltage levels that are commonly used
for power. The USA and a few copycat countries use 125V AC (well, we use about 117 VAC,
but that is a good enough approximation) and much of the rest of the world has the
sense to use something around 250V AC. The 'AC' stands for alternating current, which is
what you get from your household outlets, and NOT from a battery, and we will leave it at
that for now. 

The current ratings are really the value of interest: 5A, or 5 Amperes at 125VAC. What
use is such a low current rating switch in an electric airplane? Well, I took the
above numbers from the switch in the wiring harness supplied with the Thrustmaster motor
in a Great Planes PT Electric kit, and the harness also includes a 20 Ampere fuse.
Typical running current for this type of motor with an appropriate prop will be in the
range of 15 to 18 amps.What is going on here? Aren't we overloading the switch? Well,
yes and no. 

There are actually two current handling specifications for a switch: switching
current, and carrying current. In most applications, these are one and the same, so one
specification will do. But in our application, they are NOT one and the same. Consider
what we use the switch for: rather than using it to turn the motor on and off, we are
using it as an absolute cutoff, and under normal circumstances, you will only change the
position of the switch when the controller has already stopped the motor, so no current is
flowing. If no current is flowing when the switch changes position, then the switching
current specification is really not important. If current is flowing and the switch
position is changed, there may be arcing - a spark - across the switch as it changes
position. If the current exceeds the switching current specification, the resultant arc
may damage or destroy the switch. You may get away with it occasionally, but if you do it
often enough, the switch will fail. 

The carrying current specification is how much current the switch can handle when the
switch position is not changed. This is mostly affected by resistance of the switch
components, especially the movable contacts. In most cases, this is many, many times the
switching current specification - and this is why we can use a 5A switch in an 18 Ampere
application by never moving the switch while current is flowing. A reasonable
approximation of the carrying current may be found by simply considering it to be a power
limitation, rather than a current limitation. Power = Volts * Amps, so if the switch
is marked as 5A 125V, it can handle approximately 625 Watts. This should be adequate for
most electric flight applications - provided you do not open the switch while current is
flowing. 

I do not recommend using a switch to control a motor directly, so I am not even going
to discuss such applications. 

<hr>

The Reader's Survey 

What new electronic gadgets for electric flight would you like to see available?
Multiple output chargers? High current battery dischargers? Please email me at <a href="#skranish(at)ezonemag.com">skranish(at)ezonemag.com</a> 

<hr>

Reference Material 

Do you like to read? Do you like to collect impressive looking books in a bookcase, so
it at least looks like you read a lot? Well, this new section is for you. There is a LOT
of good technical and semi-technical information available specifically for model
aircraft, and even a lot of information for electric powered aircraft. Each month I will
discuss a book or magazine article, and the information about each of these will
eventually be collected on a separate web page. If you would like to recommend something,
please email me at <a href="#skranish(at)ezonemag.com">skranish(at)ezonemag.com</a>


<hr>

Tool of the Month 

Follow-ups on some previously mentioned tools: 

Dremel 'Mini Mite Rotary Tool Model 750' 

I did not have the Dremel catalog in front of me when I wrote about this tool, so
I did not realize at the time that they have introduced a new version with a 7.2V battery
pack. The Multipro 7700 cordless tool runs at a faster speed than the Model 750, and comes
in a case with an assortment of bits. It is somewhat bulkier and heavier, but should have
a lot more torque. 

Coverite 21st Century Iron 

I finally dug out my Fluke 52 datalogging digital thermometer, so I could
determine how much the temperature of this iron varies once it is set to a particular
temperature. With the iron set to about 200F, I measured a swing of 2F. That's right, two
degrees variation in the center of the shoe! This is one stable iron! However, I did find
that the calibration of my iron is off by perhaps 25F at this temperature. 

For the purposes of comparison, I dug out the old 'Sealector' iron that I have had
for perhaps 25 years. It is really an industrial iron, and is pretty obviously the
prototype that was copied for all the cheap irons coming from the Orient. The temperature
control is an electromechanical thermostat, and it is very hard to adjust to
a particular temperature because of the long thermal time constant. When I finally
got it adjusted to about 225F, which seemed to be as low as it would go, I measured a
swing of 6 degrees. This was more stable than I expected, but the most important
difference was that the old iron was very difficult to adjust because of the slow
response. The Coverite Iron responds very quickly to small changes in the temperature
setting. 

By the way, Coverite has recently been swallowed by Hobbico (Tower, OmniModels, Top
Flite, Great Planes, Dynaflite, etc.), so I would expect many of their products to
disappear once existing inventory is sold off. This iron could be one of them, because
they already sell cheap irons under other brand names. 

If you would like to recommend a 'Tool of the Month', please feel free to contact me at
<a href="http://rcgroups.com/shared/nospam.php?u=skranish&d=ezonemag.com">skranish(at)ezonemag.com</a> 

<hr>

<h2>Source of the Month </h2>

Multiplex, a German producer of sailplanes, electric sailplanes, and RC equipment, has
a new representative in the US. They sell a variety of electric kits, from pre-built
fiberglass and foam planes to simple, all sheet balsa Speed 400 class planes. The small
planes are particularly interesting because of their all-sheet construction and Jedelsky
wings. A Jedelsky wing is a sort of kludged undercambered airfoil made from two sheets of
balsa set at a slight angle to each other. While not perfect aerodynamically, it is strong
and very quick to build. 

I do not have any direct experience with Multiplex, but I have looked at their web
site: <a href="http://www.MultiplexRC.com#http://www.MultiplexRC.com">http://www.MultiplexRC.com</a>


If you would like to recommend a 'Source of the Month', please feel free to contact me
at <a href="http://rcgroups.com/shared/nospam.php?u=skranish&d=ezonemag.com">skranish(at)ezonemag.com</a> 

<hr>

<h2>COPYRIGHT </h2>

This document is copyrighted (c) 1997 by Steven Kranish, and may not be copied or used
in other forms of publication (electronic or paper) without written permission from the
author. I will probably grant permission, but I would like to know about it, so go ahead
and ask. 

<hr>

<h2>CONTACTS </h2>

If you have any questions, please feel free to contact me at <a href="#skranish(at)ezonemag.com">skranish(at)ezonemag.com</a>