

03132009, 11:51 AM

#61

Tech Fanatic
Join Date: Jun 2006
Posts: 872

Novak 7.5 w/ sintered rotor, smooth can. I might have stopped a little early as I was getting a little nervous about the flywheel spin speed. Motor temp at end of run, 130deg.
Now I just have my own Novak 13.5 to compare to my friends Novak 13.5 and that is all I have, so I need to round up as many motors this Saturday that I possible can.
__________________
High Desert Raceplace, Grand Jct CO, Sanwa Exzes Plus Stick Radio, Spektrum,Thunderpower 230g, Modified TriNut Novak GTB2, Ballistic 4.5t 550, 5.83lb SCTE Ten, Novak Sentry Brushless Dyno, Crossweight Setup Station, Junsi 20A power supply Icharger 20A Charger,TP610C, 22B, Novak Edge, Novak 13.5, SC10, Havoc Pro SC XDrive, Ballistic 17.5 Matthew Joseph Cordova



03132009, 12:01 PM

#62

Tech Fanatic
Join Date: Jun 2006
Posts: 872

I guess since I don't have any more motors to test, I will run a bunch of tests with the RS Pro boost timing at different settings.
__________________
High Desert Raceplace, Grand Jct CO, Sanwa Exzes Plus Stick Radio, Spektrum,Thunderpower 230g, Modified TriNut Novak GTB2, Ballistic 4.5t 550, 5.83lb SCTE Ten, Novak Sentry Brushless Dyno, Crossweight Setup Station, Junsi 20A power supply Icharger 20A Charger,TP610C, 22B, Novak Edge, Novak 13.5, SC10, Havoc Pro SC XDrive, Ballistic 17.5 Matthew Joseph Cordova



03132009, 12:15 PM

#63

Tech Adept
Join Date: Mar 2009
Location: Metro New York
Posts: 139

Predicting Peak Power from the Engineering Model.
I took Matt's first Sentry Dyno post for the Hacker 13.5, the one with voltage and current listed in the spreadsheet.
Vs  open circuit source voltage  8.4 volts
Is  start/stall current  106 Amps
Rs  internal source resistance  0 Ohms
Ra  armature resistance  Ra = Vs/Is = 0.0792 Ohms
Knowing Vs and Ra, assuming Rs = 0, I apply a standard PMDC model (prefer to keep it to myself for now) formula to compute peak mechanical power output:
Pmax = 222 Watts
Now this compares with the Novak Dyno estimate of 215 Watts for this pull. That is less than 10% error and I did not account for the loss of power due to nonzero source resistance Rs in an actual battery. The calculation of Rs from the data set published may be possible, and I might even get the 215 Watts back exactly, which would be a good confirmation of the integrity of John's linear regression or maybe just an exercise in circular reasoning from the assumed data set, I can't decide which one yet.
Last edited by SystemTheory; 03132009 at 12:39 PM.
Reason: units error



03132009, 12:35 PM

#64

Tech Elite
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777

MattI think the max power box blew up. Check the power peak on your last post and the box result. Also if you have time. You have three runs on the same motor from your earliest work. We need max power numbers on those three runs. That will help us determine the precision of the measurement.
System Theory Nice to see we are in the same ballpark.
John



03132009, 01:30 PM

#65

Tech Fanatic
Join Date: Jun 2006
Posts: 872

John, I think the max box is working fine. Were you comparing max power to the torque axis?
Here is my own, newer novak 13.5 dyno'd at 0 deg and 30 deg timing, and compared. Notice my 13.5 at 0 deg to my friends 13.5 at 0 deg. The Novak at 30 boost revs almost to 30k rpms.
__________________
High Desert Raceplace, Grand Jct CO, Sanwa Exzes Plus Stick Radio, Spektrum,Thunderpower 230g, Modified TriNut Novak GTB2, Ballistic 4.5t 550, 5.83lb SCTE Ten, Novak Sentry Brushless Dyno, Crossweight Setup Station, Junsi 20A power supply Icharger 20A Charger,TP610C, 22B, Novak Edge, Novak 13.5, SC10, Havoc Pro SC XDrive, Ballistic 17.5 Matthew Joseph Cordova



03132009, 03:01 PM

#66

Tech Fanatic
Join Date: Jun 2006
Posts: 872

John, I think you will be pleasantly surprised. I compared a previous hacker 13.5 dyno test to the one I have posted. The previous test showed that odd hump at the end of the run too, albeit more pronounced.
Here is the original:
Here is a previous test:
__________________
High Desert Raceplace, Grand Jct CO, Sanwa Exzes Plus Stick Radio, Spektrum,Thunderpower 230g, Modified TriNut Novak GTB2, Ballistic 4.5t 550, 5.83lb SCTE Ten, Novak Sentry Brushless Dyno, Crossweight Setup Station, Junsi 20A power supply Icharger 20A Charger,TP610C, 22B, Novak Edge, Novak 13.5, SC10, Havoc Pro SC XDrive, Ballistic 17.5 Matthew Joseph Cordova



03132009, 04:11 PM

#67

Tech Elite
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777

Damn! Thats really nice. Good Reproducibility. Need that third test.
I was looking at the wrong axis. Max box is OK. In excel all my axes are on the left.
So boost, moves the power peak slightly to higher RPM. It would improve speed on a long straight on a road course at the expense of some loss of efficiency and heat build up.
john
Last edited by John Stranahan; 03132009 at 09:22 PM.



03142009, 08:30 AM

#68

Tech Elite
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777

Armature Rotational Inertia
.482 inch diameter 22.1 g
4.138 x 10^7 kgm^2
I will add this to the spreadsheet in a bit. It raised the max power on my example from 215 to 216 W. The extra heavy flywheel makes this number less significant.
Please check my math.
john



03142009, 09:12 AM

#69

Tech Adept
Join Date: Mar 2009
Location: Metro New York
Posts: 139

John, my equations and simulations predict no change in peak mechanical power due to a change in motor or load inertia, which I write Jeq = Jm + JL.
If you inspect the load line (torquespeed line) in my prior posts, if the input voltage is held constant, the midpoint of this line (peak power) does not change when the inertial load Jeq changes. What changes is the spinup time, or the time to roll through this peak point on the way to equilibrium torque. When motor torque Tm = TL, load torque, equilibrium occurs.
An ineritial load exerts zero countertorque TL at maximum angular velocity (it keeps spinning with constant angular momentum according to Newton's first law), so it should not change the final operating speed either, unless it puts some force on the motor bearings and this increases the angular damping loss. When you hang a flywheel on the output shaft it puts some stress on the bearings so top rpm might change a little due to that.
The voltage and current sensors are independent of the rpm sensor, so if the sample data are accurate, it is possible to use my peak power formula to help confirm the ballpark mechanical power in your spreadsheet model.
However it is difficult to estimate internal battery resistance from a regressive technique, so accuracy of the starting voltage and current would be absolutely critical to determining source resistance Rs. With the data from Matt's post of the Hacker this gives Rs = 0.006 Ohm, or about 1mOhm per cell for NiMH six cell stack (a bit low). Is it a high power Lipo? Also the Novak site says the current sensor sensitivity is +/ 0.5 Amp up to 100 Amp, so it is just above the rated sensitivity.
I get 205 Watts peak mechanical power trying to adjust for Rs, still within 10% of your 215 Watts and based on the independently sensed values in the electrical versus mechanical system.



03142009, 09:24 AM

#70

Tech Fanatic
Join Date: Jun 2006
Posts: 872

John, I'll include the rotor inertia in the spreadsheet. All my previous tests were done with just one run based off a topped battery, plus include no current in the data. I have been taking the 3rd run and running that through the dyno because for some reason, I get higher RPMs at the 3rd run. I don't know if it is because the LIPO is 'warmed up' or something, but the 1st run is a good 2k less rpms than the 3rd run.
SystemTheory, the battery I am using is a 5200mah Tenergy 25C lipo rated at 125A continuous.
__________________
High Desert Raceplace, Grand Jct CO, Sanwa Exzes Plus Stick Radio, Spektrum,Thunderpower 230g, Modified TriNut Novak GTB2, Ballistic 4.5t 550, 5.83lb SCTE Ten, Novak Sentry Brushless Dyno, Crossweight Setup Station, Junsi 20A power supply Icharger 20A Charger,TP610C, 22B, Novak Edge, Novak 13.5, SC10, Havoc Pro SC XDrive, Ballistic 17.5 Matthew Joseph Cordova



03142009, 09:55 AM

#71

Tech Elite
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777

System TheoryThats not exactly the math check I wanted.
Because we are measuring power by spinning up an flywheel, the total inertia spinning up is important in the power number we get. Note it is only about a 1 % change.
I agree that if the motor has reached constant speed the rotor inertia will not limit power.
Matt I do expect some change as the battery cannot be the same for each test. Thats why the three other dynos use a power supply to control motor input. Maybe two good runs is all that can be expected without recharging.
john



03142009, 11:45 AM

#72

Tech Adept
Join Date: Mar 2009
Location: Metro New York
Posts: 139

John,
I realize ineritia J (or I, if you're a physicist) appears in your power formula, and I think so does the shaft speed difference deltaomega between sample times? Thus if J increases the acceleration rate, measured here as the deltaomega over a 0.1 second window, should decrease. In the continuous time system model, where sample time does not introduce calculation error, it is my understanding the peak mechanical power is limited by the battery circuit and design of the motor air gap, and does not change with inertia J, except that when the motor is designed, a good one makes more power with less Jm then a less optimized machine. That is to say, your peak power should have an expected error introduced by the long sample times, and I'm not sure data smoothing reduces this error, but I don't think actual power is effected at all by changes in Jeq.
Matt there are a number of factors related to heat that could add up to 2k extra rpm on the third pull. Probably a warmer, slightly weaker battery should reduce both the torque and speed, because voltage goes down slightly and resistance goes up at greater states of discharge. I am not an expert on lubrication, but warm bearings might have less damping, and this lets the motor rev to a greater rpm before torque equilibrium Tm = TL caused by the countertorque in the bearings. That is where you're getting max rpms in the Dyno.
In addition the steepness of the torquespeed line changes with temperature. In a good reference I found online, the steepness of the line is a function of two engineering parameters Ra and k, where k is the air gap constant in voltsecond/radian. According to this snippet from the motor expert below, a warm motor might start with less torque but gain more speed based on an increase in Ra and a decrease in k:
Quote:
A complication is that the resistance R rises with α=4 % per 10°K and that the motor constant K falls with some percent per 10°K by the temperature dependence of the magnet strength. Usually symbol for the relative decrease of the motor constant is kt [%/K]. Concerning this last dependency one has to refer to the motor data sheets, where one also will find a usual tolerance on the resistance R and motorconstant K of 5 to 10 %.

This reference also says that a better motor always has a steeper slope of starting torque over maximum speed, that is S = Ts/wmax is greater for a better motor. That means the cold motor with a fresh battery is probably coming up as the "better motor" on your first pull, and the hotter motor with a less fresh battery is coming up with a little degradation, but you have the data and I am just offering ideas based on studying one expert's advice on motor specification for control system engineers.
Last edited by SystemTheory; 03162009 at 05:45 PM.
Reason: typos; logical error corrected



03142009, 03:08 PM

#73

Tech Elite
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777

"think so does the shaft speed difference deltaomega between sample times?"
I am going to outline our Power calculation method as this is precisely one of things that I tried to eliminate.
RPM is taken at discreet data points by the sentry. It turns out the RPM values are discreet as well and selected from a chart rather than calculated individually. The sampling process and selection from a chart causes some of these values to be high and some low of the actual value.
We change RPM to angular velocity (just changing units here)
We fit a polynomial model to the angular velocity data. The fit is good. Now we have an equation that describes the angular velocity vs time. No longer are we concerned with delta (value)/delta t. We can calculate pretty exactly the angular velocity at each time point. We get instantaneous angular velocities.
We take the derivative (using calculus) of this equation. Now we can calculate the angular acceleration. We get the instantaneous angular acceleration. No delta W/delta t is involved. We are working with instantaneous velocity and angular acceleration now.
We use two formulas more
T= I x angular Acceleration
The inertia from the armature is indistinguishable from the inertia of the flywheel in this spin up test. You must add it.
If you do a steady state test like the Competition Electronics dyno that measures torque with a sensor then you do not need to add or be concerned with angular inertia.
One last formula
P=Torque x angular velocity
In each case we have instantaneous values from our fitted equation and its derivative. This is responsible for a small improvement in accuracy and smoothness.
There was once a Tekin Dyno. It calculated power from simply an RPM drop as load increased from 10 to 20 amps. It agreed with no other dyno. It came with a three page paper explaining the theory. As far as I could tell the theory was just smoke and mirrors.
I can actually follow theory if care is taken in its presentation. All the pieces have to be there. Acronyms should be described on first use in a document or just not used. Units are important. A person should not assume the other person has the same reference in his possesion or have been through the same course. I have trouble with your stuff SystemTheory. I have trouble with you getting a power number but using test values from Matt's Dyno run like resistances that are unavailable unless you do the dyno run. If you are doing a dyno run anyway then power can be calculated with the Physics printed above.
John
Last edited by John Stranahan; 03142009 at 03:19 PM.



03142009, 07:20 PM

#74

Tech Adept
Join Date: Mar 2009
Location: Metro New York
Posts: 139

John, I just "lost" a very long post, carefully explaining the numbers and units behind my peak power estimates! Luckily I've been through Anger Management courses, and failed miserably
I see the logic behind your peak power calculation and respect the integrity of this approach. Instead of hit you with my calculations and comments, let me present the model in the attached SPICE circuit. This model solves the coupled differential equations:
Tm = k*Ia = Jm*(dw/dt) + br*w (1) feedforward equation
Ia = (Vs  k*w)/(Rs + Ra) (2) feedback equation reducing Ia
Tm  motor torque generated in the air gap
k  air gap constant expressed in SI units
Ia  armature current flowing out of the battery into ESC
Jm  moment of inertia of motor plus flywheel
dw  infinitesimal change in angular velocity
dt  infinitesimal change in time
br  bearing damping assumed to generate a linear feedback torque proportional to shaft speed
w  angular velocity at an instant of time
Vs  source voltage
Rs  source resistance
Ra  armature resistance
Vbe  backemf voltage developed by generator coils in the air gap, Vbe = k*w
All units are Standard SI units, this sets k to the same value in the feedforward and feedback sources.
When I run this model, holding Vs, Rs, Ra, k, and br constant, then probe for the peak mechanical power in the "flywheel capacitor analog," this peak value does not change when I increase or decrease capacitance/moment of inertia Jm, it remains constant but the peak comes sooner or later in time. If I increase the damping loss br, this becomes a smaller resistor 1/br, it absorbs more torque and shaves power off the peak on the capacitor. Sorry if this is not clear enough, I'll try to explain better if you have questions.
PS  The power curve in the rotor/capacitor is always shaped like the sketch in BLUE.
PPS  By PMDC, I mean a brush motor model, but with a straight line torquespeed curve sampled on the Dyno, and smaller bearing losses (no brush friction), it appears the brushless BLDC motor behaves much like a brush motor in the dynamic system response to a step voltage (during a "Dyno pull").
Last edited by SystemTheory; 03142009 at 07:59 PM.
Reason: insert Vbe = k*w; add power curve sketch



03142009, 08:57 PM

#75

Tech Elite
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777

Ok. Thanks. Now I have something to study. I will take a look. At one time I used to do a ctrlC just before submitting a post to copy it into memory. I know the agony.
Here are some tips for all dyno users that I have learned through experience.
We would like repeated dyno runs to be within 2 Watts on repeated runs. Improvements are of this small size especially with stock motors.
soldering on leads heats a motor and causes low power.
Wait 5 minutes after soldering or make sure that two comparison motors are tested near the same wait time after soldering. Forget about using clips.
A motor test, heats the motor. Discard the first test. Do repeated tests every 5 minutes on the clock. This gives the motor the same time to cool and takes away that first high reading. Look out for "time trends" Power keeps going down with repeated test. This could be from heating (or with brushed motors from deterioration)
Use the same ambient temperature. Motors test better in the cold. I do my testing indoors at 70 F. Otherwise results don't compare.
And finally a safety point. Put a scattershield on that beast if you are using a flywheel at 7.4 V. I think we are reaching the limit of strength of the 1/8 inch shaft when spinning up a 3.5 motor with the tiniest imbalance. Matt noticed this already with the 7.5. I had a 3.5 really screaming on the Fantom with 7.4 volts.
john



Similar Threads

Thread 
Thread Starter 
Forum 
Replies 
Last Post 
CRC Battle Axe, GenXPro 10, 1/10th pan, Brushless, Lipo,4c, Road, Oval,TipsandTricks

John Stranahan 
Electric OnRoad 
1776 
03012016 12:18 PM 
RC10B4/T4 Forum

DerekL 
Electric OffRoad 
14331 
01012016 02:18 PM 
Dyno, Homemade, Using a Novak Sentry Data Logger, Continued, The Experimental Thread.

John Stranahan 
Electric OnRoad 
343 
06262014 04:44 AM 
Novak New Ballistic Motors

20 SMOKE 
Electric OffRoad 
11 
05112009 02:58 PM 
Idea to Use brushed motor dyno for brushless

JevUK 
Electric OnRoad 
45 
04062009 09:52 AM 
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)


Thread Tools 
Search this Thread 


Posting Rules

You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off



All times are GMT 7. It is currently 07:12 PM.
