In one of the other threads, Igor B. and I were talking about the efficiency of flying around at partial throttle. I made a comment that I was seeing roughly 5% effects in power usage by flying at partial throttle as opposed to using something closer to WOT (of course at the same rpm level).

To make it clearer, what I mean, is suppose you can fly the full pattern at a rpm which makes you happy---using a 3s battery. Is there any penalty hit by replacing the 3s with a 4s battery and leaving the motor alone, and still targeting the same rpm level? Another way of looking at this is suppose you have a 4s battery, is there any reason to try and optimize the motor by choosing a different e.g. lower--kV motor to match the 4s voltage?

Just to check this out again, I setup a test bench in the basement with my Scorpion 3020-16 motor. I mounted an APC 11-5.5 prop and ran the setup in a pusher mode--just to keep the "breeze" blowing away from all my setup stuff. I used my standard Castle Creation Phoenix 35 running in rpm governor mode (I think set at 7950 rpm). It turns out that the 11-5.5 prop running statically more or less gives the same power levels (~250 watts) that my E-Nobler uses in level flight with the APC12-6 TE prop (at same rpm).

The following narrative refers to the attached plot.

I first connected a 4s2100mAHr pack to the setup and ran for about 4 minutes. The battery voltage is shown in the white trace. After that, I quickly took off the 4s and put on a 3s4200 pack and started it back up again. This I ran for about 7 minutes until I finally detected the rpm to start to drop. At that point I was presumably at WOT since the governor couldn't keep the rpm at the set value--the battery voltage had sagged. Unfortunately for this plot, my Eagletree data recorder had run out of memory, so you don't see the rpm sag, but please take my word that the battery voltage was only a 0.1V lower than what you see in the white trace at the end of the plot.

The power draw for both batteries is the light blue trace. You can see that when using the 4s pack, the average power was 248 watts, and with the 3s pack, the average power was 237 watt, or 11 watts different. I will also note that the average rpm for the 4s pack was 7969 rpm and the 3s pack was 7954 rpm (15 rpm lower). Ignoring this slight rpm difference and attributing the full 11 watts to the intrinsic inefficiency of the throttle (not completely obvious that this is ok--but it gives the maximum inefficiency), I see that the efficiency difference is ~5%, which I am thinking is pretty insignificant in the big scheme of things!.

The yellow trace is just the average current in amps. Since the ESC was basically at WOT at the ~11 minute mark, that says my motor current at ~7050 rpm is about 23 Amps.

Since the battery volts at 11 minutes is ~10.2V (at WOT), I think that means when I am flying a 4s pack with this motor, (pack voltage ~14V near end of flight), my throttle setting is just ~10V/14V or ~70% during level flight.

There are some curiosities in the plots that I don't fully understand, especially the lower power usage at the beginning of the flight at full rpm (seen with both packs). I am guessing it is some ESC heating but am not sure.

Also there are some other side effects to using a "mismatched" motor/battery pack, mainly in battery heating which isn't really shown in this data, but in general it certainly looks like we can getaway with reasonable mismatches in motor kV (too high) and pack voltage (also can be "too high").

Finally I note that I can't use a 3s pack in a real flight with this motor since I need the extra throttle capability to handle the higher power during the maneuvers. So any attempt to match the kV a little better would bring less than 5% since you always need that extra "oomph" of more throttle when you pull the nose up.

I dont have the option, but I wonder what would show if you were to turn the same rpm with two different kv motors. say using a 900 kv motor on the 4s and an 1100 kv on the 3s so that both props turned the same rpm. This could be enlghtening as well. I hope to do this with my hacker setups in the future.

interesting Alan,
what I am really after is this, IN THEORY if I have a 4270 mah battery (which I do, actually four of them) and it works with my 900 kv motor and a 12x6 prop to fly the pattern. If I switch to a 12x4 prop and a motor that has a kv of 1100, when I get the plane flying the same lap times as the 900 kv motor was doing, would I still be using the same approximate amount of the battery capacity spinning the prop up at a proportionately higher rpm. Realizing that there are variables of course. I am after an approximation. I plan on doing physical testing, I have both motors, and systems, just need something besides 35 degree weather, IM a light weight on temperature outside, lol

Alan
thats the direction I am thinking, the increase in thrust, and braking. Hoping to be able to make it work with about the same ultimate current draw. at least to be within the range of my battery capacity.
I know your system is built around the smaller batteries you are using so that dictates to some extent what you do, as in your quest for max efficiency. When I get my new setup up and running, I will post findings.

Here is an oscilloscope trace that we made on Wednesday night of the motor current when the ESC was at partial throttle. We used a Hall Current probe that clamped onto one of the three motor leads. What you see here is one commutation cycle where the particular lead was on. The sharp spikey traces are just the Phoenix default 12kHz current spikes going into the motor. You should ignore the dipping of the baseline---the probe was AC coupled which gives rise to this instrumental effect (in other words, this is not what the motor is seeing. The horizontal scale is 200 microseconds  per major division or about 2 milliseconds for the entire trace. 12kHz just gives 83 microseconds between pulses.

I am guessing that the throttle was about 20-30% on based on how long it takes the pulse to rise from the baseline to the peak--compared to the time to the beginning of the next spike.

To be honest, I was somewhat surprised by this picture. If you look at any one spike, you can see it has a risetime to the peak and a fall time. This is (I think) just the L/R (inductance of the motor phase divided by the resistance of the same (+whatever the ESC adds in). I had thought this was going to be longer than the spacing between the spikes. Obviously it isn't.

The vertical axis is related to the current, but I didn't note the conversion in the probe. I was running the motor with a 2s pack--just to keep the prop breeze (and amps) out of the worry zone.

So I learned something about how these ESC/motors work. Just to state the obvious, the overall motor torque is being provided by the individual torque impulses. The impulses are short enough in time and frequent enough so that it appears relatively smooth and continuous. The total time of the 16-17 pulses is 1.3ms, which is basically 1/14 of a revolution (1 of the 14 magnets), so the rpm should just be 60*/(14*1.3ms) or ~3300 rpm. I wish I'd brought my tachometer to check this, but it probably is consistent with the 2s pack at partial throttle.

I didn't make a screen dump of full throttle, but when I turned it on full (100%), the spikes got a little higher in amplitude and they just all merged together into a single longer pulse which equalled the commutation period--which also got shorter in time(in other words, the rpm increased at full throttle which makes the period shorter in time).

Well look at it this way.

The purpose of an electronic speed control (ESC)  is --speed control! So how to do it.

One way would be to have a continuously variable voltage (like a power supply with a rotary knob). Max voltage would give max speed (rpm in a motor), and less voltage would give proportionally less. In the case of the motor, the electric current through the motor (which provides the torque) depends on the applied voltage.

The problem is how to make the variable voltage, or functionally equivalent a varying current. An obvious, but lousy choice, would be to drop the voltage with a variable resistor. The problem with this method is that you waste a lot of energy in the resistor (as heat = I^2*R, R being the resistor).

The way our ESC's are doing it is to chop up the current into short pulses. Each current pulse "kicks" the motor. The amount of "kick"(aka "impulse") is given by the length of the kick. The % throttle is just be the % time of the time "on" length to the "off" time.  If the kicks come fast enough (12kHz for the default Castle Creation Phoenix ESC setup) these discrete kicks will appear to be continuous. So in this way the ESC can vary the power put into the motor.

As I mentioned, what surprised me was a electronics detail. A judicious choice of rep rates (much higher frequency than 12kHz apparently!) would actually smooth out the current to almost a dc-like nature--rather than what you see in the scope trace. But given that the ESC is pretty darn efficient at partial throttle (at least no worse than the 4% I mention in the initial post of this thread.

Dean,
There are a couple of issues with your suggestion.
First the max rep rate is ~24kHz with my Castle Creations Phoenix, and I don't think that is short compared to what L/R is (just eyeballing it).--correction--now that I look, I see that I could set a rate in the 50kHz range--now I need to think about it, just for trying it sake!
Secondly the on resistance of the ESC is pretty small, but during the switching period, between "Off" and "On" you apparently go though a finite resistance which I have heard counts for much of the inefficiency (aka "switching losses") of the ESC.

On the other hand it perhaps is worth a trial run just to see what happens. Yet on my "other" other hand, I seem to be losing "only" ~4 or 5% of the total efficiency. Even more optimized would probably bring back a half of that (since I can't run at WOT at cruising speed if I want maneuvering power).

For me the interesting thing was to actually see what was going on. A little reality helps to cut down on the speculation!

Thanks Igor---I need to revisit some of those older threads. What scares me is that I usually learn something by reading my own posts!

Charles,
I think you are right about the motor eating up the power. The average "kick" is the same, but the heating is going as I^2, so it is less efficient--but not horribly so. Yet it is possible to squeeze a little more out!