THIS PROCEDURE WILL HELP YOU TURN AN OLD MOUSE INTO A LOGGING DYNAMOMETER.  WITH IT, YOU'LL BE ABLE TO PLOT YOUR MOTOR'S OUTPUT OVER THE ENTIRE RPM RANGE.

 

 

DESCRIPTION

A DYNAMOMETER, OR 'DYNO', IS A DEVICE USED TO MEASURE THE OUTPUT OF AN ENGINE.  IN INDUSTRY, TWO TYPES OF DYNAMOMETERS ARE COMMONLY ENCOUNTERED.  WE WILL BE BUILDING THE INERTIAL TYPE.  THIS TYPE OF DYNO WORKS BY ATTACHING A MOTOR TO A TEST MASS, USUALLY A FLYWHEEL.  AS THE MOTOR SPINS, SOFTWARE LOGS THE RPM VS TIME.  BY COMPARING KNOWN PARAMETERS ABOUT THE FLYWHEEL WITH THE DATA SHOWING HOW THE FLYWHEEL ACCELERATES, THE POWER AND TORQUE OF THE MOTOR CAN BE CALCULATED CONTINUOUSLY.

AS PRESENTED, THIS DYNAMOMETER WILL USE THE OPTICAL ROTATION SENSOR FROM A MOUSE'S SCROLL WHEEL TO RECORD THE FLYWHEEL'S RPM EVERY SECOND.  THE SOFTWARE SIMPLY REQUIRES INFORMATION ABOUT THE WEIGHT AND DIMENSIONS OF THE FLYWHEEL.

BUY A KIT IN THE STORE!
 

MATERIALS        

-MOUSE WITH SCROLL WHEEL (MICROSOFT INTELLIMOUSE IS BEST)
-SOME MISCELLANEOUS PLASTIC OR WOOD
-A FLYWHEEL (A CD WORKS WELL)
-A SHAFT AND BUSHINGS (INNOVATE)

EQUIPMENT        

-A COMPUTER

PROCEDURE

 

WHY?

1.   START BY PREPARING THE FRAME THAT WILL HOLD THE FLYWHEEL AND SHAFT.  YOU'LL NEED A BASE, TWO SIDES WITH HOLES FOR A BUSHING, AND TWO BUSHINGS TO HOLD THE SHAFT.

THIS WILL HOLD THE SHAFT OVER THE MOUSE SENSOR.  NOTE THAT I HAVE ALUMINUM BUSHINGS IN THE PICTURE.  THIS IS A BAD IDEA; USE BRASS BUSHINGS AND SHAFT, OR IF YOU HAVE THEM, BEARINGS.
2.  ASSEMBLE THE FRAME, MAKING SURE THAT YOUR MOUSE ELECTRONICS WILL FIT BETWEEN THE SIDES.  BE SURE THE FLYWHEEL IS ATTACHED SECURELY AND WELL BALANCED.  
3.  ATTACK YOUR MOUSE AND REMOVE THE CASING.  DISCARD ALL THE EXTRA PLASTIC, LIKE THE LIGHT PIPE, BUTTONS, SCROLL WHEEL MOUNT, ETC.  I ALSO CUT THE OPTICAL SENSOR OUT OF MINE, BUT DO SO AT YOUR OWN RISK, IT MAY KILL CERTAIN MICE.
THE ONLY THING WE'RE INTERESTED IN IS THE SCROLL WHEEL ENCODER.  I REMOVED THE OPTICAL SENSOR SO THAT SHOULD I BUMP THIS DEVICE MY COMPUTER CURSOR DOESN'T MOVE.

AS BEST I CAN TELL, THE SENSOR IN AN INTELLIMOUSE IS ONE OF THE BEST; GOOD TO 30,000 RPM.  OLDER MICE ARE MUCH LESS SENSTIVE.

4. THE SCROLL WHEEL SENSOR IS ATTACHED (ON THIS MOUSE ANYWAY) VIA A RIBBON CABLE.  IT NEEDS TO BE POSITIONED SOMEWHAT CENTRALLY SO THAT IT CAN BE PLACED UNDERNEATH THE SHAFT.  I USED SOME PIECES OF CARDBOARD GLUED TO THE MAIN BOARD AND THEN GLUED THE SENSOR BOARD ON TOP.
THIS HELPS RAISE THE SENSOR CLOSER TO THE SHAFT, AS WELL AS HOLDS IT IN PLACE. 
5.  ATTACH THE MOUSE WITH DOUBLE SIDES TAPE SUCH THAT THE SCROLL WHEEL SENSOR IS DIRECTLY BELOW THE SHAFT.  
6.  FINALLY, ATTACH A SEMI CIRCULAR PIECE OF PLASTIC OR OTHER OPAQUE MATERIAL TO THE SHAFT, SUCH THAT FOR HALF OF THE SHAFTS ROTATION THE PATH BETWEEN THE IR LED AND THE IR SENSOR IS BROKEN. THIS PIECE REPLACES THE ENCODER WHEEL THAT LIVED IN THE MOUSE.  BY LOWERING THE NUMBER OF SLOTS FROM ABOUT 20 TO JUST ONE, THE SENSOR CAN NOW RESOLVE MUCH HIGHER SHAFT SPEEDS.  THIS SACRIFICES SOME RESOLUTION, HOWEVER.

SOFTWARE

I WROTE SOME SOFTWARE THAT CONTINUOUSLY PLOTS RPM, TORQUE, AND POWER, AS WELL AS CAN EXPORT TORQUE AND POWER VS RPM CURVES TO EXCEL IN .CSV FORMAT.  OBVIOUSLY YOU COULD WRITE YOUR OWN IF YOU DESIRE; ACCESS TO THE SCROLL WHEEL IS EASILY ACCOMPLISHED WITH WINDOWS APIs.  I'LL GO OVER HOW TO USE MY OWN SOFTWARE:

DOWNLOAD SCI-SPOT.COM DYNAMOMETER
DOWNLOAD SOURCE

LAUNCH THE DYNO SOFTWARE AND CLICK SETUP:

THEN CLICK SENSOR SETUP.  ENSURE THAT THE SENSOR IS SET TO ONE ("1") SLOT:

...

NOW CLICK TEST MASS SETUP.  IN THIS WINDOW, DEFINE THE PARAMETERS OF YOUR FLYWHEEL AND SHAFT.  IF YOU'RE NOT USING A CIRCULAR FLYWHEEL YOU MAY CALCULATE THE MOMENT OF INERTIA BY HAND AND ENTER IT AT THE BOTTOM OF THE FORM.  ALSO, YOU NEED TO ENTER EITHER THE MASS OR THE DENSITY OF EACH PIECE.  YOU DON'T NEED TO KNOW THE MASS AND THE DENSITY.  (FOR PERSONS USING A CD, A STANDARD CD WEIGHS 15.6 GRAMS.  A MINI CD WEIGHS 6.7 GRAMS).

CLICK OK TO CLOSE THE SETUP BOXES.  AT THE MAIN FORM, CLICK START, AND THEN APPLY POWER TO YOUR MOTOR.  LOGGING AND PLOTTING WILL START AUTOMATICALLY:

NOTE:  THESE GRAPHS ARE SIMULATED BECAUSE I DID NOT HAVE THE DEVICE ON HAND WHEN MAKING THIS PAGE.

CLICK STOP WHEN YOUR MOTOR HAS REACHED IT'S MAXIMUM SPEED.  A BOX WILL APPEAR SO YOU MAY EXPORT THE OUTPUT VS RPM CURVES TO EXCEL.  HERE IS A SAMPLE GRAPH FROM THE EXCEL OUTPUT.  THIS IS FOR A SMALL PAGER MOTOR:

NOTE, THE PROGRAM PLOTS AN AVERAGE OF THE ACTUAL DATA COLLECTED; THIS PRODUCES SMOOTHER, MORE EASILY READ GRAPHS.  THIS DOES, HOWEVER, PRODUCE AN INACCURACY IN THE FIRST DATA POINTS IN THE TORQUE CURVE.  SIMPLY KEEP IN MIND THAT TORQUE DOES NOT EQUAL ZERO AT 0 RPM.

CALCULATIONS

THE MOTORS EFFICIENCY AT IT'S PEAK RPM CAN BE CALCULATED BY:

EFFICIENCY = MAX POWER FROM GRAPH / (VOLTS * AMPS)

FOR THE PAGER MOTOR SHOWN, THE EFFICIENCY IS ABOUT 2%.

READER QUESTIONS

JEREMY - Hello I had a quick question about your motor dyno: First, can I use a prop in place of the flywheel or would that make a difference as far as where the motor creates its most torque? Second, can I use a mouse with a USB port so I still have use of my other mouse? Thanks for the help.

ANSWER - No, do not use a propeller. You'll notice the program is set up to calculate the moment of inertia automatically for a disk only; you'd have to calculate your propeller's moment of inertia by hand. This is ignoring the major reason why you shouldn't use a propeller: the program cannot account for air drag. Obviously the drag on a simple spinning disk is low, but its HUGE for a propeller. This drag will not only produce a graph that shows lower than actual torque, but it will also shift the peak.

You may use any mouse that works in windows. Windows is very tolerant of multiple mice, just make sure the scroll wheel works (like you can scroll windows and stuff) before continuing.  The mouse I used was a USB mouse, in fact.

JAY - I am a builder experimenter of CD-ROM motors [for model aircraft use] and was given the link to your site by a friend. Currently I am using a host of measuring devices (digital scale, laser tach., watt meter, o-scope, and multimeter) to capture test data and motor constants (Io, Rm, and Kv). Would your device be applicable to my motors (is it big/tough enough) and how is the data from the dyno best used to predict motor / prop combinations?

Dynos are a new concept for me and I'm not sure what they measure?

ANSWER - You may have to 'beef it up' a bit for a CDROM motor but there no reason why the technique and software won't work.

The dyno data is useful in several ways. Torque is used to find the speed that a propeller will turn at, and power will tell you how fast it will reach that speed. Note that power is not really an important concept here, since aircraft propellers are constant speed devices, unlike RC cars for example where the acceleration of the motor directly equals the acceleration of the car.

The torque curve, however, is extremely useful. The best use of your power would be a propeller that spins at the location of the torque peak. Theoretically that combination delivers the most energy to the aircraft (this is an approximation but it is usually correct. The airfoil design of the propeller can affect this). You should be able to find a table for a given propeller that shows the torque required to turn it a given rpm. You want to choose a prop who's required torque at the peak torque rpm of your motor equals the peak torque (net torque = zero).

If you can't find a table, you can do without it. Simply test the motor with a flywheel to find and record the RPM that torque peaks at. Then start fitting different propellers, and the optimum prop *should* be the one that spins at the RPM you recorded.

Note that for electric motors the torque peak is usually at zero, which complicates the process.  Power peak provide a suitable surrogate in that case.

TIM - What actual math is done with the raw data (pulses that are directly proportional to the RPMs I assume) to get the torque and other information. I am asking because I am doing a science fair project a that involves the same stuff. Also, what "recommended reading" would you suggest for me? I am a junior in high school and am taking honors physics and calculus. Thank you in advance for any help you can provide.

Basically what I am doing is building a dynamometer for small glow engines which are used in RC cars and trucks. These can reach in speeds in excess of 40k rpm and the manufactures have unbelievably inflated horsepower and torque numbers (or so I have heard).  I'd like to find out if the numbers advertised for these engines are for real.

ANSWER - First off, a gas engine will be a good challenge, but I believe you can accomplish it. Some things to consider:

-You値l need a large test mass. If I recall my RC car days correctly a good .21 was claimed to be ~2.2 horsepower, which is about 1500 watts. You値l need to use that figure to ensure that your motor accelerates the cylinder nice and slowly. I壇 expect your flywheel to be on the order of several pounds.

-Clutches will be your largest challenge. You値l need a clutch because it痴 unlikely that you値l get the engine started with such a large mass (flywheel) attached. At the same time, measuring power and torque after a clutch could give you distorted numbers because of slippage. Of course, taking some slippage into account gives you a better view of what you値l see in your RC car, since the clutch will be a factor there as well.

-You値l need a good test stand. And by good I mean VERY good. High quality bearings, and シ or larger (3/8 or ス wouldn稚 be overkill) shafts. That motor will never reach 40k RPM loaded with a flywheel this larger, but a mass the size that you値l need is NOT something you want breaking loose, even at 10k RPM. It would easily devour your leg, the wall behind it, and any passing cars in the street.

-A reduction belt is trivial with respect to the other engineering and fabrication you値l be doing. Moreover its an excellent idea. Obviously the belt will keep the speed of the flywheel in the range of the mouse痴 optical sensor, but moreover using a belt will lower the tip speed of the flywheel, which lowers your losses due to air drag. Keeping things moving slow is always an excellent idea. Go belt.

Now for the math:

Some constants:

w = d(phi) / dt = angular velocity = RPM * 2 *pi ゥャ (note the 2 * pi, not phi.)

dw / dt = angular acceleration in radians

T = torque

W = work = change in energy

t = time

n = number of samples recorded

dn / dt = samples recorded per second = n / sample rate

Some fundamental physical truths:

dw / dt = T / I

I = ス m * (Ro^2 + Ri^2) where m is the mass of the flywheel and Ro and Ri are its inner and outer radii.

P = T w

Practical equations:

As you know, the program samples the mouse sensor every few milliseconds. The mouse reports back with the number of samples its recorded since last time it was asked. This number is n. The program asks the mouse for this data every 妬 seconds (i is chosen as a random variable, you get my point). Thus:

dn / dt = n / i

Now, a parameter of your device will be the number of 壮lits you put in on the shaft. Each slit will count as a sample (n) as it passes the sensor. If your disk has one slit, then the mouse will report one sample for each rotation. If you have two slits, then you see two samples per rotation etc. The advantage of using a single slit is that the mouse can read ~10k samples per second, which is 10k RPM. If you use two slits, then this 10k samples corresponds to only 5k RPM. The advantage to using more slits is you get higher resolution data, and lower error. One slit is fine for the speeds involved here. Consider S as the number of slits on the shaft, thus:

w = 2* pi * n / (i * S)

Now we can find w many times a second by querying the mouse sensor and applying that equation. For human minds, RPM is easier to understand and can be found with:

RPM = 60 * n / (i * S)

As stated above, dw / dt = T / I which means T = I * (dw/dt). dw/dt can be found by querying the mouse for w, and then comparing the calculated value of w with the value calculated the previous time you asked for it (w0).

dw / dt = (w w0) / i

Recall 妬 is the number of seconds between queries of the mouse sensor, and 努0 is the w calculated from the previous sample. Now simply:

T = torque
   = I * dw /dt
   = ス * m * (Ro^2 +Ri^2) * (w w0) / i

Voila, half the battle is done. Power is simply P = T w so:

P = T w
   = I w dw /dt
   = ス * m * (Ro^2+ Ri^2) * w * (w - w0) / i


And there you have it. Regarding actually getting this data from the mouse into vb, simply download the source code for my program. It壮 available below the program itself on the Dyno page.

As far as required reading is concerned, I致e never been one for reading my text books, so I知 not very knowledgeable.  One online resource of impeccable quality is HyperPhysics (http://hyperphysics.phy-astr.gsu.edu/hbase/hframe.html) which has answered more questions for me over the course of my education than I could count.

RICH - I have been fiddling with your little Dyno and have a few suggestions.

1) Maybe use a r/c car transmission to reduce the rpm's to a safe level. An Associated TC3 has a transmission ratio of 2.5 which would decrease the expected Stock motor rpm's from say 22k to 8.8k

2) Using the already assembled transmission gives you a more reliable mounting means

3) You would get an error with respect to torque due to the loading of the gearbox.

4) You could also use an existing gearbox/outdrive to power the CVD to the wheel axle and use a standard TC wheel as your flywheel and also it acts as your counter through the IR sensor.

Other dyno's I have seen use a hall effect device to measure rpm's but I am not sure how they run the torque curves.

Also, the master slave dyno using the drive motor to power a slave motor which is where a resistive load is used as a torque load may work too.  What do you think?

ANSWER - Thanks for the suggestions, I like them a lot!

1,2,3) Using a transmission is a fantastic idea with respect to the reduction gearing, but perhaps even more so for the motor mount. In fact the losses due to friction and rotating mass of the transmission may not even be a bad thing. For RC car applications anyway, the transmission is something you can't live without, so reading the power and torque after the tranny gives a better idea of how your motor's performance translates to the car's performance.

4) I had thought on this, the only issue would be calculating the moment of inertia of the wheel. The shape of the rubber tire varies with speed so while possible, my software doesn't support it, and it would be computationally complex. If that problem were overcome you could lock the differential(s) and then measure the torque curves while the car is fully assembled. I may be nitpicking about the tire expansion; I'll bet you can get good data even without worrying about it. Still, calculating the moment of inertia for an irregular body like a wheel & tire seems like it'd be a brow-furrowing experience.

Regarding the slave motor there would be two ways to approach that situation, both of which I didn't want to touch even though it may offer certain advantages. The easiest way would be to hook the second motor up and simply monitor the voltage produced by the slave motor. The relationship between RPM and voltage is non linear so you'd need to calibrate the motor with a controlled RPM source (ick). The better way would be to hook the second motor up to a variable power supply. The idea would be to increase power (load) to the second motor until the RPM's held constant. Again, you'd have to calibrate your test motor with respect to torque vs input voltage (double ick). I *think* that the later method is the way most professional dynos work, but admittedly I'm not well versed in the advantages it offers. Physically I can find no reason why all method wouldn't produce the same results.

If I was to try using a hall effect RPM sensor, or using a slave motor, I'd almost certainly log the data with the analog input that every computer is blessed with: the soundcard.

GOT A QUESTION OF YOUR OWN?  CLICK HERE TO ASK IT!