What is the problem

[Peter Beeney]

I think that I might tend to consider that if my ESC was beginning to overheat at partial throttle openings, and particularly if this was happening even though I’d ensured that everything was in spec., at least to the best of my ability, then I would feel that I would need to question the ability of the said ESC. I really cannot seriously believe that if an ESC is rated at say 40A and with a steady current supply below that figure it then starts to overheat at lower throttle settings it really is a worthwhile piece of kit. Would this happen to a quality piece of kit, such as Kontronic, Aveox or Schulze? I would be extremely surprised if it did. So maybe it’s just a symptom of some manufacturers, sometimes the performance is related to what you spend.

Incidentally, are we totally convinced about this throttle signal being a Pulse Width Modulation, (PWM) type? PWM is defined as the pulse repetition rate remaining steady and the width of the pulse varying, Pulse Frequency Modulation would be interpreted as the pulse width staying steady and the frequency varying. If this the case, throttle signal being PWM, then would the switching rate in fact actually increase? Or decrease? However, I also I have to consider that using PFM the frequency is possibly going to at it highest at full throttle, (or perhaps much more unlikely at closed throttle), so that implies that that any switching losses, and thus heat dissipated, will be at a maximum at this point. So that too, is unlikely.
There are a few variations on this theme, the transmitter modulation, PPM, Pulse Position Modulation, is a form of pulse frequency modulation, so the frame rate is constantly varying, depending on what position the sticks are in.

I’ve recently borrowed a motor and some props so that I could tinker with something else, but I just happened to notice that the switching frequency appeared to be highest at a point somewhere between half and full throttle; I will go back and have another look, also I can try and do some fairly accurate temperature rise checks, I have a thermometer. But I guess to really get a handle on the switching situation an oscilloscope would be very desirable, and that I don’t have. I can only conclude that if the temperature rise at half throttle is correct, and I’m sure that may well be so, there have been many reports of this, then there is some form of interaction between the commutated pulses and the throttle pulses. Difficult to see how the switching rate increases, as the motor slows the commutating switching becomes proportionally less frequent.

Simon, Just to be pedantic, and this is only a casual observation, without prejudice, but I think that really and truly all the losses are down to the resistance of the components. You are of course absolutely right about the heat produced being proportional to the resistance and the current flowing through it, but that also applies to the switching losses also, surely?

This seems to have the makings of an interesting conundrum, and I’m sure the answer is out there somewhere, someone else I’m sure will have been into this one!

PB

[WolstonFlyer]

This seems to be one explanation of how an ESC works....he says using frequency modulation. I am not sure if it helps?

How-an-ac-motor-and-its-esc-works

[Tony Smith 7]

Posted by Peter Beeney on 07/09/2012 14:25:41:

Simon, Just to be pedantic, and this is only a casual observation, without prejudice, but I think that really and truly all the losses are down to the resistance of the components. You are of course absolutely right about the heat produced being proportional to the resistance and the current flowing through it, but that also applies to the switching losses also, surely?

This seems to have the makings of an interesting conundrum, and I’m sure the answer is out there somewhere, someone else I’m sure will have been into this one!

PB

I think you could distinguish switching losses as being those due to the cut-off not being a perfect square wave, so there's a short period during each on or off during which the semiconductor is dropping some of the voltage.

[Peter Beeney]

WolstonFlyer, I’m afraid I’m not at all convinced by Luke Warm’s line of thinking. I shall have go back and read it throughly, but it’s the applied voltage that controls the speed of the motor. Hence kV, or revs per volt. In my book it’s the motor that actually controls the timing of the ESC pulses, not the other way round. But the ESC controls the speed of the motor by varying the applied voltage, it does this in a series of varying pulses, the full voltage is applied but over a cycle the average is less. For instance, if a pulse is one second long, and 12 volts are applied to a load for half a second and nothing for the other half, then the second consists of 2 half seconds, thus 12 + 0 = 12. Then divided by 2 = 6. So over the one second the average voltage is 6. But because the pulses are very short the motor appears to all intents and purposes to be one continuous movement.

This is exactly Chris’s statement that the ESC is switching two things.

I’m tending to think there are a few crossed up wires here to untangle……

PB

[Simon Chaddock]

This is from the EE Times describing power losses in MOSFETS.

Major causes of power loss

Power loss in a MOSFET comes from two sources. Every MOSFET has a resistive element, so it dissipates power as current is conducted through the device. The resistive parameter is described as on-resistance, or R_DS(ON). These conduction losses are inversely proportional to the size of the MOSFET; the larger the switching transistor, the lower its R_DS(ON) and, therefore, its conduction loss.

The other source of power loss is through switching losses. As the MOSFET switches on and off, its intrinsic parasitic capacitance stores and then dissipates energy during each switching transition. The losses are proportional to the switching frequency and the values of the parasitic capacitances. As the physical size of the MOSFET increases, its capacitance also increases; so, increasing MOSFET size also increases switching loss.

Nice phrase "parasitic capacitances"!

[PatMc]

Posted by WolstonFlyer on 07/09/2012 14:44:25:

This seems to be one explanation of how an ESC works....he says using frequency modulation. I am not sure if it helps?

How-an-ac-motor-and-its-esc-works

I'm afraid he's completely wrong. We don't apply ac to our motors, we apply pulsed dc via the ESC. The ESC constantly senses the rotor position in relation to the "active" winding, uses this to time when to apply a pulse the repeats this sequentialy around the windings.

If our motors & ESCs worked in the way he explains we would be able to use a single ESC for multi motors limited only by the current capacity. Also the Kv wouldn't be relevant as the motors would run at a speed determined by the number of poles & frequency of supply. (minus some "slip"

[Peter Beeney]

Yes indeed, I’ve always thought in my simple way that what really sorts out a boy fet from a man fet is it’s on resistance. The lower, the better, but I suspect the cost is perhaps to a small extent at least, subject to some sort of law squared, to get a FET with only half as much resistance it will cost you four times as much.

I’ve always had the switching FET impedance slightly different, I’ve always understood that the resistance cannot go from infinity, - that’s switched off - to none - switched on - in no time at all, so during the transition time there is some dissipation of power, or heat. Not much of a problem at low revs, but up around the 4 MHz region, such as in some switching regulators, it begins to have an effect.

However, I’m sure the ‘parasitic capacitance’ is the correct answer, but does that make make any difference? As I remember, from a very long time ago, the capacitive reactance Xc, is measured in ohms anyway, as a resistance, so the I squared R losses, as heat, will be calculated in just the same way. In the end, it always comes down to simple resistance.

It certainly seems as though Mr Warm’s ideas are inclined to be a little bit different, I’m afraid I’d have some difficulty in getting my head around some this. Some fairly straightforward statements here, but with no slightly more in-depth explanations; he’s turned conventional thinking on it’s head. Some of it would appear to be at least partly a correct statement, but I reckon that’s only a lucky guess!

But, as always, I too always stand to be corrected, but I usually need an explanation that convinces me with sound reason and explanations, rather than just bald statements.

Which brings me to the fact that I’ve had this ‘discussion’ with Pat before, about the old chestnut, sorry, query, - ‘is it AC, or is it DC’ applied to the motor? My view has always been that it’s DC from the battery to the ESC, but it’s AC from the ESC to the motor. That’s AC in as much that the current travels first in one direction at a given point, and then in the opposite direction at the same given point.

We did amicably agree to disagree on this subject last time, and so I’m sure that we will do so again, and again after that, after all it doesn’t make a lot of difference how we think things work, because they still function correctly regardless……

PB

[PatMc]

Peter, Luke Warm's "explanation" is complete nonsense. I don't think he has a clue about how our brushless motors &/or ESCs work.

For example his diagram of what he thinks would be seen on a 'scope shows +11.1v, 0v & -11.1v.

That's 22.2v pd from an 11.1v battery .

BTW it's got to be DC because there is never any -ve voltage applied anywhere in the circuit.

Edited By PatMc on 07/09/2012 20:58:47

[Peter Beeney]

Pat, I do wholeheartedly agree with you 110% about Luke Warm, it’s just that from experience I’ve learnt that it’s sometimes not really much of a good idea to be too emphatic about my point of view too quickly. I have to say I’m very cautious, I’ve found it’s very easy to drop off the edge before I’ve realised I’m anywhere near. For instance, I wouldn’t really comment about the ‘scope, because I haven’t got one to try, and so I can’t really say for certain. But perhaps I should make an effort to rectify this. I did have a poke about a while back, but there doesn’t seem to be many that run on a Mac. So it might have to be a stand alone box.

With regard to the AC/DC, read again like wot I rote…..

PB

[John Olsen 1]

It is probably only of academic interest, but here are a few points about power dissipation in switch mode circuits. There are several different sorts of losses:

* power dissipation in the on state due to the resistance of the FET. Usually called I squared R loss, and usually quite important.

*power dissipation in the off state due to leakage...usually negligible with modern devices.

*power dissipation during the switching transition. The peak at this point is half the supply voltage times half the switched current, so would be significant except that because the devices switch very fast, it only occurs for a very short time. So a dissipation of fifty Watts might seem a lot, but if it only occurs fo say 1 nanosecond, it is not a lot of actual energy. This will happen every time the device switches so as the switching rate (sometimes called sequency) goes up it will start to add up.

*power dissipation due to charging and discharging the parasitic capacitances. These should be reasonably low.

*power that was stored in load inductance and has to be dissipated when the load is switched. When the power to an inductor is switched off, current will try to keep flowing and a path often has to be provided for this, often a diode. The energy that was in the inductance ends up being disipated in the diode (which may in fact be an inherent feature of the construction of an FET) and so ends up in hte ESC.

So you can see that the actual power dissipation is going to be the sum of a lot of different sorts of losses.

The actual coils in the motor will only see AC, there will be no net DC voltage across them. If there was, the DC current would saturate the inductor. What wil actually be hapening is that for each connection on the motor, there are three possible states...1, connected to the positive rail, 2, conected to the negative rail, and 3, not connected to either rail, but with current paths available through protection diodes to enable current to continue flowing. Now, the peak voltage between a pair of connections in one direction will be when one end is connected to plus and the other to minus, say 11 volts or so in hte example above. The peak voltage in the opposite direction will be when the connections are reversed. So if you look at the voltage across a pair of connections, you can easily expect to see a peak to peak voltage of 22 volts from an 11 volt supply.

John

[Simon Chaddock]

John

Is the polarity to the coils actually reversed by the ESC in a brushless motor?

I am not at all sure it is.

[Erfolg]

Hmmm

Seem to have a problem originally on the same propeller, I was getting

160w @ 16 amp @ 8300 revs

now

124w @ 11 amp @ 4215 revs.

From a fully charged Lipo unfortunately I have not measured the volts when under load.

Something has changed, can it be all down to the ESC, is the battery failing. Perhaps reduce the propeller pitch, or increase dia?

[Erfolg]

I think the Lipo is on the way out, the volts are down to 10v in about 15 seconds,

I think they should stay near to 12 v for about say 30s, sagging slowly?

[Erfolg]

I have put in an order for a new Lipo. I will then see if that is the source of low power. I am just so far from the rated max and the values which I could easily reach.

It seems if it is not one thing then it is another, or maybe the problems are caused by a common source?

[PatMc]

Erfolg, it would probably be worth re-calibrating the throttle range just to be certain that's OK. It only takes a couple of minutes.

[PatMc]

Posted by John Olsen 1 on 07/09/2012 23:45:37:

The actual coils in the motor will only see AC, there will be no net DC voltage across them. If there was, the DC current would saturate the inductor. What wil actually be hapening is that for each connection on the motor, there are three possible states...1, connected to the positive rail, 2, conected to the negative rail, and 3, not connected to either rail, but with current paths available through protection diodes to enable current to continue flowing. Now, the peak voltage between a pair of connections in one direction will be when one end is connected to plus and the other to minus, say 11 volts or so in hte example above. The peak voltage in the opposite direction will be when the connections are reversed. So if you look at the voltage across a pair of connections, you can easily expect to see a peak to peak voltage of 22 volts from an 11 volt supply.

John

How can the motor see AC when there is only pulsed DC ?

Why would DC saturate the inductor (motor coils ?) ?

I don't understand your ref to protection diodes. Where are these diodes ?

Where does the 22v come from when the supply is 11v ?

[PatMc]

Posted by Simon Chaddock on 07/09/2012 23:59:22:

John

Is the polarity to the coils actually reversed by the ESC in a brushless motor?

I am not at all sure it is.

The motor coils are delta connected. When coil A is pulsed it is in parallel with coils B & C which are in series with each other. The current flow through B & C is half the magnitude & the reverse to A's.

The next pulse is through B in the same direction as the previous one through A (i.e. the current through B is reversed & doubled in magnitude. But of course A & C are in parallel. The current flow in A is also reversed & half the previous magnitude. The current flow through C is the same as the previous pulse.

This reversal in current direction does not constitute AC since the pulses are always positive to zero never negative.

[Peter Beeney]

With regard to the AC/DC dilemma posed by a permanent magnet motor, do you think it might help if we consider the very basic action? This would be my version of a standard brushed can, as a start.

We have a magnet with a north and south pole situated opposite one another in the shape of a tube, and inside is an armature, a rotating cylinder with two coils on it, consisting of just one long wire. The coils are arranged to be close to the magnets, with the windings arranged so that when a current flows though them they are opposite polarities. Now we connect a DC electrical supply to the two ends of the wire, such that the current flowing through the coils turns them into electro-magnets, with the polarity now the same as the permanent magnets, i.e. N to N and S to S. On the basis that like poles repel, the magnets will try to force themselves apart and thus the armature will start to turn. Once the armature has reached the 90 degree position away from the magnets they have lost their influence but the momentum keeps it going. As it gets further round, toward the 180 degree spot, the now unlike magnet poles begin to attract one another, and as the magnets become closest to each other the the attraction is greatest and so the armature stops dead?? …. So that’s not much of a motor, then… DC is a bit of a failure here!

I think perhaps we need to start again, let’s connect up our DC once again but this time just as the poles reach 180 degrees, the point where they want to become stationary we arrange for the current flowing through the coils to be reversed, i.e. made to flow in the opposite direction, this reverses the magnetic poles in the coils and so they now continue to repel the permanent magnets; and so the armature now continues on it’s way until it gets to the 360 degree point where the current direction is changed back to the original flow; and now the motor will continue to turn, with the current flowing in the armature first one way and then the other; changing over at whatever speed the motor is running at.

This change over is effected by having the two ends of the coil wires on two separate contacts, the commutator, which turns as the armature turns, and there are two stationary carbon brushes touching these contacts to transfer the electricity. As it turns so each brush bears against each contact in turn, thus automatically changing, or reversing the current flow. These are placed so that the changeover occurs at exactly the right moment to keep the motor turning.

Thus we can see that whilst we have DC up to the brushes, it is actually AC flowing in the armature, the frequency depending on the speed of the motor.

If we take this idea further on to a brushless motor then it’s possible to arrange for the armature to stay stationary, it’s now become the stator, and the magnets to rotate, they then become the rotor. The two wires from the stator coils can now be just simply extended out to a controller, where instead of having moving contacts there are solid state switches which turn the current on and off to the coils. The controller gets it’s information when to switch and reverse the current from feedback from the armature. But there is absolutely no doubt that the current has to reverse once each time the motor turns, if it didn’t we can see from the first example that the motor would never function.

[Peter Beeney]

PART TWO

Of course, our brushed motor is always actually a bit more sophisticated, it has multiple magnets and poles, this keeps the rotation even and smooth but exactly the same principle applies to each circuit. Likewise the brushless motor, this does the same trick by having a number of magnets and 3 phases, each running 120 degrees apart from each other. But as before, each circuit is reversed at 180 degrees round the can, simply to enable it to function as a machine that will constant revolve in one direction.

Another proof, if it were needed, perhaps, that the current driving our brushless motor is AC is if we could run our motor up to speed and then arrange for it to be driven on and speeded up just a little bit faster we would find that the motor is now a generator, and that it is producing AC, the current going first one way and then the other, exactly the same as what was required to drive it. This is because the coils are cutting the two magnetic poles, North and South, once in every revolution. As before, this is AC, the current going first one way and then the opposite way in every cycle. But, if we apply this AC to a diode bridge we can very easily ‘steer’ this constantly changing current so that it’s all now going in the same direction, and so becomes DC. Of course, if we turn the motor at any speed it will always act as a generator, but indubitably AC, this is where the kV comes from, how many volts generated per 1000 revs. In the case of our brushed motor, when this becomes a generator, or dynamo, the commutator acts like the diodes and thus changes the AC generated in the armature to DC at the brushes. Exactly the opposite operation it did when it was a motor.

I tend to be very cautious when measuring AC with a voltmeter, for a start it’s most likely that a basic instrument will only measure the rms value of a sine wave, such as the mains. If you want to measure other waveforms you need a meter that will measure true rms values; it can be done, of course, but it has a cost, in money! The waveform supplied to an electric motor to make it run is most unlikely to be a sine wave, so you might get some very unlikely results here, for instance.
If we were measuring the peak value of AC then the value would simply be the zero to peak value, the wave form is symmetrical and therefore if we add the positive and negative figures they would amount to zero, likewise the average current, adding both halves adds up to zero, so we only ever take these values over half a cycle, but if there were some DC superimposed on the AC this shifts the AC vertically by a given amount then we might want to measure the peak to peak value, which I believe in this case is often of use.

Now, I guess, we must now be getting into some very cold, deep and murky waters indeed……

PB

[Erfolg]

Patmac

You are correct and it is much appreciated the suggestion to re-calibrate the throttle range.

However, I am some way from solving the issue. At the field today, the the motor would not start and then all the servos stared to jitter wildly. A couple of restarts were tried with same result.

Seems unlikely that two different ESC's would have the same issue in being faultyon the same set up.

I am now starting to suspect the Futaba 6017 Rx.

Retrying in the house, all was well, after dismantling the equipment so as to see it. I have however noted that the elevator servo is bit jittery in movement. I will replace it this evening.

What started as a minor annoyance, has grown into a model threatening issue.

I am enjoying the tech discussion, when agreed, will one of you guys write a short piece on how they work, for us numbties?

Edited By Erfolg on 09/09/2012 15:13:16

[Erfolg]

I am now a little concerned. I do not like co-incidences and tend not to believe in them, however statically likely.

I have temporally taken an ESC out of a another model and used it on the Nobler set up. It works.

I have two ESC's now which have rapidly failed, on the same model (3 in total) can it just be coincidence, or is something causing the ESC to fail?

Any Ideas, or can it be just co-incidence?

[WolstonFlyer]

Peter your explanations are great, takes me back to physics class many years ago.

I guess it would be good to put an esc on an oscilloscope and see what the output looks like.

Erfolg: Could be the motor causing the problem? Perhaps a short in the windings?

[PatMc]

Despite the long winded "explanation(?)" a DC brushed motor only has DC flow & our DC brushless motor ditto. In both cases the current is commutated to be presented to the coils as pulses of DC.

The fact that a brushless motor will produce AC when driven as an alternator is not proof of that it is driven by AC as a motor. As an alternator each phase produces current in a sinusiodal wave varying between +ve & -ve in equal magnitude. As a motor each coil receives pulses of DC always above zero volts.

[Erfolg]

How would I check for a short?

Measure the ohms across each coil, looking for an inbalance?

Or something else?

[Thomas Barwick]

ON the hobbywing ESC there is a timing option in the instructions it mentions this , and i have had a esc catch fire due to setting it to the wrong one. Maybe this is causing an issue with one motor but not the other as they need differnet timings. ?