What is the problem

[Sparks]

On the basis that a picture is worth a thousand words, here is an actual 'scope trace of the three waveforms at the motor/ESC I took some time ago (with added text to show what is going on).

The motor was an E Flite 32, Black Mantis ESC and 3S LiPo.

Full Load:

... and part load:

.. and some info concentrating on one phase.

may the debate continue....

Sparks

[John Olsen 1]

OK, those are the waveforms across the switching devices in the ESC...your scope common was attached to the negative rail. The Voltage across each motor coil connection is the difference between each of these waveforms. eg A-B, B-C and C-A. This is the exact reference problem that I referred to in my last post. The negative (or positive) battery connection is not relevant to the voltage across the motor coils since they are not connected to it. (They are sometimes connected to it but they are sometimes connected to positive and sometimes not being driven at all.)

For a similarly misleading result, take your scope, connect the common point to the negative of a DC supply from a bridge rectifier, then take the waveforms for each of the AC connections to the bridge. You will find that they are rather like your first set of waveforms, eg a sine wave shape, but offset so that it does not go significantly below zero. So the transformer is not supplying AC at all, it is supplying sine wave shaped pulses of DC. Well no, if you apply the scope across the transformer you will see that it is actually AC as would be expected. But it does not look like AC if you use the wrong reference.

John

[Peter Beeney]

Just to try and justify what I said previously about AC from a DC battery, and if I may, perhaps further attempt to illustrate John’s post, I’ve done a little drawing. This is a H bridge. This is one way of obtaining AC from a steady DC battery supply. Apologies for the quality of these in advance for these, I’m afraid I’m not much of an artist.

The H Bridge.

It comprises of 4 switches - j, k, l and m, a resistor R, as a load, plus two voltmeters 1 and 2, connected as shown. We’ll give it a voltage level, 10 volts. In sketch (a) all the switches are open, and thus both meters will read zero, so that’s a start level in the output trace to the right.

If we now close switches j and m, as in sketch (b), current will flow from left to right though the resistor. Voltmeter 1 will read a positive 10 volts, and v.m. 2 will also read +10 volts. The output can be shown on the graph at the right. Then opening these two switches and simultaneously closing k and l, as in sketch (c), this then shows the current has reversed direction and is flowing from right to left through the resistor. Voltmeter 1 will now read a negative 10 volts, but voltmeter 2 significantly will read only zero volts. Again the output is shown on the graph to the right. If we now repeat this on a regular basis we have a square wave AC output, but as you can now see, you have to have the voltmeter connected to the correct frame of reference to see this. The square output trace on the right shows the output.

Looking back at the first waveforms again, the rather more unrefined sketches on the right, if we add a bit of PWM to this, to reduce the voltage a bit, we can see that v.m. 1 will still show the full waveform but if we are looking at it with the second voltmeter we get the second view. Which as we can now see is missing it’s better half, which will be completely misleading. We could in fact perhaps imagine this is DC. If this were two oscilloscopes this is exactly what we might see on their traces.

Commutation.

The two sketches (f) and (g) on the second picture show how revolving the switch, or commutator, will also reverse the direction of the current through the winding every 180 degrees. But again you would have to have the voltmeter or scope across the winding to actually see this in operation, and that’s perhaps only obtained with some difficulty.

The colour code is black for 0 volts, normally the DC base reference level, red is anything positive with respect to 0 and blue is anything negative w.r.t. 0. I’ve used green to show the direction of current flow. If we could arrange to turn on the switches gradually we could start to turn the waveform into a curved wave shape, perhaps even to a sine wave. After all a sine wave is only a square wave with all the harmonics removed. With regard to a sine wave, in exactly the same way we can see from the square wave that the average, or mean, voltage over one full cycle is only ever going to be zero, which is true of any waveform. Adding equal plus and minus numbers is only ever going to equal zero. So the average is taken over one half cycle. Because all roots are positive the root-mean-squared value can be calculated over the complete cycle but it will be projected as two positive half cycles. So one half will suffice anyway. Because the heating effect is not affected by AC the r.m.s. value has the same power content as an equal DC voltage. So generally the AC figure is a r.m.s., such as the mains voltage. In the vast majority of cases I don’t think the actual wave shape is vitally important, it doesn’t seem to bother many appliances very much. It’s the fact of being AC or DC that matters.

So is the next step perhaps deciding on how and what makes the motor actually turn?

Now all that remains to do is to substitute the resistor in the H bridge for a small brushed electric motor, arrange for the switches to be turned on and off by a control stick via a radio link and the motor turning an arm through a system of gears, and we’ve got ourselves a method of remotely controlling a model aeroplane……

PB

[Erfolg]

I have flown the model again on Sunday, using an old (although unused) 30 A ESC.

With additional ventillation to the ESC there does not appear to be any issues now. This is a change from a position where neither the motor, Lipo or Esc were ventilated.

Examination after landing did not indicate any rise in temperature in any of the equipment, appearing to be near to ambient temperature.

I have started to try and operate at either full throttle of about 1/3 to lower the work rate on the thyristors, as outlined in the thread,

The motor (Runthump) itself seems to be working fine now that I have replaced the drive shaft.

The discussion has been most informative. Although we now seem to be into the area of discussing the age old "Angel on a Pinhead" conundrum. Everything else seems to be agreed, how the thyristors are made top work in an ESC. It is agreed that the current flow reverses in a brushless winding in the same way as a brushed motor. Therafter, well, it is a discussion for those who know the intricacies of definitions.

For me, the source is DC, the motor coils function as a DC motor, what else is there to discuss?

[Chris Bott - Moderator]

Nice one Erfolg, good to hear it's all working as it should now.

Let's not get hung up on the fact that the power devices in ESC's are MOSFET's not Thyristors

[kc]

What a wonderful discussion! Plenty of food for thought here from all the " usual suspects" ( i.e those few people who comment usefully on anything technical and make ModelFlying a great forum )

But what is the answer - does the ESC work harder at low throttle? ( obviously I have not understood the diagrams! )

[John Olsen 1]

Kc, the answer to that is "possibly". There are a lot of factors that determine the power being lost in the switches and their drive circuits. The one that might be more significant at medium speeds is the actual switching loss. It takes a finite time for a device to go from on to off or off to on, and for that time it is dissipating power. There is also power used in the drive circuits to charge up and discharge the gate capacitance .So the more often you switch the device on or off then higher the dissipation from this cause will be. As you will see from the waveforms above the devices are switching more often at medium power than at high power. So they could get hotter at medium speeds. But it is hard to be definitive, since the dissipation due to the current (I^2R loss) will be higher at full throttle. So it will depend on the design of the particular ESC. The cooling airflow will also be less at medium speeds.

Don't be too hard on Erfolg about the thyristors, the circuits for AC motor controllers back when I was working on them used the same bridge layout except with thyristors. There was some additional complexity to allow turning the thyristors off. They may well use MOSFETS these days, especially in the smaller sizes, or else possibly GTO devices. But they work in much the same way, chopping up the DC supply to produce three phase AC for the motor to run on. One difference is that they do not sense the rotor position since they do not have permanent magnet rotors.

John

[Peter Beeney]

kc, - Recently I borrowed some bits and bobs, namely an ESC, motor and props etc. to have a little play. I noticed that the most activity apparently seemed to occur just as I started to close the throttle. I say ‘apparently’ because I don’t really have the right kit to see exactly what is really happening, and I wasn’t really checking on this anyway. So this is largely supposition on my part, but it might be a bit of a handle on what is going on.

One way of changing the speed of the motor is to vary the voltage as applied by the ESC. It can do this by adjusting the average voltage over the full cycle waveform. Going back to our square wave, we said the average voltage could be taken over one half of one complete cycle. In our example the voltage is present for the whole of the half cycle so it will be 10×1 = 10 volts. If we now cut this in half, and switch it off for half the time, 0.5t off, 0.5t on, it will now be 10×0.5 = 5 volts. Off for 0.75 of the time, thus 0.75t off, 0.25t on, so that’s 10×0.25, 2.5 volts. This is ok, but the motor might not fully appreciate these big chunks of power being taken out in one go so we could split this up into a number of shorter periods, switching on and off, varying over time. Lets chose an arbitrary figure of say 20 times, spread equally across the half cycle, each one long enough so that when they are added up together they amount to the half cycle switched half off and half on. So it’s an average of 5 volts applied to to the motor again. Because these are short periods, now it has enough momentum to easily keep going through the off spaces. So we can now see that the speed, or in other words, the average applied voltage, can easily be changed by adjusting the width of these periods, or pulses. Hence Pulse Width Modulation, PWM. One advantage is, because the on pulse is always the full voltage, each on pulse is full power. We note this is not AC, it’s not changing direction, just on and off, and looking at the second half of the cycle, we find this will be an identical mirror image, of course. I suspect the pulse rate remains constant, so if our half cycle was half a second in duration, and contained 20 pulses, if it was shortened to a quarter of a second it would only contain 10 pulses, but the effect would be the same. The frequency is constant. There is a much lesser used technique of speed control which keeps the width constant but alters the frequency, Pulse Frequency Modulation, PFM.

When the throttle is fully opened the full voltage will be continually applied, so it’s possible the control will be switched right out, but as soon as the throttle starts to close it will be required. So the transistors start switching on and off and because very nearly the maximum power is still required maybe this is the ‘worst case’ with regard to those switching losses. As I said, this is a bit of a guess, but I’d say that around seven eighths to three quarter throttle is perhaps where the ESC is working hardest. Thinking about it, when the motor is running slowly the transistors will be switched off most of the time, much more time to cool down, and also the change over switching is at it’s minimum too. So I’d think that in general the ESC is not always working that hard at low throttle.

I’d think that the switching losses is yet another debate on it’s own. And, as I said, something of a bit of guesswork on my part. One slightly crude test I could do is to cobble my colleague’s motor and ESC together again and give it a run, carefully monitoring the surface temperature of the ESC. It won’t tell us much how it works, but I think it will give us a bit of an indication as to where it’s working the hardest.

If this overheating is a problem, is it a case that the ESC is not really up to the job anyway, perhaps? One answer might be to shell out some extra shekels on something like a Kontronic ESC. You could run a steamroller over this and it would still go on working. I don’t think you have to consider part throttle openings here.

PB