The FET's (Field Effect Transistors) that are used in brushless speed controllers are the little electronic switches that turn on and off, cycling the power to the 3 phases of the motor to make it spin. FET transistors have a very unique property that makes them well suited to this type of use. When they are turned off, their resistance is very high, measured in the millions of Ohms, so they conduct virtually zero current. When they are turned on, they have an extremely low resistance, measured in milli-ohms, so they are very close to being a dead short. Because of these two characteristics, when they turned of they draw almost no power, and when they are turned on, there is very little power loss through them.
The only time that a FET has any real measurable resistance is during the brief period of time while it is switching from either off to on, or on to off. During this time period, the FET has a measurable resistance, and has the full phase current passing through it. Because of this, the only time that the FET really has to dissipate any real heat is while it is in the transition period while it is either switching on or switching off.
When a motor is running at full speed, Each FET in the ESC turns on and off one time per switching cycle. So when the ESC is running at full speed this is the time when there are the fewest number of on and off cycles per FET.
When a 3-phase brushless motor runs at less that 100% power, the way the ESC slows the motor down is by taking each phase pulse from the ESC and chopping it up into smaller pieces. So when you run at less that 100% throttle, instead of the FET's switching on and off once per phase cycle, they might switch on and off 2, 3 or even 4 times depending on how high you have the PWM frequency set. Since most of the heat generated by the FET's occurs during the time where they are switching on or off, if you double or triple the number of switching cycles per phase cycle, you double or triple the heat output of the FET's.
Most people have a misconceived notion that as you throttle back, the current through the FET's decreases, however this is not the case. When the ESC turns on, the entire battery voltage is placed across the motor whether it is running at full throttle or 1/4 throttle. At full throttle it just does it for a longer period of time per cycle, and at lower throttle settings it does it for a shorter period of time. Because of this, the average current is lower at lower throttle settings, but during each pulse of power, the current through the FET's is the same.
The worst case scenario for an ESC is to run at around 2/3 throttle setting. In this case, the transistors are carrying a heavy load and switching on and off several times per cycle, with no time to rest between cycles. It is during operation in this range that the ESC will get the hottest.
The ESC is always sending a series of pulses to the motor as each phase is energized while the motor spins. At full throttle, each of these pulses is like a rounded off square wave with only a turn-on at the beginning of the pulse and a turn-off at the end of the pulse. As you slow down the motor, these pulses get longer, but each individual pulse starts getting chopped up to reduce the power output. So at reduced power settings, the FET's may turn on and off multiple times during the total pulse width.
The set-up that was used consisted of the following equipment:
A 3-cell 1800mah X-Caliber Li-Po battery that was partially discharged and measured 11.8 volts.
A Scorpion 110 amp Commander ESC.
A E-Max 2826 motor that has a Kv of about 810.
The ESC was driven with a servo driver box so the pulse width of the input signal could be set to the ESC.
The test consisted of running the motor at full throttle with a 2.0mS pulse from the servo driver. Then it was tested at 1.8mS, 1.6mS, 1.4mS and 1.2mS to get readings at 100%, 80%, 60%, 40% and 20% throttle settings. The waveforms shown below are measured with the test probe on one of the motor phase connections, and with the ground reference on the negative terminal of the battery.
This first photo is a screen shot from a digital storage oscilloscope with the motor running at full throttle.
- ESCThr100.jpg (17.08 KB) Viewed 2983 times
In this photo the voltage scale is 2 volts per division, and the time scale is 0.2mS per division. The ground reference was set for 1 square up from the bottom of the screen, so if you count up to the top of the waveform, you will see that it is 5-1/2 squares, which equals 11 volts. The power input is 11.8 volts, and the junction of the FET's act like a diode when they are turned on, so 11 volts is just about right when you consider the diode drops in the circuit.
If you count the duration of the waveform from one peak to the next, it is about 4-1/2 squares or 0.9ms This gives a rep rate of 1111 Hz. This is the number of times the phases are firing per second on each of the motor phases at full speed. To convert this value into RPM, you can first multiply it by 60 to get the number of pulses per minute, and that equals 66660. Now, since the motor has 14 magnets, and during each cycle of the ESC, the motor moves past 2 magnets, it takes 7 cycles to make the motor go around 1 time. So now if we takr 66,660 and divide it by 7 we get 9522 RPM.
To verify this mumber, if we take the Kv of the motor, which was 810 and multiply it by the voltage applied, which was 11.8 we get 9558, which is pretty close to the measured value of 9522. SInce we are doing a rough approximation from the values on the scope screen, anything within 5% it great.
Let's see what happens when you start reducing the throttle. Here is a graph of the motor at 80% throttle.
- ESCThr80.jpg (18.46 KB) Viewed 2980 times
Now you can see that the FET's are switching on and off once or twice per cycle to reduce the motor speed a bit. It is still spinning at around 9500 RPM, but now instead of one on and off per cycle, we are now getting 2 or 3.
For some of you, the graph may appear upside down, because you were expecting to see the motor turn on for the pulses instead of turning off. When the motor runs, the phases are normally all at the battery voltage, and the FET's pull the phase wire down to ground to get the current to flow. Here you can also see that the full 11 volts is still going to the motor phase, it is just on for about 80% of the time now and off for 20% of the time.
One other thing to note is that you can also see the PWM frequency in this graph. If you look at the 2 spikes that are in the top right section of the display, you will notice that they are a little less than one square apart. It is actually somewhere between 0.6 and 0.7 squares apart. Actually they are exactly 0.625 squares apart:
Remember that the time division per square is 0.2mS, and this is displayed in the top right corner. To calculate the time difference between these two spikes, you take 0.2mS per division times 0.625 divisions, and you get 0.125mS. Now if you take that time, and divide 1 by that number, it equals 8000 or 8 KHz. Why does that number sound familiar? Because it is the PWM frequency. The 8 KHz PWM frequency is the rate at which the ESC chops up the phase signal going to the motor in order to slow it down. The ESC fires as fast as the motor is turning, but the amount of power that goes into each cycle is controlled by the PWM modulation of the signal.
Keep reading to get the overview. It will start to make sense...
Now lets look at the waveform at 60% throttle and see how things change.
- ESCThr60.jpg (19.64 KB) Viewed 2971 times
There are a couple things that are different now. First the motor has slowed down. If you look at the time between peaks on the signal it is now 5 squares where in the above graphs it was 4-1/2 squares. This means that we are now having a time of 1.0 mS per pulse which is equal to 1000Hz. Converting this to RPM like we did earlier gives a motor speed of 8570 RPM.
Since the pulses are now a bit longer, you can fit more PWM cycles into each pulse. In the first power pulse you can see that the FET's are switching onn and off 5 times per cycle, and that the signal is still pulling all the way down to ground.
Next let's look at what the waveforms look like at 40% power.
- ESCThr40.jpg (19.77 KB) Viewed 2984 times
Now if you look at the time duration between peaks, it is about 5-1/2 squares now. This is a time duration of 1.1mS or 909 Hz. Converting this to motor RPM as we did before give an RPM of 7790. If there was a prop on this motor, it would be slowing down more, but in the interest of safety, the prop was left off during the test, but this is an inadvisable operation. Now with the motor running at a lower speed, the 8KHz PWM frequency can fit in 6 pulses per cycle. You will also notice that the waveform is starting to round out on the bottom pulses. Here you are getting to a point where the motor is not fully turning on any more because it does not need to to maintain the RPM. The signals would look different under the load of a prop.
Finally, here is the motor at 20% throttle. Things are really changing now!
- ESCThr20.jpg (19.21 KB) Viewed 2969 times
The motor has slowed down considerably, and now the time between peaks has increased to about 8 squares or 1.6mS. This is a frequency of 625 Hz which equals 5350 RPM. Now that the motor is turning slower, there is enough time for 9 PWM pulses per cycle. You can also see that the motor is only getting about half power to the phase between PWM pulses. If you look close, you can also see a little ripple on the top of the pulse while the phase is off. These little ripples are happening at an 8KHz rate in response to the power pulses that are surging through the motor during each PWM pulse.
As you can see, there is a LOT going on inside the ESC and motor while it is running, and this is with no prop installed and the motor only pulling a couple amps of no-load current. If you put a prop on the motor and start pulling some real current, the waveforms get pretty ugly!
Another thing that many people make a mistake at is in sizing their ESC to a motor. Many times people will take a motor that is rated for 90 amps max current, and run it on a small prop so it only pulls 50 amps. Then they figure since they are only pulling 50 amps, that a 55 amp ESC will be fine. BIG MISTAKE!
A larger motor has larger coils of wire in it around the stator, and because of this, it presents a greater load for the FET's to switch when they turn on and off. The amount of time that it takes for a FET to turn on depends on the size of the load. The coil of a brushless motor acts like an inductor, and inductors try to prevent rapid changes of voltage by their very nature. This is why they are used as noise filters in DC circuits.
If you have a small motor with small coils, the resistance to voltage change is less, and the FET's can turn on and off rather quickly. If you have a larger motor the inductance of the motor fights the FET's harder and makes it harder for them to switch on. This increases the amount of time that the FET's take to turn on, and since they FET's only get hot during the time they switch, it increases the heating of the parts. For this reason, if you use an ESC that is rated for less than the maximum motor current, even if you run the motor at less than the max current for the ESC, the ESC can still overheat because it is fighting such a large load.
So if you have a motor that is rated for 90 amps, use a 90 amp ESC or even better oversize it by 50%, even if you plan on only running the motor at 50 amps. This way the number of FET's on the ESC is matched to the load that the motor will put on the ESC, and everything will run in an optimized fashion.
Reprinted with kind permission of Scorpion Motors
