R/C Tech Forums

Go Back   R/C Tech Forums > General Forums > Electric On-Road

Reply
 
Thread Tools Search this Thread
Old 03-30-2009, 12:52 PM   #106
Tech Adept
 
Join Date: Mar 2009
Location: Metro New York
Posts: 139
Default

I found this post from a forum about Brushless dyno back in 2005:

http://forums.radiocontrolzone.com/a.../t-210520.html

Quote:
We at Novak built a very basic dyno when we first started designing the Super Sport System and still use it today to compare our brushless motors with brushed motors or other brushless motors. It basically is a flywheel dyno that measures the rpm and current from a standing start ( zero RPM) to full speed and will run a standard brushed motor thru a brushed speed control or a brushless motor thru a brushless speed control. The one problem that we have with it is when we try to dyno a sensorless brushless system. On startup all the sensorless brushless systems that we have tested will make the flyweel oscillate back and forth and will never get started. You have to manually spin the flywheel by hand to get it started. This is the main reason we decided that sensorless was not the way to go for cars. The heavier the load the worse the problem.
Bob Novak
I think the rpm sensor is rated for 500{rpm} and up, but its sensitivity is probably lower at lower rpm. The hiccup on start-up might be caused by instability in the electronic drive based on the information in the above post.

Last edited by SystemTheory; 03-30-2009 at 12:54 PM. Reason: added URL
SystemTheory is offline   Reply With Quote
Old 03-30-2009, 01:04 PM   #107
Tech Adept
 
Join Date: Mar 2009
Location: Metro New York
Posts: 139
Default

Matt, John, others:

Recently I took a very casual look at the brushless speed sensors, such as insert in 1 phase and detect the frequency using a zero-crossing detector. The sensitivity at low rpm is not assured.

It seems from the other thread these "load start-up" problems were well known in earlier brushless systems, and have not been fully resolved.

If the electronic speed control has trouble sensing low rpm then it might introduce oscillations during start-up of a heavy load, as you see in the data.

That would tend to confirm the accuracy of the rpm sensor if electronic commutation is isolated as the cause of oscillations.

Last edited by SystemTheory; 03-30-2009 at 01:31 PM. Reason: clarification
SystemTheory is offline   Reply With Quote
Old 03-30-2009, 07:07 PM   #108
Tech Elite
 
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777
Trader Rating: 27 (100%+)
Send a message via AIM to John Stranahan
Default

What sensorless speed controls have the most trouble with is knowing which way to start. Until the rotor has made a partial turn the sensorless speed control has no idea which way the rotor is is turning. In the RC car you will see the car run backwards then forward about half the time by chance on a standing start. This is very agravating on the track for me. This may be causing the problem if the hacker is sensorless. I will note that Novak (sensored)Motors on a Fantom dyno (used only for RPM) and a brushless LRP speed control do not have this initial problem.
John
John Stranahan is offline   Reply With Quote
Old 04-02-2009, 02:50 PM   #109
Tech Adept
 
Join Date: Mar 2009
Location: Metro New York
Posts: 139
Default

John and Matt:

In the absence of a voltage regulator with a high current pass transistor, it may be impossible to eliminate the battery parameters from impacting the measured flywheel power.

The attached plot shows theoretical terminal voltage VT pretty close to your measured values around peak power.

One option may be to use this sensed voltage around the peak power point (red oval) to calculate the peak watts per measured volt.

I don't know if doing a smooth of volts or amperes gains anything, but it might assist with your effort to "standardize" the measured value for use in the sensitive tuning context.

This is a "ballpark simulation" of the Hacker 13.5 data using approximate motor circuit model. The circuit diagram and relevant plots are attached.

:Battery Parameters --------
Vs = 8.4;
Rs = 0.012;
:Motor Parameters ---------
k = 0.003319;
Rm = (0.0741 - Rs);
Req = Rs+Rm;
Jm = 4.138E-07;
br = 1.505E-05;
:Flywheel Load ------------
JL = 3.45E-04;
:Pre Netlist Calculations -----
Jeq = Jm+JL;
Vbe = k*w;
Ia = (Vs-Vbe)/Req;
Tm = k*Ia;
alpha = (1/Jeq)*(br + k**2/Req);
tF = 5*(1/alpha);

WATTS PER SQUARE-VOLT (?)

Theoretically the peak air gap power is regulated terminal voltage squared, divided by motor resistance Rm, multiplied by 1/4. So it might be better to normalize the watts per volt-squared?
Attached Thumbnails
Novak Sentry as Brushless Dyno-hacker_13_5_ckt_diag.png   Novak Sentry as Brushless Dyno-hacker_13_5_ckt_plot.png  

Last edited by SystemTheory; 04-02-2009 at 03:05 PM. Reason: W/V^2 better?
SystemTheory is offline   Reply With Quote
Old 04-02-2009, 04:03 PM   #110
Tech Elite
 
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777
Trader Rating: 27 (100%+)
Send a message via AIM to John Stranahan
Default

Nice graph
Watts per Volt might be a useful output of the spreadsheet. It would be easy to include. For brushed motors the Power output of the motor vs volts is a straight line. I have dyno data to show this. No need to complicate things with square volts.


I had another note from Charlie (formerly from Novak). He had very stable rpm readings with the amp section pegged while running some 1/8 scale stuff that drew more than 108 most of the time. We need to look elswhere for the spurious results on the first few points. It may just be a property of non sensored motors and speed controls as we know they have trouble on start.

John
John Stranahan is offline   Reply With Quote
Old 04-03-2009, 12:44 PM   #111
Tech Adept
 
Join Date: Nov 2007
Location: Vine Grove, KY
Posts: 228
Trader Rating: 4 (100%+)
Default

It seems like you guys are on to something here... Is there any way you guys could tests some speed controllers, under different profiles and tuning options, and then post the results? It doesn't concern me as much if your machines are producing accurate numbers (although I'm sure they are), but precision is important and it would be great to know which speed controllers and which motors will work better in different scenarios. I'd love to see a 4.5, 8.5, 10.5, 13.5, and 17.5 tested under different controllers and settings and then everything plotted and put on a spreadsheet.

This is intuitive, but it looks like the timing "boost" on the Tekin units shifts the power band to the right and decreases torque a bit. How do other changes in speed controller affect performance. It would be great if you guys could do some testing and make these generalizations. Otherwise, I may have to copy your work and do the tests my self.

Well done with your machines and calculations!
richard.bratton is offline   Reply With Quote
Old 04-03-2009, 02:36 PM   #112
Tech Adept
 
Join Date: Mar 2009
Location: Metro New York
Posts: 139
Default

John,

In the ideal brush motor model both torque and speed get boosted by an increase in voltage, so I'm fairly certain power increases with voltage squared. My logic is presented symbolically below.

Nominal/Dyno Values
Vnom - nominal input voltage {V}
Ts - start/stall torque {N-m}
wIDL - ideal maximum velocity {rad/s}

Peak Nominal Power
Pmpk = 0.25*Ts*wIDL {W}

Voltage Adjusted Peak Power
Vs - ideal source voltage
Pmpk = 0.25*(Vs/Vnom)*Ts*(Vs/Vnom)*wIDL {W}

This scales the power up or down by (Vs/Vnom)^2, and gives back nominal peak power when Vs = Vnom. So it looks right.

One challenge is the frictional power loss may be non-linear with voltage too, so one would not measure the ideal power scaling at the flywheel.

This suggests a power normalization algorithm on the Sentry might have to rely on the current sensor and motor circuit estimates based on sensor data. It could become fairly complex and thus less reliable.

Last edited by SystemTheory; 04-03-2009 at 03:02 PM. Reason: clarify terms
SystemTheory is offline   Reply With Quote
Old 04-03-2009, 06:35 PM   #113
Tech Elite
 
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777
Trader Rating: 27 (100%+)
Send a message via AIM to John Stranahan
Default

Richard-Thanks. Thats a whole lot of time you have requested. I will comment on some of it. Comparing speedos on the same motor is somewhat useful. These days there is so little difference between top quality speedos, that I think time would be better spent looking at motors.

When you dyno different winds of motors the difference in track performance is very difficult to determine by looking at the ouput. Peak power numbers are certainly helpful. Matt might be posting some more of those later. He has taken quite a few runs.

I made some progress predicting track performance from dyno data on stock motors with different kinds of armature. The performance differrence was most clear looking at Power vs Amp draw graphs. That graph is not the Forte of this dyno, though, I tried one. Peak Power is the useful output here as well as those differences in power that the speed control can modify.

"system Theory"
"In the ideal brush motor model both torque and speed get boosted by an increase in voltage, so I'm fairly certain power increases with voltage squared. My logic is presented symbolically below.

In the real world power increases in a straight line with voltage but the two are not proportional. Alas the straight line does not go through zero nor near zero. You would not really be "normalizing" the motors with a Watts/volt nor W/volt^2 (graph on request). It may be somewhat different on brushless due to lessened friction. The real beauty of this dyno is that you are getting very realistic max power numbers by using an actual RC pack. If you use a 4 cell pack you would get Ideal numbers for the 4 cell crowd. There is not an accurate conversion concerning voltage like we would like.
john

Last edited by John Stranahan; 04-03-2009 at 06:47 PM.
John Stranahan is offline   Reply With Quote
Old 04-03-2009, 09:41 PM   #114
Tech Adept
 
Join Date: Mar 2009
Location: Metro New York
Posts: 139
Default

John,

I'm having trouble visualizing a linear relationship between voltage and power as you describe. I'll continue to think about this.

I attach a plot of sampled current versus time for the Hacker 13.5 on Sentry Dyno run. This was made in Open Source Calc using the graphing function. The line fit is just to test the Cubic Spline with resolution 20 over 3 points (I don't know what that means, but that's the default setting in Calc). It puts a spike up at 0.1 second but fits the data everywhere else.

I note the downward inflection near the start. A clean step voltage input to a brush motor would not produce such inflection. Informally this could be attributed to a person pulling the throttle (ramping the voltage up over a few tenths of a second) and/or current limiting via electronic commutation. The inflection should not be introduced by clipping in a Hall Effect sensor.

I note one "dip" in the current that looks like sensor noise, otherwise current is decreasing monotonically in uniform steps. I assume these steps of two or three constant current values in a row are due to the digital resolution of the Sentry, as a smooth decrease of current should occur due to the continous increase in shaft speed bringing up the back-emf voltage. If the digital resolution is at its limit, the values would sample constant for a few tenths of a second.

It might be instructive to throw out sensed current values above 100 amps and do a regressive curve fit. I would be interested to see the predicted start/stall current at 0 seconds. This would be no less reasonable then smoothing the speed curve to calculate starting torque on the flywheel, although both are approximations in the face of sensor limitations.

I would plot power graphs versus time, with option to add current:

Estimated Air Gap Constant
k = Ts/Is (air gap constant at zero shaft speed, assumes zero stiction torque)

Estimated Air Gap Power
Pgap = k*w*Im {W} (where k is estimated, speed w and current I sampled)

Estimated Bearing Dissipation
Pbd = Pgap - Pflywheel {W} (where air resistance is negligible at low speeds)

If air resistance is negligible on the flywheel at relatively lower speeds, these plots would be a fair approximation for the particular motor/ESC combination. One might then discover a reasonable method to "normalize" the flywheel power per sensed volt for a specific motor/ESC on the Sentry (other combos could only be self-normalized).

Although this is an approximate power analysis it is still useful to gain insight into how the system functions both on the Dyno and out on the track.
Attached Thumbnails
Novak Sentry as Brushless Dyno-hacker_13_5_current.png   Novak Sentry as Brushless Dyno-hacker_13_5_power_ex.png  

Last edited by SystemTheory; 04-03-2009 at 10:03 PM. Reason: attach graph, clarify, second graph
SystemTheory is offline   Reply With Quote
Old 04-03-2009, 09:59 PM   #115
Tech Elite
 
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777
Trader Rating: 27 (100%+)
Send a message via AIM to John Stranahan
Default

"I'm having trouble visualizing a linear relationship between voltage and power as you describe. I'll continue to think about this."


It takes about 1.5 V just to get an RC brush type motor turning. A straight line develops from this point and increases. This is a graph out of an RC book that I wrote. It is no longer available. I have not tested brushless motors this way but you can see the problems adapting strict theory to RC motors. Click the graph then click once more for a larger view of the graph.
john
Attached Thumbnails
Novak Sentry as Brushless Dyno-power-vs-voltage001.jpg  

Last edited by John Stranahan; 04-03-2009 at 10:16 PM.
John Stranahan is offline   Reply With Quote
Old 04-03-2009, 10:25 PM   #116
Tech Adept
 
Join Date: Mar 2009
Location: Metro New York
Posts: 139
Default

John,

I assume you increased the applied voltage in discrete steps, and measure peak flywheel power on the Dyno? The motor is allowed to cool off, then you do another Dyno pull? Or is it some other process where voltage is dialed up continuously and you're able to pick off another peak power measure?

I see the graph, thanks. I wondered about this constant amp setting on a Dyno. I know in control system engineering, to get constant torque out of a brush motor within some limited speed range, one designs a constant current regulator. The constant torque would look like the horizontal line in this graph I posted above, and the power output may be different than a motor operating in the voltage controlled mode. If a motor is operating in the dark gray constant torque region it will make less power than if it climbs to the midpoint of the torque-speed line, unless the constant torque test is designed to hit the mid-point of the torque-speed line? Anyway thanks for the information as it is useful.
Attached Thumbnails
Novak Sentry as Brushless Dyno-brushlesscurve.gif  

Last edited by SystemTheory; 04-03-2009 at 10:45 PM. Reason: graph comment, add graph and comments
SystemTheory is offline   Reply With Quote
Old 04-03-2009, 10:32 PM   #117
Tech Elite
 
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777
Trader Rating: 27 (100%+)
Send a message via AIM to John Stranahan
Default

This is done on a Competition Electronics Brake Dyno. This is probably the finest RC measuring instrument ever produced. You can choose to vary voltage in 5 discreet steps. The dyno averages power for 4 s at each level. Extremely reproducible down to 1 W if you prep and cool your motor properly. Voltage is held to good quality voltmeter standards.

John
John Stranahan is offline   Reply With Quote
Old 04-04-2009, 02:20 PM   #118
Tech Adept
 
Join Date: Mar 2009
Location: Metro New York
Posts: 139
Default

John,

Attached is a simulation of the Hacker 13.5 approximated "brush motor" model driven by a constant 30 ampere source current for 4 seconds, as if it were on the current regulated dyno you describe.

Notice my curves come up very similar to yours. I did not add anything to force a non-zero turn on threshold voltage, and I was surprised to see it appear in the result. I must study it further.

The power is roughly linear to voltage only because the current is constant on the Dyno. This could be an advantage for sensitive motor tuning, but it does not predict power at a given voltage under speed control on the track.

The Sentry Dyno uses a voltage source, not a regulated current source, and the power is going to be proportional to the voltage squared, since a much larger start/stall current is permitted to flow at increasing input voltage. There are practical limits on this theoretical model, but I think it is correct.
Attached Thumbnails
Novak Sentry as Brushless Dyno-ce-dynockt.png   Novak Sentry as Brushless Dyno-ce-power_vs_volts.png  
SystemTheory is offline   Reply With Quote
Old 04-04-2009, 06:04 PM   #119
Tech Adept
 
Join Date: Mar 2009
Location: Metro New York
Posts: 139
Default

I ran some models in my engineering simulator to test my "Watts per Square Volt" battery exclusion hypothesis. For brevity I present typical results for Case A versus B below.

A. Hacker 13.5 on Sentry Dyno
:Battery Parameters --------
Vs = 8.4;
Rs = 0.012;
:Motor Parameters ---------
k = 0.003319;
Rm = 0.062;
Jm = 4.138E-07;
br = 1.505E-05;
:Flywheel Load ------------
JL = 3.45E-04;
:Pre Netlist Calculations -----
Jeq = Jm+JL;
Req = Rs+Rm;
Vbe = k*w;
Ia = (Vs-Vbe)/Req;
Tm = k*Ia;
alpha = (1/Jeq)*(br + k**2/Req);
tF = 5*(1/alpha);


B. Hacker 13.5 on 5V Regulated Dyno
:Battery Parameters --------
Vs = 5.0;
Rs = 0.0;

Power divided by Square Voltage Metric:

Case A = 216.4/(7.65*7.65) = 3.6977 Watts per Square Volt
Case B = 92.9/(5.0*5.0) = 3.716 Watts per Square Volt

Percent Error < 0.5%

This assumes the source voltage and resistance change but no change in the motor parameters due to temperature, mechanical wear, etc. In practice a motor drawing more current at higher voltage may make a bit less actual power due to self-heating in the copper coils and magnets increasing Rm and decreasing k.
Attached Thumbnails
Novak Sentry as Brushless Dyno-hacker_13_5_ckt_diag.png   Novak Sentry as Brushless Dyno-sentrypowermetric0.png   Novak Sentry as Brushless Dyno-sentrypowermetric.png  
SystemTheory is offline   Reply With Quote
Old 04-04-2009, 08:30 PM   #120
Tech Elite
 
Join Date: Aug 2002
Location: Houston, Texas
Posts: 3,777
Trader Rating: 27 (100%+)
Send a message via AIM to John Stranahan
Default

If only you had measured the power at two voltages on a dyno and then compared this to your voltage squared hypothesis ( or theory) and then reported the error, then I would believe it. The motor theory is shown lacking again and again even with your Power vs volts graph. Again and Again you try to use theory to predict this maximum power (by taking outlandish liberties with some of the factors). It is close, just not close enough to be useful. Just use the dyno. That is what its for. I still don't understand what you are trying to accomplish. All we want is a good power number and maybe a few useful graphs to look at. The other discussions have driven people away.
John
__________________
[email protected]
TeamCRC.com Mikes-HobbyShop.com
John Stranahan is offline   Reply With Quote
Reply

Similar Threads
Thread Thread Starter Forum Replies Last Post
RC10B4/T4 Forum DerekL Electric Off-Road 14339 12-13-2017 11:20 PM
Dyno, Homemade, Using a Novak Sentry Data Logger, Continued, The Experimental Thread. John Stranahan Electric On-Road 344 12-30-2016 05:22 PM
CRC Battle Axe, GenXPro 10, 1/10th pan, Brushless, Lipo,4c, Road, Oval,TipsandTricks John Stranahan Electric On-Road 1776 03-01-2016 01:18 PM
Novak New Ballistic Motors 20 SMOKE Electric Off-Road 11 05-11-2009 03:58 PM
Idea to Use brushed motor dyno for brushless JevUK Electric On-Road 45 04-06-2009 10:52 AM



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off



All times are GMT -7. It is currently 05:12 PM.


We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.
Powered By: vBulletin v3.9.2.1
Privacy Policy | Terms of Use | Advertise Content © 2001-2011 RCTech.net