As you can see, adding a 1000 uF capacitor and a faster operational amplifier drastically reduced the dissipated power of the transistor (I have Ptr and have this power).

But which thing contributed what benefit? Three things changed at once (amp, gate resistor, cap on power).

 

Am I missing something here? The original question was about the battery overheating, but most of the discussion seems to be about transistor dissipation. Would the transistor issues also cause battery heating, or are these separate issues?

 

See

 

The first thing you should do is add a diode and change (decrease) the average current consumption with a potentiometer. Adding a capacitor will reduce the battery current, because The energy for heating the transistor is taken from the battery. Elements of size AA allow an average (long) current of 100 mA. And only for a short time 500mA. If you need a large current (for fast engine rotation) use galvanic cells or larger batteries. The corrected circuit will allow to reduce the average current when the total voltage rises, together with the adjustment of the potentiometer (decreasing the duty cycle).

 

You should gratefully accept everything that was mentioned above. After all, this all increases the efficiency of the scheme. On what rated voltage is the motor calculated and what is the current (or power)? I admit that the battery should warm up.

 

Am I missing something here? The original question was about the battery overheating, but most of the discussion seems to be about transistor dissipation. Would the transistor issues also cause battery heating, or are these separate issues?

No, you're not missing anything. The circuit performance has been improved but the original problem probably still exists.

Transistor and battery dissipation are separate issues, provided that the switching loss is not so great as to be making a significant difference to the total power from the battery. If you look at Bordodynov's table at #23, you'll see that at step 3 in the pbatt entries, loss due to the slower amp is adding about 8% (column 3 vs column 1). That is certainly not going to make much difference in the heating of the battery, but is not insignificant to overall efficiency.

Note step 3, column 2 - big power loss in the FET due to the three ways that battery impedance makes a mess - two interfering with comparator hysteresis and the third being essentially negative feedback for the gate drive (FET begins to turn on, gate drive voltage falls).

The LM358 using CMOS hex Schmitt trigger as a gate driver (1 as inverter driving other 5 in parallel) would probably outperform the TLE2141 at lower parts cost, though higher parts count and board area.

My original idea with adding the cap across the battery was to bring the RMS current from the battery down by averaging, to some extent. It clearly does that, but whether it improves things much is dubious with the load used in the simulations. At the time I first suggested the 1 mF cap I hadn't looked at the issue of the 3-way detrimental negative feedback due to source impedance.