Dyno, Homemade, Using a Novak Sentry Data Logger, Continued, The Experimental Thread.
06-05-2009, 05:10 PM
#181
I'm not going to get to worried about the measured current values. My datalogger is 5+ ears old so It could be any of the things you mentioned or a bad solder joint or wire somewhere.
The point of the program is to predict car acceleration versus time. If you look at the speed versus time graphs the results are very close . 40mph measured versus 41.5 mph predicted. That's 3.75% error. Very acceptable if you ask me.
What I'm going to do is try and build some motor models based on John's testing. I'll see what I can come up with. Then you can dyno a motor and use that data to build a computer model and then tweak gearing to optimize performance.
John if you have some Excel Dyno data you could send me that would help.
The point of the program is to predict car acceleration versus time. If you look at the speed versus time graphs the results are very close . 40mph measured versus 41.5 mph predicted. That's 3.75% error. Very acceptable if you ask me.
What I'm going to do is try and build some motor models based on John's testing. I'll see what I can come up with. Then you can dyno a motor and use that data to build a computer model and then tweak gearing to optimize performance.
John if you have some Excel Dyno data you could send me that would help.
06-05-2009, 05:18 PM
#182
Bob-I'll be happy too. Tell me what motors you want and I will attach it to your e-mail. [email protected]. Nice Results.
06-15-2009, 11:21 PM
#184
We have had good luck with a steel flywheel 3.472" diameter, 3.395 x 10^-4 kgm^2/s^2, 359.1 g for those motors.
Also go here.
http://www.rctech.net/forum/5698920-post46.html
Mattnin post the thickness if you are still around.
Also go here.
http://www.rctech.net/forum/5698920-post46.html
Mattnin post the thickness if you are still around.
Last edited by John Stranahan; 06-16-2009 at 10:24 PM.
07-01-2009, 07:54 PM
#187
Well, believe it or not progress with the spreadsheet and simulators continue and show more promise.
First Bob used some log polynomials to improve the fit of the model to data late in the run that I normally just truncate or get rid of. We had spoken of this, but he actually had the skill to effect it.
We had some communication on how accurate this tail end data was as it appeared to me that the motor was working much harder by the sound of it and by the ferocious amount of air streaming off the flywheel than the data suggest. For large number of coils (stocks, specs) the error might be on the order of only 8% from blowing the air. At the 70,000 RPM of a 3.5, the error might in fact be several times more than this. Anyway Bob gave me a link to a good article on this type of loss. He might add a correction for this as time permits. At this point (after a correction for air friction) I would be happy to include the truncated data.
Now the good part.
Bob had written a straight line acceleration run simulator. Similar to Joe's oval simulator in some respects. The output of the straight line acceleration simulator looks very similar to the velocity curves vs time given by data collecting radar that some of the RC magazines have.
I sent Bob my dyno data from the 10.5, that I race, along with a 46 mph radar measured top speed with a 72/23 gear ratio. The following steps were applied.
the log polynomial was applied to widen the power curve that was available to the simulator.
The power curve is input directly into the simulator to "power it" so to speak. Car mass, gear ratio, and other parameters are entered.
A simulated 46 mph was obtained.
The gear ratio was tinkered with.
A new ratio of 72/21 was predicted as getting to 46 mph in a shorter period of time. Now that is the exact gear I used to like on our old asphalt and track. I put those gears in the car (along with some other changes that are partially responsible for the increase) Now I am at 48 mph. That is a huge difference. Those MPH at that speed do not come easy. Air resistance is increasing as the square of the speed. Combined with seat of the pants feel, I will have to conclude most of my increase was because of gear change. Acceleration out of corners is much improved on a good traction surface. The dull feel of the 72/23 may still be preferred by some when it is dusty. A 72/23 may also lead to 190 F plus motor temps for some racers (not those with front diffusers on board). My cars motor is at a happy 150 F at 100 air temps and 140 F track temp. No motor fans.
Again in a second completely different discipline (different than 4 cell/ 13.5 oval), I gear not using max motor temp, but some dyno output results that indicate lower gears are more appropriate. Gearing, until the motor is almost smoking, is again found to be slower. I am now using 2s LiPo and a 10.5 in a pan car that is very similar to the oval car.
Anyway that is just thrilling results. Who would have thunk it. Stay tuned. I will give more details as I actually use the simulator myself and tinker with its parameters. These will be slow ongoing progresses.
Thanks bob.
First Bob used some log polynomials to improve the fit of the model to data late in the run that I normally just truncate or get rid of. We had spoken of this, but he actually had the skill to effect it.
We had some communication on how accurate this tail end data was as it appeared to me that the motor was working much harder by the sound of it and by the ferocious amount of air streaming off the flywheel than the data suggest. For large number of coils (stocks, specs) the error might be on the order of only 8% from blowing the air. At the 70,000 RPM of a 3.5, the error might in fact be several times more than this. Anyway Bob gave me a link to a good article on this type of loss. He might add a correction for this as time permits. At this point (after a correction for air friction) I would be happy to include the truncated data.
Now the good part.
Bob had written a straight line acceleration run simulator. Similar to Joe's oval simulator in some respects. The output of the straight line acceleration simulator looks very similar to the velocity curves vs time given by data collecting radar that some of the RC magazines have.
I sent Bob my dyno data from the 10.5, that I race, along with a 46 mph radar measured top speed with a 72/23 gear ratio. The following steps were applied.
the log polynomial was applied to widen the power curve that was available to the simulator.
The power curve is input directly into the simulator to "power it" so to speak. Car mass, gear ratio, and other parameters are entered.
A simulated 46 mph was obtained.
The gear ratio was tinkered with.
A new ratio of 72/21 was predicted as getting to 46 mph in a shorter period of time. Now that is the exact gear I used to like on our old asphalt and track. I put those gears in the car (along with some other changes that are partially responsible for the increase) Now I am at 48 mph. That is a huge difference. Those MPH at that speed do not come easy. Air resistance is increasing as the square of the speed. Combined with seat of the pants feel, I will have to conclude most of my increase was because of gear change. Acceleration out of corners is much improved on a good traction surface. The dull feel of the 72/23 may still be preferred by some when it is dusty. A 72/23 may also lead to 190 F plus motor temps for some racers (not those with front diffusers on board). My cars motor is at a happy 150 F at 100 air temps and 140 F track temp. No motor fans.
Again in a second completely different discipline (different than 4 cell/ 13.5 oval), I gear not using max motor temp, but some dyno output results that indicate lower gears are more appropriate. Gearing, until the motor is almost smoking, is again found to be slower. I am now using 2s LiPo and a 10.5 in a pan car that is very similar to the oval car.
Anyway that is just thrilling results. Who would have thunk it. Stay tuned. I will give more details as I actually use the simulator myself and tinker with its parameters. These will be slow ongoing progresses.
Thanks bob.
Last edited by John Stranahan; 07-02-2009 at 07:18 PM.
07-02-2009, 04:55 PM
#188
Tech Adept
This new development sounds promising. I'm eager to see how it pans out.
I'm going to revisit my study of the sweet spot in the gear ratio. Maybe I can come up with a closed formula that predicts the optimum gear ratio and helps explain these results.
I'm going to revisit my study of the sweet spot in the gear ratio. Maybe I can come up with a closed formula that predicts the optimum gear ratio and helps explain these results.
07-04-2009, 02:10 PM
#189
The explanation is fairly simple, well sort of. It all boils down to two factors acceleration and top speed. The torque delivered to the rear wheels translates directly to acceleration. More torque more acceleration. To generate the maximum torque you want the highest gear ratio you can tolerate before wheel spin becomes a problem.
Generating top speed is exactly opposite to torque. Lower gear ratios will provide higher top speeds until you reach the point where the power output of the motor cannot overcome air resistance, drivetrain friction, rolling resistance and so on. You need a long straight to reach this point.
What we want for on road is the best compromise between the two. Good acceleration out of the corners while achieving maximum speed at the end of the straight. If you gear too high you loose acceleration out of the corners and may not gain a lot in top speed. If you gear based solely on motor temp that can easily happen.
I attached the acceleration, speed and distance traveled graphs that illustrate the case John is talking about. The car model is not right I just used a TC model and set the gear ratio to the values John was talking about. You can see that the higher gear ratio has better initial acceleration and both cars reach the same speed after about 2 seconds. After that point the lower gearing will start to pay off assuming you have a very long straight.
I try and gear so speed is pretty much maxed out by the end of the longest straight. Usually motor temp is OK with this method. If you are still accelerating at the end of the straight then you may be over geared. Motor temp is only a guide to gearing and in my view should not be the only consideration.
Generating top speed is exactly opposite to torque. Lower gear ratios will provide higher top speeds until you reach the point where the power output of the motor cannot overcome air resistance, drivetrain friction, rolling resistance and so on. You need a long straight to reach this point.
What we want for on road is the best compromise between the two. Good acceleration out of the corners while achieving maximum speed at the end of the straight. If you gear too high you loose acceleration out of the corners and may not gain a lot in top speed. If you gear based solely on motor temp that can easily happen.
I attached the acceleration, speed and distance traveled graphs that illustrate the case John is talking about. The car model is not right I just used a TC model and set the gear ratio to the values John was talking about. You can see that the higher gear ratio has better initial acceleration and both cars reach the same speed after about 2 seconds. After that point the lower gearing will start to pay off assuming you have a very long straight.
I try and gear so speed is pretty much maxed out by the end of the longest straight. Usually motor temp is OK with this method. If you are still accelerating at the end of the straight then you may be over geared. Motor temp is only a guide to gearing and in my view should not be the only consideration.
07-05-2009, 01:39 PM
#190
Tech Adept
The graphs look right based on a cursory review. I agree that as the speed reducing gear ratio, G, goes down the top speed in a frictionless system goes up, assuming a constant tire radius r.
But in a system with friction in the driveline and air drag, I think there is a sweet spot in the gear ratio, such that going to a lower G beyond this best ratio would result in a reduction in top speed on an infinite run, i.e., on the salt flats. This is a problem of mechanical power transfer similar to the well studied electrical system theory of power transfer and impedance matching.
John's data seems to bear this out with 48{mph} being greater than 46{mph} at a gear ratio that, without friction, should predict a lower top speed. John is getting more speed with the G going higher, not lower, unless I am interpreting it wrong?
The rotating shaft of the motor is a source of torque, and the slope of the torque-speed line is equivalent to an internal source resistance, although some of this is due to motor shaft damping Dm, some is due to the linear air gap constant Kv generating reverse voltage at higher speeds, and non-linearities are attributed to magnetic circuit hysteresis, windage at high rpm, etc. Thus there is a mechanical impedance matching problem of the source to the load through the gearbox, which behaves like a load impedance transformer. This should create a sweet spot or optimum gear ratio producing peak velocity at the load (load voltage maximization in an electrical circuit-analog, since voltage is an analog for speed under the torque-current analog).
http://en.wikipedia.org/wiki/Maximum_power_theorem
http://en.wikipedia.org/wiki/Impedance_matching
As I said it will take some thought to sort this through, and I am always open to being set straight via a compelling derivation to the contrary.
But in a system with friction in the driveline and air drag, I think there is a sweet spot in the gear ratio, such that going to a lower G beyond this best ratio would result in a reduction in top speed on an infinite run, i.e., on the salt flats. This is a problem of mechanical power transfer similar to the well studied electrical system theory of power transfer and impedance matching.
John's data seems to bear this out with 48{mph} being greater than 46{mph} at a gear ratio that, without friction, should predict a lower top speed. John is getting more speed with the G going higher, not lower, unless I am interpreting it wrong?
The rotating shaft of the motor is a source of torque, and the slope of the torque-speed line is equivalent to an internal source resistance, although some of this is due to motor shaft damping Dm, some is due to the linear air gap constant Kv generating reverse voltage at higher speeds, and non-linearities are attributed to magnetic circuit hysteresis, windage at high rpm, etc. Thus there is a mechanical impedance matching problem of the source to the load through the gearbox, which behaves like a load impedance transformer. This should create a sweet spot or optimum gear ratio producing peak velocity at the load (load voltage maximization in an electrical circuit-analog, since voltage is an analog for speed under the torque-current analog).
http://en.wikipedia.org/wiki/Maximum_power_theorem
http://en.wikipedia.org/wiki/Impedance_matching
As I said it will take some thought to sort this through, and I am always open to being set straight via a compelling derivation to the contrary.
Last edited by SystemTheory; 07-05-2009 at 01:43 PM. Reason: never can remember how to spell "hysteresis"
07-06-2009, 01:35 PM
#191
Tech Adept
Brute Force Maximum Velocity Gearbox Study
The attached graphic is a load line analysis to predict top speed for a wide range of installed gear ratios. There is a sweet spot in terms of "salt flat" gearing, but note, this does not include a thermal model (overheat) analysis!
Apply Newton's Second Law and D'Alembert's Principle. This gives the law known as Conservation of Dynamic Force, stated: the sum of forces acting at a point minus the mass times acceleration of a body equals zero.
Maximum velocity occurs when drive force F is equal and opposite to the sum of resistive forces, R = B + D, as shown in the continuous time velocity feedback block diagram (top right).
The equivalent rolling mass mR includes mass plus inertia seen at the rolling axles, converted via tire radius r into some additional equivalent mass, but this does not appear in the code model because top speed is the steady state solution, and mass only effects the transient response (acceleration rate, braking rate), not the steady state boundary condition of top speed.
A velocity array is created as an independent variable, from 0 to 65+{mph}, to generate the velocity feedback dynamic load resistance R = D + B, here shown in Newtons.
Drag Force: 1/10 Scale Dodge Viper, Sea Level Static Air
c = 0.5*rho*Cd*Af (drag feedback system constant in SI units)
D = c*v^2 (v in meters per second, D in Newtons)
Driveline Resistance Force (See ERROR? - footnote/Edit)
DL = 3.0E-3 {N-m-s} estimated driveline damping
b = DL/r (r is tire radius in meters)
B = b*v (estimated driveline feedback resistive force in Newtons)
Resistive Load Force
R = B + D
Generic 19T Brush Motor Torque-Speed Line
Ts - (Ts/wf)*w (source of torque starting at Ts, velocity feedback -Ts/wf)
Ts - start/stall torque at full voltage in Newton-meters
wf - final shaft speed on Dyno or per model in radian per second
w - angular motor shaft speed in radians per second
Gearbox and Tire Radius Act as a Source Transformer
Fs - (G^2/r^2)*(Ts/wf)*v (source of axle force starting at Fs, slope S)
S = - (G^2/r^2)*(Ts/wf) (G is gear ratio, r is tire radius)
When you change the installed gear ratio G, the starting force Fs = Ts*(G/r) changes, and so does the slope of the axle force curve S proportional to G^2.
My brute force study assumes parameters for the car that determine dynamic resistance R from the chassis velocity independent of the installed power plant. The power plant has a Dyno torque-speed line at maximum voltage.
I let G increase over a reasonable range in increments of 0.01, and find the velocity where F = R and Net force = 0. This is maximum velocity for that G.
I plot all the velocities in the final graph against the changing G on the horizontal, and a maximum velocity is predicted at the sweet spot value of G.
In practice, this may exceed a thermal limit of the motor, although it looks close to the data for set up at the Velodrome. Most of the time one would gear for a cooler motor, particularly with a lot of infield, and going to G lower tends to bog down the acceleration (reduce wheel spin) but warm the motor, so it might be OK, as John indicates, on a very slick track, but maybe there is a better solution (take voltage off the low rpm range on the ESC profile).
I've tried to explain a pretty complex study in clear terms. If anyone has questions, fire away.
Apply Newton's Second Law and D'Alembert's Principle. This gives the law known as Conservation of Dynamic Force, stated: the sum of forces acting at a point minus the mass times acceleration of a body equals zero.
Maximum velocity occurs when drive force F is equal and opposite to the sum of resistive forces, R = B + D, as shown in the continuous time velocity feedback block diagram (top right).
The equivalent rolling mass mR includes mass plus inertia seen at the rolling axles, converted via tire radius r into some additional equivalent mass, but this does not appear in the code model because top speed is the steady state solution, and mass only effects the transient response (acceleration rate, braking rate), not the steady state boundary condition of top speed.
A velocity array is created as an independent variable, from 0 to 65+{mph}, to generate the velocity feedback dynamic load resistance R = D + B, here shown in Newtons.
Drag Force: 1/10 Scale Dodge Viper, Sea Level Static Air
c = 0.5*rho*Cd*Af (drag feedback system constant in SI units)
D = c*v^2 (v in meters per second, D in Newtons)
Driveline Resistance Force (See ERROR? - footnote/Edit)
DL = 3.0E-3 {N-m-s} estimated driveline damping
b = DL/r (r is tire radius in meters)
B = b*v (estimated driveline feedback resistive force in Newtons)
Resistive Load Force
R = B + D
Generic 19T Brush Motor Torque-Speed Line
Ts - (Ts/wf)*w (source of torque starting at Ts, velocity feedback -Ts/wf)
Ts - start/stall torque at full voltage in Newton-meters
wf - final shaft speed on Dyno or per model in radian per second
w - angular motor shaft speed in radians per second
Gearbox and Tire Radius Act as a Source Transformer
Fs - (G^2/r^2)*(Ts/wf)*v (source of axle force starting at Fs, slope S)
S = - (G^2/r^2)*(Ts/wf) (G is gear ratio, r is tire radius)
When you change the installed gear ratio G, the starting force Fs = Ts*(G/r) changes, and so does the slope of the axle force curve S proportional to G^2.
My brute force study assumes parameters for the car that determine dynamic resistance R from the chassis velocity independent of the installed power plant. The power plant has a Dyno torque-speed line at maximum voltage.
I let G increase over a reasonable range in increments of 0.01, and find the velocity where F = R and Net force = 0. This is maximum velocity for that G.
I plot all the velocities in the final graph against the changing G on the horizontal, and a maximum velocity is predicted at the sweet spot value of G.
In practice, this may exceed a thermal limit of the motor, although it looks close to the data for set up at the Velodrome. Most of the time one would gear for a cooler motor, particularly with a lot of infield, and going to G lower tends to bog down the acceleration (reduce wheel spin) but warm the motor, so it might be OK, as John indicates, on a very slick track, but maybe there is a better solution (take voltage off the low rpm range on the ESC profile).
I've tried to explain a pretty complex study in clear terms. If anyone has questions, fire away.
Last edited by SystemTheory; 07-06-2009 at 07:18 PM. Reason: ERROR?: apparently b = DL/r^2, not DL/r, to get B{N}. Some units, numbers would change. Principle of this study is correct.
07-07-2009, 05:38 PM
#192
Tech Adept
Brute Force Gearbox Study - Corrected
The graphic shows corrections to my code in the left panel, and an additional block of code to plot John's ratios against my estimated 1/10 car parameters.
This model suggests the opposite of John's report: the 2{mph} gain in top speed comes at a higher gear (lower G), as one usually expects, rather than at a lower gear (higher G), as I think John reports above.
This discrepancy could be explained if John's system permits more velocity at these ratios, and if he is reading the speed gun on a shorter acceleration run.
Then the car could accelerate to higher velocity on the short run by using a lower gear, in accord with Bob's explaination.
Anyway in the interest of clarity I post the corrections.
This model suggests the opposite of John's report: the 2{mph} gain in top speed comes at a higher gear (lower G), as one usually expects, rather than at a lower gear (higher G), as I think John reports above.
This discrepancy could be explained if John's system permits more velocity at these ratios, and if he is reading the speed gun on a shorter acceleration run.
Then the car could accelerate to higher velocity on the short run by using a lower gear, in accord with Bob's explaination.
Anyway in the interest of clarity I post the corrections.
07-09-2009, 10:42 AM
#193
Tech Rookie
What we want for on road is the best compromise between the two. Good acceleration out of the corners while achieving maximum speed at the end of the straight. If you gear too high you loose acceleration out of the corners and may not gain a lot in top speed. If you gear based solely on motor temp that can easily happen.
Even if it is an even match, you'll be passing people primarily by out-accelerating them at the slower speeds, where it's easier to go around people or take a line that is not quite optimal. Someone who has his car set up to pass you on a long straight will have to do so at high speed, which is harder or more risky. Even if you are constantly passing each other throughout the race, you are taking less risk in doing so than he is.
"This discrepancy could be explained if John's system permits more velocity at these ratios, and if he is reading the speed gun on a shorter acceleration run."
Don't forget about the error of the speed gun. They aren't that accurate... if a speed gun was used to measure the 46mph vs. 48mph difference... is that even a meaningful measurement? If you want truly accurate speed measurements you have to use a speed trap or something similar.
07-09-2009, 01:55 PM
#194
Tech Adept
Access,
I agree with your comments. I think +/- 1{mph} was the resolution of one speed gun specification I read, but I can't recall the make and model. This is obviously a bigger percentage error at low speeds.
Acceleration runs are best conducted with a tape measure and speed trap, or stop watch if automatic triggers are not available. I think human error on the stop watch is relevant if you're trying to measure 0.1{s} improvement.
My cursory study of NASCAR engineering is that a car can be geared for acceleration in the passing points or overtaking points on a given track, very similar to your description. You don't need a better gear in places where it is easy to block the car behind, so gearing can become very strategic. It is the link between tire tests, traction conditions, track strategy, and Dyno curves.
John is very good with measurements and observations that make sense in a racing context. I'm sure if it doesn't help bring down lap times, he'll question its utility, as you suggest is reasonable.
I agree with your comments. I think +/- 1{mph} was the resolution of one speed gun specification I read, but I can't recall the make and model. This is obviously a bigger percentage error at low speeds.
Acceleration runs are best conducted with a tape measure and speed trap, or stop watch if automatic triggers are not available. I think human error on the stop watch is relevant if you're trying to measure 0.1{s} improvement.
My cursory study of NASCAR engineering is that a car can be geared for acceleration in the passing points or overtaking points on a given track, very similar to your description. You don't need a better gear in places where it is easy to block the car behind, so gearing can become very strategic. It is the link between tire tests, traction conditions, track strategy, and Dyno curves.
John is very good with measurements and observations that make sense in a racing context. I'm sure if it doesn't help bring down lap times, he'll question its utility, as you suggest is reasonable.
07-09-2009, 08:00 PM
#195
Speed guns are very reproducible. The actual value may be off a bit, but lap after lap they record the same number. This means also that the cars top speed on the straight is quite reproducible in spite of varying actual paths taken by the car on each lap.
Stop watch accuracy is much better than .1 seconds. We have discussed this recently. The human is able to use anticipation to hit the stop watch button when the car crosses a line. There is no reaction time involved like when a dragster driver sits at a pro tree and a light flashes to green.
Some time in the future I will be gathering Sentry data with the two gears I tried. Interestingly at the moment I am geared right in between.
Thanks for the discussions.
Stop watch accuracy is much better than .1 seconds. We have discussed this recently. The human is able to use anticipation to hit the stop watch button when the car crosses a line. There is no reaction time involved like when a dragster driver sits at a pro tree and a light flashes to green.
Some time in the future I will be gathering Sentry data with the two gears I tried. Interestingly at the moment I am geared right in between.
Thanks for the discussions.