

03072009, 02:38 PM

#46

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

Here is the status of the spreadsheet. I think Mattnin and I agree more or less now. I have input the momentum of the new flywheel. This is the same dyno run that Mattnin posted up above.
Note amps and efficiency are crap as we don't have the amps sensor hooked up yet.
I have also include the graph Torque vs Angular momentum if it is of interest.
I think there may be some torque control evident on this middle output when compared to the output I showed for the Fantom earlier (3rd across). I suffered the same with an LRP speed control that was not set to maximum output.
john



03082009, 01:44 PM

#47

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

BLDC vs PMDC Efficiency
Fred questions efficiency for BLDC on the PMDC dyno, so here are some thoughts.
Efficiency: eta = (Pm/Pe)*100
Pm  mechanical power output in watts
Pe  electrical power input in watts
PMDC motor: Pe = Vt*Ia
Vt  terminal voltage
Ia  armature current
During a Dyno pull the time curves for Pe and Pm are similar to the attached sketch. The area under the RED line is the energy depleted from the battery, and the area under the BLUE line is the kinetic energy invested in the car at greater velocity. Efficiency of the motor is not as important as its ability to accelerate parts up to greater velocity while taking less energy from the battery, which occurs over the whole curve, because Racing does not occur at the maximum efficiency point of any motor.
BLDC motor: Pe = three phase power measurement
If the BLDC operates similar to the PMDC torquespeed curve, it makes power similar to the curve shown in BLUE. However there is no DC current to sample in the 3phase lines. I suspect a twowattmeter arrangement is required to measure the electrical power input as shown in the second attachment.



03082009, 02:17 PM

#48

Tech Fanatic
Join Date: Jun 2006
Posts: 872

I have the current sensor in finally, and I will be running a test of the hacker motor later this evening and even possibly the Novak 13.5 for comparison. The test parameters will be 0 deg boost, fully topped battery, and 3 full dyno tests in one file, includes voltage and current so efficiency data will be available. Then John and I are going to take the raw data from the sentry, run it through the spreadsheet, and set up all graphs. We are close to having a final spreadsheet for the Sentry dyno. Soon, we will post it for download and a howto use the spreadsheet in case you want to do this yourself.
__________________
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



03092009, 05:08 AM

#49

Tech Addict
Join Date: Mar 2008
Posts: 649

Quote:
Originally Posted by SystemTheory
Fred questions efficiency for BLDC on the PMDC dyno, so here are some thoughts.
If the BLDC operates similar to the PMDC torquespeed curve, it makes power similar to the curve shown in BLUE. However there is no DC current to sample in the 3phase lines. I suspect a twowattmeter arrangement is required to measure the electrical power input as shown in the second attachment.

When I ran, the current was measured at the speedo inputs and not the phase wires. That's another reason why I questioned the high efficiency numbers that I got on the runs.
As you said, measuring current at the motor would be a little difficult for what we have to work with.
The runs were still great for comparison but they were not correct.
I'm interested in playing with the spreadsheet a little. Might just be able to get rid of the old Turbodyno.



03092009, 07:23 PM

#50

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

Electrical power output from the battery can be measured using a True RMS Wattmeter to find Pe.
Attached is a block diagram for a four quadrant meter, with voltage and current sampling, continous integration, then A to D conversion. This can measure actual DC, pulse width modulated DC, and the RMS power output from a battery into the BLDC motor controller.
The Power Analyzer Pro by Medusa Research appears to have a wattmeter and other functions necessary to build a BLDC dyno by adapting a flywheel, rpm sensor, and calculations per this thread, in case anyone's interested in measuring efficiency accurately in a BLDC. I don't know the sample frequency, however.



03112009, 11:02 PM

#51

Tech Fanatic
Join Date: Jun 2006
Posts: 872

The Sentry Dyno Calc spreadsheet is almost completed. Here you go guys, big preview. Be sure to give John a lot of kudos for this, he is a mathematical genius!
http://arcadechamp.net/radio/Hacker,...2%20x%202.html
__________________
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



03122009, 07:16 PM

#52

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

Quote:
Originally Posted by John Stranahan
Now if Joe can tell me how they get away with that huge gearing on a 17.5 on the banked oval. I would be enlightened. I would think that gearing the motor to produce max power at max speed would be the way to go. It was for my flat oval 13.5 car. I have a 17.5 in hand now.
John

I'm curious about this comment. Do you mean max power at max speed to push air and overcome driveline friction (top gear as on a salt flat)? If not, how do you identify max speed for the given track?
The Dyno curve looks good, and I notice predicted alpha and omega versus measured alpha and omega, so I assume John is applying some correction factor to the sampled data, particularly when rpm are low and the sensor is not accruate. Correct?



03122009, 07:32 PM

#53

Tech Fanatic
Join Date: Jun 2006
Posts: 872

Here is data comparing the Novak 13.5 vs the Hacker 13.5. This is not the final spreadsheet yet, but the #'s are accurate.
Hacker 13.5
http://arcadechamp.net/radio/Hacker,...try%20Data.pdf
Novak 13.5
http://arcadechamp.net/radio/Novak13...try%20Data.pdf
One thing that is readily apparent is that the claimed Hacker wattage ratings are way overinflated. Hacker claims peak wattage of 370W. Maybe if the calculations were done like I showed previously which calculates torque based on a RPM reading with the tangent of the acceleration taken at a midpoint you will get those high numbers. But to do it right like John has shown, find the tangent of the acceleration curve AT the RPM reading and NOT the midpoint to calculate torque.
Seems like Novak's published power ratings are slightly lower than what I was able to see. Novak says their 13.5 produces only 165W of power. The Novak does seem to have a wider powerband than that of the Hacker, although the Hacker has a quite a bit more torque.
Also, further analyzing the graph, the Novak seems to have a much smoother efficiency curve. I don't know what that means however. I do know that the Novak in the same setup will run cooler, and the Hacker gets quite warm.
__________________
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



03122009, 08:00 PM

#54

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

Quote:
Originally Posted by SystemTheory
I'm curious about this comment. Do you mean max power at max speed to push air and overcome driveline friction (top gear as on a salt flat)? If not, how do you identify max speed for the given track?
The Dyno curve looks good, and I notice predicted alpha and omega versus measured alpha and omega, so I assume John is applying some correction factor to the sampled data, particularly when rpm are low and the sensor is not accruate. Correct?

Joe and I have run some oval. We can easily calculate average speed. On the velodrome the average speed is probably very close to top speed. On my flat asphalt there is some variance, but not a whole lot when running 13.5 four cell. So the question is how to gear the beast. This is what I found. Oval guys with 15 years more oval experience than me were gearing there 13.5/ four cell 5 to 6 tooth higher than me. Their motors were coming off like a frying pan especially if we ran an afternoon race. 210230 degrees. Mine was coming off at 140150. Our fast laps could be distinguished by a tooth pick length probably. Then I dynoed that 13.5. I was geared right for maximum power at maximum speed. This is what you do on the salt flats. I don't know all the factors that oval guys use to gear the motor, but definitely there were few walks to the timing sheets. I think they are going for more of a dead feel all around the track. I want to be informed on this so I can write about it more intelligently. That mysterious upturn in the Sentry data near the end of the run caused me to wonder, but it was an artifact of a poor fit to my model when an extended run is made.
The other answer is yes the Linear Regression smooths the response data very well. The very first data derived from RPM is the angular velocity; this is smoothed by the model. The model can then predict angular velocities (RPM's) in those places where the Sentry hiccupped. This number also tells me how good the model is fitting angular velocity data. Measured and predicted values should be close. There are other ways as well. Then a derivative (using Calculus) of this model can accurately calculate Acceleration at each recorded RPM number. This little trick right here is responsible for the improvement over a point to point approach. the model I used is totally empiricle. No motor theory involved. It fits the data very well though. I am pleased with the result and await data from a 17.5. Then we can talk gearing some more.
I think we still need to add the armature inertia. I will work up a number tomorrow.
John



03122009, 08:46 PM

#55

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

John, once again I value your description of the gearing and the linear regression technique. In my experience with Automatic Test Engineering, one must triangulate between the expected system model, the measurements, and the numerical methods applied, or errors multiply everywhere. Statistics and calculus are theoretical tools. Newton invented the derivative to relate velocity to acceleration, originally, and the blasted bankers extended this concept to the differential equations for controlling the Money Supply
The reason banking blows up is positive feedback in the extension of credit during a bubble, and positive feedback in credit contraction during a bust, so the whole system is engineered for periodic failure (maybe psychology limits this but I suspect a better system could be invented some day).
I may put together some brushless motor models from your data, and see what my software Dyno comes up with. If I do, I'll post results. What estimate are you using for the rotor inertia in each machine?
I need to think through the gearing issue for further discussion, because at this hour it will just be noise, and I'm likely to go off on another tangent ...



03122009, 11:01 PM

#56

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

"In my experience with Automatic Test Engineering, one must triangulate between the expected system model, the measurements, and the numerical methods applied"
?
It's easy to tell if our mathematical model is accurate. You just plot the models predicted values over the experimental data points. If the models result almost goes right through the middle of each data point then you have a good empirical model. If the correlation coefficient R^2 comes out 1.0 or close to it you have a good model. If the list of deviations from the experimental data points to the predicted data points are small you have a good model. I have looked at all of these. We have not discussed previously all of these. Error in modeling does not creep into this process, it is plain to see in these outputs. 95% confidence bands on the power curve can be plotted from the excel results. It is not really neccesary for our purpose. I think the power numbers are as good as can be obtained from this dyno. The spreadsheet will also work with the Fantom dyno or Robitronic dyno. Now later we can go ahead and run the 3 runs through the dyno and see the repoducibility. That is most critical. I think this power number is much better than you can calculate with only a piece of paper and calculator. I have seen plenty of bad examples of that.
The derivative is a simple mathematical tool. Not just theoretical. If you take the derivative of velocity you get acceleration like you mentioned. It is right handy having a formula for the motors angular acceleration and angular velocity. It gives us an accurate value of torque and then power.
Torque = Momentum x angular acceleration
Power = Torque x angular velocity
If you do a point to point set of calculation which I have done and plot the results, every weakness in the Sentry data is there to see. The results are randomly high and low of a smooth line. The linear regression smooths this out by minimizing deviations between a smooth line (the model) and the data points.
This discussion might not help but it delays the game slightly. Matt is under the weather. If I can zip that excell file I can post it for download. I'll see.
Here is the file in excel. I bundled two files so it would be zipped. The file ending in .xls is the latest spreadsheet. Feel free to save it first with a new name and then tinker with the new one. The way it will work is first you will copy and paste (or type, god forbid) your data onto page one in the green area. Then you will delete any unused rows. Then save. Then reload and the calculations and graphs should be updated and visible on page two. Some instructions like this paragraph will eventually come with it. Some knowlege of excel or Open Office (file to come later) will be needed to copy and paste in the data. Not too hard. Instructions will be there.
This is more or less to give you our state of progress. The finished version should come in a few days or so.
John
Last edited by John Stranahan; 03132009 at 11:12 AM.



03132009, 07:51 AM

#57

Tech Fanatic
Join Date: Jun 2006
Posts: 872

__________________
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, 08:46 AM

#58

Tech Fanatic
Join Date: Jun 2006
Posts: 872

__________________
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
Last edited by mattnin; 03132009 at 09:27 AM.



03132009, 08:59 AM

#59

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

John I don't doubt your statistical analysis skill exceeds mine. If I recall the title of my course in this stuff, it was Random Variables and Probability Theory. If it is not a theory then I wonder why the experts in the field call this a theory?
This attachment is a sketch of the No Load Motor Response, or Dyno test setup. The model can be built into a multildomain system simulator, or you can convert the mechanical system to an electrical analog, and run it in a circuit simulator (SPICE). The motor circuit does not include the armature inductance (La), since this is more significant in designing the electronic speed control then it is in predicting the mechanical response at DC voltage level Vs.
The strength of this approach is that one can change system parameters in silicon and gain insight into changes in the "Dyno" curves. Particularly in a Racing scenario the current in the motor can be studied under various conditions, and since current is proportional to torque and heat in the coils, it helps one begin to understand how the Dyno curves translate to track performance based on gearing changes, etc. All the literature I've read on Control Systems with a PMDC motor plant starts with this type of model and gets more complicated from there ...
Comment: Armature current Ia begins at a maximum and decays exponentially over time due to the increase in backemf voltage Vbe with increasing shaft speed omega. Power dissipated as heat in the battery is Ia^2*Rs, and power dissipated in the motor coils is Ia^2*Ra. There are two kinds of mechanical load that reflect back to the motor circuit through the air gap, inertial load and frictional loss (drag, driveline friction). Gear selection will impact armature current by changing the (1) response time of the system; and (2) equilibrium point at which torque in the motor equals countertorque due to friction and drag.
Last edited by SystemTheory; 03132009 at 09:23 AM.
Reason: Attachment to gif; added SPICE model; comment



03132009, 11:05 AM

#60

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

MattThat motor comparison is really nice. Who would have thought you would get such nice output from the Sentry.
Fantom owners. There is a setting in the Fantom software to save the motor file in several formats. One of them is I believe ASCII. This one can be opened by excel and then a copy and paste made into the green area of the spreadsheet. If this is your plan there is a chance we could rearange the column order and make the spreadsheet more Fantom compatible. Make a post or PM if you would like this to happen. As of now I have arranged the columns so Sentry data can be inputed the most easily.
The Robitronic has a similar feature as I know I have plotted Robitronic results with excel.
System TheoryI can follow your electrical analog circuit. I can say the most similar thing I recall is the complete mathematical analog used to model the inerter suspension device for formula 1. Boomer supplied the article, I made and tested the device in an RC car and it actually worked. The report is in my CRC thread. Good luck!
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 01:20 PM.
