The Model 3 uses switched reluctance and AC induction motors, both of which are electronically commutated.
They aren't commutated at all. They are three phase motors. They have 3 windings of 4 poles each connected in a wye connection.
3.5kw is an insane amount of heat for the inverter to dump
. Each percent inefficiency of the inverter (in normal operation) represents 500 watts at a 50 kW draw (which isn't unusual in my car). So in normal operation I would expect a couple of kW and going to 3500 would hardly be a stretch. I wonder if you have ever done a semiconductor thermal load calculation. If one has extra heat to dispose of he designs his system to have lower thermal impedance to the sink and thus prevent excessive rise. Are you aware that these inverters use silicon carbide transistors which are pretty tough when it comes to temperature.
and it would be the worst available option for heating by a considerable margin. The drastic thermal stresses would dramatically accelerate wear and shorten inverter life likely orders of magnitude if they would tolerate it at all.
I wonder why you say that? Do you have any relevant literature, experience or even reasoning to back that up? Reasoning is OK. At least it forms the basis of a discussion. And as Tesla isn't revealing design details that's all we have.
SiC packages like used in the rear inverter are capable of carrying high currents, but only if they are efficient. They can’t dump considerable heat super effectively.
This implies that you at least know about silicon carbide. Transistors are perfectly well capable of being operated inefficiently if they are in a low thermal impedance path to a sink.Interestingly enough the very first site I come up with typing "silicon carbide transistor" into the search bar offers:
"As an alternative to traditional silicon MOSFETs, silicon carbide MOSFETs offer the advantages of higher blocking voltage, lower on-state resistance, and
higher thermal conductivity."
High efficiency means low heat.
Yes.
An electronically commutated motor does not magically spin when power is applied,
The ones in my house do. No magic involved.
it requires knowledge of the rotor position and then it coordinates the waveforms applied to hit the desired motion.
Rotor position is not required for the motor to spin. If you want to do vector field control it is and we do in this application.
This is very complex, and not super easy to do efficiently.
I grant you the intricacies of DQ control give me a headache but once you have the transforms sorted out and have DSP chips that will do them fast enough the basic concept is pretty simple. The inverter constructs sin waves at the proper frequency, amplitude and phase to give the desired torque and field. D and Q are like the field and armature currents in a DC motor and so can be controlled by PI controllers making it possible to do all the neat stuff the Tesla cars can do.
There’s tons of ways for this to be marginally less efficient,
Yes, as the simplest of them is to simply slow the rise time of the transistor gating waveform and/or allow some common mode current which is why I am quite confident that this is what they do. Besides that I have seen it mentioned in other places that this is the technique and as it makes so much sense I have come to accept that though I readily admit that Occams Razor is not proof.
Can you suggest a method for changing the inverter's gating in such a way that the motor itself becomes less efficient? I may be slow but other than locking the rotor and running DC I can't think of a way to do it. Actually something just came to mind: harmonic distortion and phase imbalance. This would lead to negative sequence currents which reduce inefficiency. Is that what you have in mind?
and the primary heating impact is going to be on the motors windings which are quite capable of tolerating heating, at least short term. The thermal cycling doesn’t typically cause as considerable fatigue as it does with a mosfet, as well as the dramatic increase in surface area and thermal mass. It would be the logical choice as a heater by a considerable gap.
Have you considered that in normal operation the inverter transistors are subject to constant thermal shock as inverter power demand can go from 0 to 150 kW (or more typically 100 kW) and then back to 30 kW in a matter of seconds? Have you considered that the transistors, heat sinks and motor steel are all part of the same thermal mass?