Sunday, February 6, 2011

Choosing the Right Motor - Reality vs. Sales Pitch

There are a lot of people mundanely peddling motors on the common RC message boards. If you were in the business, wouldn't you?  The sales people are easy to identify, they make 1,000s upon 1,000s of posts, often 10s of thousands on many different boards to subtly, or more often brashly push their products to mostly unsuspecting victims.  The favorite method of collusion between board and salesman is the simple rigged review, usually fluffed with all kinds of silly to ridiculous claims that make no aerodynamic sense whatsoever.  Their tactics wouldn't be as bad if they didn't so flagrantly trash their competition, often taking aim at products which are demonstrably better than their merchandise.

So how do you distinguish a sales pitch from reality?  It's easier than you think. If it is permitted to appear on one of the larger forums, it is a sales pitch.  Just because a product is being pushed without revealing associations to paid sponsors doesn't make it bad, though it does make one wonder why they feel the need to misbehave. 

With that in mind, is there anything one can do to determine which motor might be the best choice for a particular airplane?  Yes!  Understand the common misconceptions, usually spread by the salesman/reviewers themselves who lack any technical background to make sound judgements, and it is easy to separate the wheat from the chaff.

Let's start by debunking the sales pitch of choice, the ground tested "data base" of good and bad motors, complete with charts and graphs of statistical noise.  Some people have seemly dedicated their lives to creating these long lists of meaningless gobbledy gook.  Let's be clear, you cannot determine anything useful from a ground test of any motor in isolation if that motor is intended to fly.  Looking at data bases of ground test data is a total waste of time, even if the creators' intent wasn't to trash superior competition.  Why?

Where to start?  How about if we start by assuming ground test data bases are not complete garbage, and see how far it gets us?  Let's be intelligent, and start by defining some notional requirements.

Let's say we want to power a model with the goal of (A) flying very casually, and (B) achieving the longest possible flight times.  We need (C) a power reserve for takeoff and landing with a go around option in a field boxed by trees.  The battery compartment is (D) fairly small, so we have a very limited selection of suitable batteries.  The ESC is up to us, we'd like to (E) minimize cost.

So let's go to a motor data base and absorb all the wonderful misinformation.  Here's one that takes the time to describe how they rate engines (big mistake), let's examine their own example:

The above graph (minus annotations) was produced with the following link here

1) Vertical labels indicate the current the noted prop would draw at the given voltage. Note the diamond is drawn on the blue efficiency curve, but they could have been correctly placed on any of the curves since the diamond indicates the current draw that prop will draw on the noted motor at the noted voltage. In this case, putting a Graupner Miniprop 4.3x2.0 motor on a HiMaxx HA2025-4200 motor and powering it from a 10V supply will result in the motor consuming just over 18A, the prop producing 962g of static thrust at a pitch speed of 69 MPH. Prop data used in the graph is measured data imported from the excellent work done in Drive Calculator.

If the prop label is in black, then the prop RPM is estimated safe. If the prop label is gold/yellow, then the prop might be operating at an unsafe RPM and you should consult the manufacturer's specifications on that prop. If if the prop label is red, then the prop RPM is exceeding the manufacturer's RPM recommendation.

2) Reading motor efficiency is done by noting intersection of current and the blue efficiency curve. With the 4.3x2 prop, efficiency is about 83%. Note that ESCs and batteries and the the prop are not factored in here. This is simply the Pout/Pin of the motor, which is a measure of how good the motor is at converting electrical energy into mechanical energy.

3) Noting where the green line intersects the operating current will yield the power input to the motor. With the 4.3x2 prop, it's about 185W of point going into the motor.

4) Noting where the gold line intersects the operating current will yield the RPM for the prop. In the case of the 4.3x2 prop, it'll consume just over 18A at 10V on this motor, and turn about 36K RPM at that voltage.

5) Thermal power is a measure of how much power is converted into heat. The closer you operate to peak efficiency of a motor, the harder you'll be able to push the motor because less of the power is being converted into heat. Currently, the motor weight is used to determine how much heat the motor can dissipate and for how long it can dissipate it. Ideally we should be characterizing various motor cases to determine Theta for case-to-ambient. T = Ta+ Pd * Thetaca. For now, the closer you operate to the red region, the hotter the motor will get. Inside the red region may damage the motor. I really depends on an enormous number of variables.

6) These labels note the parameters used to generate the graph.

7) A text summary of the red region of the graph.



9) The title idicates the manufacturer, model of motor, and the voltage.
So let's start with assertion # 1 that this motor will draw just over 18A and generate 962g of static thrust.  True?

Of course not.

The motor will pull far less than 18A once airborne, because any fixed pitch prop is nowhere close to its design AoA at zero airspeed.  Every prop is optimized for a certain combination of airspeed and RPM (and all are different).  The combination of airspeed (the forward vector) and RPM (the horizontal vector) determines a resultant relative wind vector.  When the resultant wind vector is compared to the prop airfoil's chord line, a certain Angle of Attack (AoA) is achieved.   When the AoA is correct, the prop is operating as designed.

So when the airspeed is zero, no useful data can be gleaned or compared, especially if tests are conducted in a trade space that includes various props.  Every prop will operate at a dramatically different delta from it's design lift and drag, depending on it's pitch, airfoil, chord and twist, and there is no constant relationship inherent to any "data" we might measure.  In general, the steeper the prop's pitch, the farther out of whack the test will be.  High pitch airfoils, or parts of them, will simply be stalled during the test, while lower pitch props will be closer to but never achieve their design AoA, so they'll artificially perform completely differently on a test stand, usually generating more thrust in a relative sense but with no measurable relationship to reality.

Even though all the "information" extracted is wrong in the first place, the longer the "test" is allowed to run, the more the air column will start to move, like a hovering helicopter creates re-circulatory downwash.  If the test is conducted inside a house or room, the size and shape of the room will change results your results, because an RC motor typically creates airflow between a few mph and 100 mph or more, requiring a massive venue to avoid changing the garbage data output.  To quantify this one of so many compounding errors, I did a test in my 4 car garage (3 spots are in tandem).  Placing the test airplane in the middle of the long lane colored the known-wrong thrust measurements by almost 10% (roughly 3 oz delta over 40 oz thrust) simply by opening and closing the garage door.

To conceptualize this concept, imagine a motor boat anchored in the middle of a still pond. Run the throttle to full. The force on the anchor chain is substantial, but the propeller is mired in turbulent water, the engine is artificially taxed, and the prop blades are largely stalled even through the drag from the boat is zero. If I anchor the same boat in the middle of a flowing river, where the water current is perfectly matched to the design speed of the boat's power system, then you instantly see the absurdity of static "thrust" testing. If the motor was working as designed, then there would be no measurable thrust (i.e. the anchor chain would go slack).

So much for making any progress on requirements A, B and E.

So stop right  there, ground test data bases are positively useless. But let's keep going just to humor the novices that create these impressive monuments to aeronautical engineering ignorance.

Thrust is 962 grams, right?   Of course not.  The net force pushing an airplane forward is "thrust." The motor only accounts for part of that equation.  Whenever you blow air on something, it feels a force.  That force is called drag.  Whenever you run up a motor, especially in a conventional tractor config, there is a lot of airframe drag.  Since the propeller has to pull the plane (too bad it cannot proceed along by itself) the plane is usually attached to it.  When the prop pulls one way on the shaft, the airplane pulls back the other way.  The net result, not the prop's contribution taken in isolation, is "thrust."   What this data base is telling us is UNinstalled thrust, which has no relationship whatsoever to installed Thrust, or Thrust Available, T(A).

If you mount the motor in the middle of a flat board, the motor will produce no thrust.  If you mount the motor using a tiny suspended nacelle that has no aircraft form behind it, you might come closer to the uninstalled number, in exchange for possible  handling issues and increased structural requirements.  A conventional nose-mount might lose 30% to 80% of uninstalled thrust to installation error, depending on the aircraft form and the relative scale, twist, and airframe match of the selected prop.

Propellers must be well matched to the aircraft, first.
This one is designed to produce maximum thrust outside of the
cowling boundary, while producing just enough inside air flow for 
cooling during ground ops with the cowl flaps open.  The 3-blade 
design transmits a lot of engine power to the air  while reducing
the necessary diameter of the prop to avoid prop strikes with
the ground.  3-blade (and 4, 5, 6...) props are easier to spin, per
blade, during static runs because they create more turbulent
wake that reduces drag (and produce less lift), but are overall
more difficult to turn once forward speed allows 
each blade to bite clean air. 

So stop right  there, again.  All ground test data bases are positively useless.  But let's keep going anyway to see what other misconceptions are pervasive when trying to push ill-suited merchanside. 

The next discussion is about how "efficiency" is measured, because it is presumably good.  Then they go on to embarrass themselves by using motor weight as the driver for heat reduction.  In other words, the heavier the motor, the better it is for aviation.  Smart.  Not.

But even if this wasn't, obviously, backwards, the very assumption that efficiency is good is preposterous.  Good for what?  Well, good for being efficient, I guess.  But who says we desire efficiency?  Did they use telepathy to divine that everyone wants to drive a Prius?  Some people actually want high performance and are willing to compromise efficiency.  Why, some people even enjoy driving '60's muscles cars around!  Shame on you!!  Don't you know the car is not efficient?!

But let's humor their absurd assumption, because I want to show that their methods are backwards even if their assumptions were sound, as laymen musings often are (just look at the Prius driver trying to save trees by minimizing their output of life-giving CO2).   What makes an electric motor efficient?  They've tried to estimate it with their egregious assumption that motor weight is good for efficiency, noting that a motor's ability to "dissipate heat and how long it can dissipate it" is somehow a good thing.  So... if I take a motor that converts 100% of it's fuel into it to heat, and incorporate a 90 lb heat sink, then I've achieved efficiency Nirvana, right?  Efficient motors don't dissipate heat well, they create a minimal amount of heat in the first place.

So let's ignore that backwards logic, too.  Well, we already disposed of the notion that everyone wants an efficient plane, that's bunk, but what if you actually do want an efficient plane - kind of like our requirement A, which I graciously aligned with their false assumptions and methods for a reason. 

There is another level where they've boarded the opposite direction train. How do electric motors generate less heat per unit power?  Generally speaking, by minimizing resistance to lots of current.  For any given voltage, that means heavy gauge, highly conductive coils (not heavy cases or heatsinks as they assert).  Ah ha, so we doooo want some heavy parts inside the motor, so maybe generating less heat during a ground test could be an ok metric on some level, right?  Wrong.  What we want, even for our carefully chosen, efficiency-oriented requirement A, is the smallest motor that remains reasonably efficient spinning the prop at it's design AoA (which is, remember, a combination of forward airspeed and RPM), where that airspeed also happens to match the most efficient airspeed for the airframe.  The motor should not incorporate coils so heavy as to be too efficient at full power.

The "efficient airspeed" of the airframe depends solely on our requirements.  In requirement B we specified our goal is long flight duration, so we want to optimize motor efficiency at the airplane's L/D)max airspeed, not at full speed.  L/D)max is the airspeed that minimizes the sum of Induced Drag (caused by lift creation) and Parasite Drag (caused by aircraft form, skin friction, interference, and with fast airplanes, wave drag).

Does the motor "data" base know any of this?  No.

Turbojets are very efficient.  Turbofans are very efficient.  Turboprops are very efficient.  See a pattern?  All motors are very efficient within certain performance parameters.  All motors are very inefficient within certain performance parameters.  Efficiency is a meaningless word without aircraft design parameters.  If one is dumb enough to attempt to optimize motor efficiency in a vacuum, there is about a 100% chance they will drive the overall solution toward woeful inefficiency.

Even if we ignore the fact that it is a clueless recommendation, and blindly choose the heaviest motor assuming that heavy casing or a heavy shaft is somehow related to efficiency in addition to the purity and conductivity of the design the motive parts, we'd always choose a motor that is massively over-designed (read: too big, too heavy and too power hungry) for a given aircraft.  The assumption that max motor efficiency (both incorrectly defined and incorrectly measured) is somehow good, drives us to install anvils that aren't working nearly hard enough to be light, and thus efficient, as part of the overall flying solution.
Smaller/lighter motors = higher efficiency in the overall flying solution.
The motor on the left weighs 100% more and draws 13% more power,
while producing 11% more installed thrust.  As the shadows indicate,
the silver front and back portions of the left motor's case are empty.
Worse, much worse, is the aerodynamic design reality that for every 1 unneeded oz, you must add about 6 ozs of unneeded support.  More weight means you need bigger wings, that means more lift, more lift means more drag, more drag mandates more thrust, more thrust needs more fuel and beefier support structure and power control systems, more wing, motor, and fuel and support systems means even more weight, and that additional weight means more wing, motor and fuel and support systems, and so on, and so on, and so on....  The effect of adding that extra oz finally ends as these infinite integral summations mathematically settle, and that happens around 6 to 1.

Point being, for efficiency's sake, we don't want a heavy motor that feels particularly relaxed most of the time, or we are hauling more dead weight all of the time.  We want a light engine that must work very hard, hard enough to pull its own weight.  We also don't want a motor that overheats and dies on takeoff.  But truth be told, and for true efficiency sake, under-design, not over-design, is generally the mundane path that leads to high efficiency, complete flying solutions.  In other words, the most efficient solutions are at the opposite end of the spectrum from what most toy motor testers are erroneously trying to optimize, in the name of efficiency.

We've already "stopped right here" enough, so I won't repeat it anymore.

What about the whole idea of characterizing motor performance without understanding the scale and type of aircraft.  Duh.  If one motor/prop produces 50 oz of static thrust and another produces 40 oz, is that good?  Maybe.  Maybe not.  We have no idea without tons more information.  Even if we assume a static test has some unknown but useful relationship to actual flight, if some motor/prop combo pumps out 50 oz and weighs 10 oz more than the 40 oz solution and also requires 5 oz more battery, then it probably sucks by comparison--especially after the 6:1 rule mandates increasing the size of the airframe to achieve proper handling.  You can't put little balsa wood fins on a 1,600 hp Rolls Royce Merlin and expect it to fly right.  Not caring a whit about the scale and weight of the power system as a whole (= weight of the motor, prop, battery, Rx, ESC, wiring, and all airframe structure required to do more than glide without it) is actually quite bizarre. 

And what about that wide scale radial cowl, perhaps designed to minimize drag using cooling airflow over conventional cylinders?  Or what about the airplane's particular design and form? Gee, does the airplane, and the purpose of its full scale parents matter at all when selecting an RC motor?  I wonder...

I once read  a toy tester who goes by the name of "Dr" Kiwi (tens of thousands of mostly erroneous posts on several internet chat boards = salesman) suggest that the best diameter prop for a certain person's 480 motor was 9", since it was "the best performing prop" on his homemade static thrust tester.   I saved the person from having to repair a crashed airplane by pointing out that a 9" prop would hardly clear the large scale radial cowl of his plane.

The airplane actually matters. 

But let's not even pick apart the idea that you can choose a motor and prop without knowing the desired purpose, type, or scale of aircraft, and finish up with this:  Matching a prop to the design goals and desired purpose of the airplane is a black art that almost always produces some unexpected results.  Anyone, anyone, who tests an isolated aircraft motor or prop on the ground, without knowing the exact airframe and performance objectives, and the weight of the power system as a complete flying solution, is beyond clueless.

So how do you pick the right motor for your application?  Unfortunately, its a complex decision and the process itself changes from aircraft application to application.  The answer starts with defining your own unique requirements for the overall solution.  It ends with ignoring anyone who thinks they've solved all or part of your problem in advance--that is a sure path to picking the pig of the litter.

Lastly, I suppose I might go easier on ignorant toy testers if they weren't mostly slimy salesmen.  Slimy is to be expected in sales, but I really hate when I know a gazillion times more than a salesperson who's only job is to learn about what they are selling.