Aug 18, 2006, 12:54 AM
Registered User
Joined May 2003
768 Posts
Quote:
 Originally Posted by flieslikeabeagl The second part is related to the ESC: how does the efficiency of the ESC change as it goes from full throttle to part throttle?
I think the easier test here at this point might be to run an ESC at full throttle and half throttle while encased in styrofoam with a thermocouple inside stuck on a fet and nothing hooked up to the BEC. Measuring the temp after a few minutes at each throttle position will give a definitive answer for a particular ESC. What do you think?
 Aug 18, 2006, 01:47 PM Got shenpa? Los Angeles Joined May 2004 10,940 Posts Matttay, thats a clever idea. For low temperature rises above ambient, the power dissipation is proportional to the temperature rise. (At high temperatures the relationship is not linear - Stefans Law of radiation comes into play; but for small temperature rises, Newtons law of cooling is valid). I don't think you need the styrofoam - if the ESC was perfectly insulated, the temperature would keep rising continuously proportional to the amount of energy (integral of the power) dissipated. Instead you're looking for a temperature rise proportional to power (not energy), which I think one gets if the ESC is allowed to dissipate heat into the air and reach a steady-state temperature. This technique would certainly work to figure out what throttle setting maximises waste heat in the ESC, etc, but I still can't see the answer to one question - how do you calibrate the temperature rise, so we know that X degrees of temperature rise corresponds to Y watts of dissipation? I have an idea for this (mount the ESC on a big heatsink, mount a power resistor on a second identical heatsink, and use the resistor to calibrate temperature rise vs watts dissipated), but it is a bit involved. There must be a better way! -Flieslikeabeagle
 Aug 18, 2006, 08:08 PM Dieselized User Chicagoland Joined Feb 2000 7,727 Posts I'm not all that concerned where in the system the loss is. In my test we had 16W more watts to turn the prop the same RPM. The ESC ran a bit hotter than the full throttle setting and the motor was quite a bit warmer. We can guess that most of the loss went occured in the motor. The resistance of the ESC is quite low compared to the motor, and that is the only significant source of heat in the ESC. We know that ESC's aren't burning up left and right. ESC manufacturers know this too. If the 16W was lost in the ESC it would have burned up. That would be roughly equal to 130A if we use the published resistance of the PH80. I want to find out how we can determine the system losses There should be a way to estimate the loss based on test data. Estimating the loss of the ESC by temperature might be problematic because the copper wires transfer a good amount of heat. If we look at AstroBob's book we see that peak ESC loss is at ~80% duty cycle for high frequency ESC's. That is at the half power point. I would guess that is varies a bit with switching frequency and motor inductance. The actualy lowest efficiency of the ESC is at a much lower throttle setting but total system power is much lower so the loss is smaller. Greg
Aug 18, 2006, 11:53 PM
Got shenpa?
Los Angeles
Joined May 2004
10,940 Posts
Quote:
 Originally Posted by gkamysz I want to find out how we can determine the system losses There should be a way to estimate the loss based on test data.
Well, you can do that with a motor dyno, by measuring electrical power into the motor, as well as motor rpm and motor torque.

The trouble is, most of us don't have dynos!

-Flieslikeabeagle
Aug 18, 2006, 11:58 PM
Got shenpa?
Los Angeles
Joined May 2004
10,940 Posts
Quote:
 Originally Posted by gkamysz If the 16W was lost in the ESC it would have burned up. That would be roughly equal to 130A if we use the published resistance of the PH80.
Isn't that a steady-state resistance? I believe some of the heat dissipation in the ESC occurs while the FET's are switching between on and off states, during which transistions the resistance is much higher. It takes time to clear all those holes out of the channel in the (P-channel) FET, and of course the inter-electrode capacitances have to be charged and discharged too.

-Flieslikeabeagle
Aug 19, 2006, 02:09 PM
Senile Member
Moab, Utah, USA
Joined Apr 2003
6,671 Posts
Quote:
 Originally Posted by gkamysz I'm not all that concerned where in the system the loss is...I want to find out how we can determine the system losses There should be a way to estimate the loss based on test data.
With that thought in mind I did some empirical testing using an Atlas AM2312/07 motor, a Jeti ECO 12 ESC, a PQ 3s 2200 mah battery, a PQ 2s mah battery and a 10x7 wood prop. I measured the battery current, battery volts and prop RPM at full throttle, 75% throttle, 50% throttle and 25% throttle. I attached a table of my measurement results.

Since I don’t have a dyno, I have no way of determining absolute output power so I cannot determine overall system efficiency. However, knowing that the propeller power changes with the cube of the RPM, I can calculate how much the output power changes at partial throttle by dividing the partial throttle RPM by the full throttle RPM and raising the result to the 3rd power.

I did make some interesting observations:
1. Partial throttle efficiency appears to be about the same as full throttle efficiency. Except at the lowest throttle settings the output power goes down at approximately the same rate as the input power. If there was a change in efficiency at partial throttle, the output and input powers would change at different rates.
2. ESC output voltage does not correspond to throttle position. If it did I would expect to see the prop RPM to be more in line with the throttle position.
3. Output power corresponds roughly to the throttle position, more so for the 3s battery.

Now, this is only one motor-ESC combination and I realize it may be different for other combinations. When I get a chance, I will try it with the same motor/battery/prop but a different ESC and see if I get similar results.

Larry

# Images

 Aug 19, 2006, 02:32 PM Senile Member Moab, Utah, USA Joined Apr 2003 6,671 Posts I was just reviewing my data and made another observation. With the 3s battery it takes 48 watts of input power to spin the prop at 4100 RPM. With the 2s battery it only takes 43 watts of input power to spin the prop at 4100 RPM indicating a higher efficiency. Now we all know that motor efficiency decreases with lower voltage. Does this mean that the ESC efficiency actually increased with the lower voltage more than the motor efficiency decreased giving us a net increase in efficiency? Maybe this deserves more investigation. Larry
Aug 19, 2006, 02:46 PM
Tony Rogers
Bath, UK
Joined Sep 2003
2,198 Posts
Quote:
 Originally Posted by Lnagel I was just reviewing my data and made another observation. With the 3s battery it takes 48 watts of input power to spin the prop at 4100 RPM. With the 2s battery it only takes 43 watts of input power to spin the prop at 4100 RPM indicating a higher efficiency. Now we all know that motor efficiency decreases with lower voltage. Does this mean that the ESC efficiency actually increased with the lower voltage more than the motor efficiency decreased giving us a net increase in efficiency? Maybe this deserves more investigation. Larry
I saw that too but it could just mean that you are at different places on the efficiency curve for the two setups.
 Aug 19, 2006, 10:16 PM Senile Member Moab, Utah, USA Joined Apr 2003 6,671 Posts Giz, you have a point there. Looking at the data yet again, with the 2s battery and the throttle at 75% the input power goes to 87% and the output power goes to 93% indicating an increase in efficiency. At lower throttle settings the output power falls faster than the input power, indicating lower efficiency. It just may be that 5.5 amps is the most efficient current for this motor with a 2s battery. That means this motor is propped for max efficiency at 3/4 throttle and it answers one of the questions posed in the first post of this thread. 'Would it be better to prop for max power at full throttle or max efficiency at 2/3 throttle?' Since this is a 12 amp continuous motor and it pulls only 6 amps at full throttle when propped for max efficiency, it appears that propping for max efficiency at 3/4 or 2/3 throttle will leave you far short of the motor's maximum power capability at full throttle. This seems to be substantiated with both the 2s and 3s batteries. The power changes only between 7% to 14% from full throttle to 75% throttle, whereas the max power current of a motor is usually 200% to 300% of max efficiency current. In short, you can prop for max power or max efficiency, but ner' the twain shall meet. Larry Last edited by Lnagel; Aug 19, 2006 at 10:22 PM.
Aug 19, 2006, 10:36 PM
Dieselized User
Chicagoland
Joined Feb 2000
7,727 Posts
Quote:
 I was just reviewing my data and made another observation. With the 3s battery it takes 48 watts of input power to spin the prop at 4100 RPM. With the 2s battery it only takes 43 watts of input power to spin the prop at 4100 RPM indicating a higher efficiency. Now we all know that motor efficiency decreases with lower voltage. Does this mean that the ESC efficiency actually increased with the lower voltage more than the motor efficiency decreased giving us a net increase in efficiency? Maybe this deserves more investigation. Larry
That essentially duplicates my test of the Aveox. Now we are just going around in circles. The motor is NOT in a different part of the efficiency curve. The losses are DIRECTLY related to part throttle operation.

We know operating at part throttle decreases system efficiency regardless of the motor efficiency curve. If we compare a lower duty cycle to a higher duty cycle for the same power output for any given motor, the lower duty cycle will ALWAYS show a lower efficiency.

The question is how much.

Greg
 Aug 20, 2006, 09:59 AM Senile Member Moab, Utah, USA Joined Apr 2003 6,671 Posts Mommy, why am I going around in circles? Shut up or I'll nail your other foot to the floor. Greg, I reviewed your test and I fully agree that for a given power out the switching action of the ESC is going cause efficiency to decrease as input voltage rises. However, motor efficiency will increase with increased voltage. Neither your test nor mine differentiated between motor efficiency and ESC efficiency. They just looked at overall system efficiency. Which is as it should be. Because trying to analyze brushless motor operation without the ESC is like trying to analyze brushed motor operation without the brushes and commutator. In the case of burshless motors it seems that ESC efficiency has the upper hand. Larry
 Aug 20, 2006, 10:27 AM Dieselized User Chicagoland Joined Feb 2000 7,727 Posts I give up. Greg
 Aug 20, 2006, 10:45 AM Senile Member Moab, Utah, USA Joined Apr 2003 6,671 Posts Please don't give up, Greg. My brain must have been in park with my previous reply and your inputs help me put it into drive. On re-thinking the matter, the results of your test do eliminate motor efficiency since the effective volts into the motor from the ESC have to remain the same to maintain the same RPM. Therefore, the motor efficiency is the same. So at higher ESC input voltages the ESC takes more overhead to keep the output voltage constant, thus lower efficiency. So what I am seeing my data is that with both the 3s and 2s batteries the effective output voltage of the ESC is the same to spin the prop at 4100 RPM but the ESC is less efficient at the higher input voltage because of the lower duty cycle required to maintain that output voltage. I think I got it Larry Last edited by Lnagel; Aug 20, 2006 at 10:56 AM.
Aug 20, 2006, 10:56 AM
Registered User
Joined May 2003
768 Posts
Quote:
 Originally Posted by gkamysz That essentially duplicates my test of the Aveox. Now we are just going around in circles. The motor is NOT in a different part of the efficiency curve. The losses are DIRECTLY related to part throttle operation. We know operating at part throttle decreases system efficiency regardless of the motor efficiency curve. If we compare a lower duty cycle to a higher duty cycle for the same power output for any given motor, the lower duty cycle will ALWAYS show a lower efficiency. The question is how much. Greg
I just looked at some data form International Rectifier that showed measured thermal performance on a variety of SO8, DPAK, MLP and DirectFET packaging and it's all running 35-55'C/W for footprint areas around what you'd expect on an ESC, and they all assume free air.

ESC aren't free air, which will reduce their thermal performance even further.

I'd submit wasted power of an ESC that is sucking 200W is 5-6W max at any throttle position, which is 2-3% total efficiency loss. At that point of operation, you'd expect any one of the six mosfets to be 160'F on the outside of the mosfet's case (inside the shrink wrap).

I'd also submit that of that 5-6W lost, the MOSFET I2R losses are much greater than the switching losses at anything beyond 50% throttle.

The rest of the lost system power (20-30%) is wasted in the motor as I2R losses.

ESC is a small part of the equation, regardless of throttle position. And yes, I still submit you gain more by running at a higher voltage at part throttle then lower voltage at full throttle, because the ESC losses are much much smaller than the motor I2R losses. Outrunners can show a 10% efficiency improvement going from 2S to 4S, which is huge.

Anyway, the above is all speculation. I'll hopefully measure directly on a mosfet this week some time and report back.
Aug 20, 2006, 12:26 PM
Got shenpa?
Los Angeles
Joined May 2004
10,940 Posts
Quote:
 Originally Posted by matttay I'd submit wasted power of an ESC that is sucking 200W is 5-6W max at any throttle position, which is 2-3% total efficiency loss. At that point of operation, you'd expect any one of the six mosfets to be 160'F on the outside of the mosfet's case (inside the shrink wrap).
Excellent work!
Quote:
 Originally Posted by matttay And yes, I still submit you gain more by running at a higher voltage at part throttle then lower voltage at full throttle, because the ESC losses are much much smaller than the motor I2R losses. Outrunners can show a 10% efficiency improvement going from 2S to 4S, which is huge.
I agree. Not just outrunners, the same is true for inrunners and brushed DC motors too, which is why Vintage1 started that "can motors on 3S lipo" thread way back when.

The graphs I attached to post #146 in this thread show how much the efficiency of that (hypothetical) motor drops when the supply voltage is reduced from 10V to 5V.

If you look at the motor equations, the peak efficiency of a motor is proportional to the square root of the supply voltage. This assumes the idle current is constant and independent of supply voltage, which is not quite true for many (most?) motors.

As I mentioned before, using a bigger propeller is another way to make substantial increases in powertrain efficiency. If you download Javaprop from Martin Hepperle's website and play with it, you will find it is very easy to drop below 50% propeller efficiency by using too small a prop. Most small outrunners have Kv's high enough to require the use of a too-small propeller. So we may be worrying about losing 15% of the total power in the motor and ESC, while we are loosing 50% of the available power output due to the propeller!

-Flieslikeabeagle